Search

Phenotypic plasticity, the ability to produce multiple phenotypes from a single genotype, represents an excellent model with which to examine the relationship between gene expression and phenotypes. Analyses of the molecular foundations of phenotypic plasticity are challenging, however, especially in the case of complex social phenotypes. Here we apply a machine learning approach to tackle this challenge by analyzing individual-level gene expression profiles of Polistes dominula paper wasps following the loss of a queen. We find that caste-associated gene expression profiles respond strongly to queen loss, and that this change is partly explained by attributes such as age but occurs even in individuals that appear phenotypically unaffected. These results demonstrate that large changes in gene expression may occur in the absence of outwardly detectable phenotypic changes, resulting here in a socially mediated de-differentiation of individuals at the transcriptomic level but not at the levels of ovarian development or behavior., Connecting genotypes to complex social behaviour is challenging. Taylor et al. use machine learning to show a strong response of caste-associated gene expression to queen loss, wherein individual wasp’s expression profiles become intermediate between queen and worker states, even in the absence of behavioural changes.

Identifying the genes and mutations that drive the emergence of tumors is a critical step to improving our understanding of cancer and identifying new directions for disease diagnosis and treatment. Despite the large volume of genomics data, the precise detection of driver mutations and their carrying genes, known as cancer driver genes, from the millions of possible somatic mutations remains a challenge. Computational methods play an increasingly important role in discovering genomic patterns associated with cancer drivers and developing predictive models to identify these elements. Machine learning (ML), including deep learning, has been the engine behind many of these efforts and provides excellent opportunities for tackling remaining gaps in the field. Thus, this survey aims to perform a comprehensive analysis of ML-based computational approaches to identify cancer driver mutations and genes, providing an integrated, panoramic view of the broad data and algorithmic landscape within this scientific problem. We discuss how the interactions among data types and ML algorithms have been explored in previous solutions and outline current analytical limitations that deserve further attention from the scientific community. We hope that by helping readers become more familiar with significant developments in the field brought by ML, we may inspire new researchers to address open problems and advance our knowledge towards cancer driver discovery.

This study aims to explore biomarkers associated with vitiligo and analyze the pathological role of immune cell infiltration in the disease. We used the robust rank aggregation (RRA) method to integrate three vitiligo data sets downloaded from gene expression omnibus database, identify the differentially expressed genes (DEGs) and analyze the functional correlation. Then, the comprehensive strategy of combined weighted gene coexpression network analysis (WGCNA) and logical regression of the selection operator (LASSO), support vector machine recursive feature elimination (SVM-RFE), and random forest (RF) machine learning algorithm are employed to screen and biomarkers associated with vitiligo. Finally, the immune cell infiltration of vitiligo was evaluated by CIBERSORT, and the correlation between biomarkers and infiltrating immune cells was analyzed. Herein, we identified 131 robust DEGs, and enrichment analysis results showed that robust DEGs and melanogenesis were closely associated with vitiligo development and progression. TYR, TYRP1, DCT and LARP7 were identified as vitiligo-related biomarkers. Immune infiltration analysis demonstrated that CD4 T Cell, CD8 T Cell, Tregs, NK cells, dendritic cells, and macrophages were involved in vitiligo’s pathogenesis. In summary, we adopted a comprehensive strategy to screen biomarkers related to vitiligo and explore the critical role of immune cell infiltration in vitiligo. Abbreviations: TYR, Tyrosinase; TYRP1, Tyrosinase-related protein-1; DCT, dopachrome tautomerase; LARP7, La ribonucleoprotein domain family, member-7; RRA, robust rank aggregation; DEGs, differentially expressed genes; WGCNA, weighted gene coexpression network analysis; LASSO, logical regression of the selection operator; SVM-RFE, support vector machine recursive feature elimination; RF, random forest; GWAS, Genome-wide association study; FasL, Fas-Fas ligand; Tregs, T-regulatory cells; NK, natural killer; GEPCs, gene expression profiling chips; GO, gene ontology; GSEA, gene set enrichment analysis; FDR, false discovery rate; AUC, area under the curve; ROC, receiver-operating characteristic; BP, biological process; CC, cellular component; MF, molecular function., GraphicalAbstract

Although DNA methylation variation of autosomal CpGs provides robust age predictive biomarkers, no male-specific age predictor exists based on Y-CpGs yet. Since sex chromosomes play an important role in aging, a Y-chromosome-based age predictor would allow studying male-specific aging effects and would also be useful in forensics. Here, we used blood-based DNA methylation microarray data of 1,057 males from six cohorts aged 15-87 and identified 75 Y-CpGs with an interquartile range of ≥0.1. Of these, 22 and six were significantly hyper- and hypomethylated with age (p(cor)

Rationale: This paper aims to show how the focus on eradicating bias from Machine Learning decision-support systems in medical diagnosis diverts attention from the hermeneutic nature of medical decision-making and the productive role of bias. We want to show how an introduction of Machine Learning systems alters the diagnostic process. Reviewing the negative conception of bias and incorporating the mediating role of Machine Learning systems in the medical diagnosis are essential for an encompassing, critical and informed medical decision-making. Methods: This paper presents a philosophical analysis, employing the conceptual frameworks of hermeneutics and technological mediation, while drawing on the case of Machine Learning algorithms assisting doctors in diagnosis. This paper unravels the non-neutral role of algorithms in the doctor's decision-making and points to the dialogical nature of interaction not only with the patients but also with the technologies that co-shape the diagnosis. Findings: Following the hermeneutical model of medical diagnosis, we review the notion of bias to show how it is an inalienable and productive part of diagnosis. We show how Machine Learning biases join the human ones to actively shape the diagnostic process, simultaneously expanding and narrowing medical attention, highlighting certain aspects, while disclosing others, thus mediating medical perceptions and actions. Based on that, we demonstrate how doctors can take Machine Learning systems on board for an enhanced medical diagnosis, while being aware of their non-neutral role. Conclusions: We show that Machine Learning systems join doctors and patients in co-designing a triad of medical diagnosis. We highlight that it is imperative to examine the hermeneutic role of the Machine Learning systems. Additionally, we suggest including not only the patient, but also colleagues to ensure an encompassing diagnostic process, to respect its inherently hermeneutic nature and to work productively with the existing human and machine biases.

Early risk assessments and interventions for metabolic syndrome (MetS) are limited because of a lack of effective biomarkers. In the present study, several candidate genes were selected as a blood-based transcriptomic signature for MetS. We collected so far the largest MetS-associated peripheral blood high-throughput transcriptomics data and put forward a novel feature selection strategy by combining weighted gene co-expression network analysis, protein-protein interaction network analysis, LASSO regression and random forest approaches. Two gene modules and 51 hub genes as well as a 9-hub-gene signature associated with metabolic syndrome were identified. Then, based on this 9-hub-gene signature, we performed logistic analysis and subsequently established a web nomogram calculator for metabolic syndrome risk (https://xjtulgz.shinyapps.io/DynNomapp/). This 9-hub-gene signature showed excellent classification and calibration performance (AUC = 0.968 in training set, AUC = 0.883 in internal validation set, AUC = 0.861 in external validation set) as well as ideal potential clinical benefit.

Background The current COVID-19 pandemic is unprecedented; under resource-constrained settings, predictive algorithms can help to stratify disease severity, alerting physicians of high-risk patients; however, there are only few risk scores derived from a substantially large electronic health record (EHR) data set, using simplified predictors as input. Objective The objectives of this study were to develop and validate simplified machine learning algorithms that predict COVID-19 adverse outcomes; to evaluate the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and calibration of the algorithms; and to derive clinically meaningful thresholds. Methods We performed machine learning model development and validation via a cohort study using multicenter, patient-level, longitudinal EHRs from the Optum COVID-19 database that provides anonymized, longitudinal EHR from across the United States. The models were developed based on clinical characteristics to predict 28-day in-hospital mortality, intensive care unit (ICU) admission, respiratory failure, and mechanical ventilator usages at inpatient setting. Data from patients who were admitted from February 1, 2020, to September 7, 2020, were randomly sampled into development, validation, and test data sets; data collected from September 7, 2020, to November 15, 2020, were reserved as the postdevelopment prospective test data set. Results Of the 3.7 million patients in the analysis, 585,867 patients were diagnosed or tested positive for SARS-CoV-2, and 50,703 adult patients were hospitalized with COVID-19 between February 1 and November 15, 2020. Among the study cohort (n=50,703), there were 6204 deaths, 9564 ICU admissions, 6478 mechanically ventilated or EMCO patients, and 25,169 patients developed acute respiratory distress syndrome or respiratory failure within 28 days since hospital admission. The algorithms demonstrated high accuracy (AUC 0.89, 95% CI 0.89-0.89 on the test data set [n=10,752]), consistent prediction through the second wave of the pandemic from September to November (AUC 0.85, 95% CI 0.85-0.86) on the postdevelopment prospective test data set [n=14,863], great clinical relevance, and utility. Besides, a comprehensive set of 386 input covariates from baseline or at admission were included in the analysis; the end-to-end pipeline automates feature selection and model development. The parsimonious model with only 10 input predictors produced comparably accurate predictions; these 10 predictors (age, blood urea nitrogen, SpO2, systolic and diastolic blood pressures, respiration rate, pulse, temperature, albumin, and major cognitive disorder excluding stroke) are commonly measured and concordant with recognized risk factors for COVID-19. Conclusions The systematic approach and rigorous validation demonstrate consistent model performance to predict even beyond the period of data collection, with satisfactory discriminatory power and great clinical utility. Overall, the study offers an accurate, validated, and reliable prediction model based on only 10 clinical features as a prognostic tool to stratifying patients with COVID-19 into intermediate-, high-, and very high-risk groups. This simple predictive tool is shared with a wider health care community, to enable service as an early warning system to alert physicians of possible high-risk patients, or as a resource triaging tool to optimize health care resources.

Background Accurate, objective pain assessment is required in the health care domain and clinical settings for appropriate pain management. Automated, objective pain detection from physiological data in patients provides valuable information to hospital staff and caregivers to better manage pain, particularly for patients who are unable to self-report. Galvanic skin response (GSR) is one of the physiologic signals that refers to the changes in sweat gland activity, which can identify features of emotional states and anxiety induced by varying pain levels. This study used different statistical features extracted from GSR data collected from postoperative patients to detect their pain intensity. To the best of our knowledge, this is the first work building pain models using postoperative adult patients instead of healthy subjects. Objective The goal of this study was to present an automatic pain assessment tool using GSR signals to predict different pain intensities in noncommunicative, postoperative patients. Methods The study was designed to collect biomedical data from postoperative patients reporting moderate to high pain levels. We recruited 25 participants aged 23-89 years. First, a transcutaneous electrical nerve stimulation (TENS) unit was employed to obtain patients' baseline data. In the second part, the Empatica E4 wristband was worn by patients while they were performing low-intensity activities. Patient self-report based on the numeric rating scale (NRS) was used to record pain intensities that were correlated with objectively measured data. The labels were down-sampled from 11 pain levels to 5 different pain intensities, including the baseline. We used 2 different machine learning algorithms to construct the models. The mean decrease impurity method was used to find the top important features for pain prediction and improve the accuracy. We compared our results with a previously published research study to estimate the true performance of our models. Results Four different binary classification models were constructed using each machine learning algorithm to classify the baseline and other pain intensities (Baseline [BL] vs Pain Level [PL] 1, BL vs PL2, BL vs PL3, and BL vs PL4). Our models achieved higher accuracy for the first 3 pain models than the BioVid paper approach despite the challenges in analyzing real patient data. For BL vs PL1, BL vs PL2, and BL vs PL4, the highest prediction accuracies were achieved when using a random forest classifier (86.0, 70.0, and 61.5, respectively). For BL vs PL3, we achieved an accuracy of 72.1 using a k-nearest-neighbor classifier. Conclusions We are the first to propose and validate a pain assessment tool to predict different pain levels in real postoperative adult patients using GSR signals. We also exploited feature selection algorithms to find the top important features related to different pain intensities. International Registered Report Identifier (IRRID) RR2-10.2196/17783

Motivation Recent developments in technology have enabled researchers to collect multiple OMICS datasets for the same individuals. The conventional approach for understanding the relationships between the collected datasets and the complex trait of interest would be through the analysis of each OMIC dataset separately from the rest, or to test for associations between the OMICS datasets. In this work we show that integrating multiple OMICS datasets together, instead of analysing them separately, improves our understanding of their in-between relationships as well as the predictive accuracy for the tested trait. Several approaches have been proposed for the integration of heterogeneous and high-dimensional (p≫n) data, such as OMICS. The sparse variant of canonical correlation analysis (CCA) approach is a promising one that seeks to penalize the canonical variables for producing sparse latent variables while achieving maximal correlation between the datasets. Over the last years, a number of approaches for implementing sparse CCA (sCCA) have been proposed, where they differ on their objective functions, iterative algorithm for obtaining the sparse latent variables and make different assumptions about the original datasets. Results Through a comparative study we have explored the performance of the conventional CCA proposed by Parkhomenko et al., penalized matrix decomposition CCA proposed by Witten and Tibshirani and its extension proposed by Suo et al. The aforementioned methods were modified to allow for different penalty functions. Although sCCA is an unsupervised learning approach for understanding of the in-between relationships, we have twisted the problem as a supervised learning one and investigated how the computed latent variables can be used for predicting complex traits. The approaches were extended to allow for multiple (more than two) datasets where the trait was included as one of the input datasets. Both ways have shown improvement over conventional predictive models that include one or multiple datasets. Availability and implementation https://github.com/theorod93/sCCA. Supplementary information Supplementary data are available at Bioinformatics online.

Significance: Screening and early detection of oral potentially malignant lesions (OPMLs) are of great significance in reducing the mortality rates associated with head and neck malignancies. Intra-oral multispectral optical imaging of tissues in conjunction with cloud-based machine learning (CBML) can be used to detect oral precancers at the point-of-care (POC) and guide the clinician to the most malignant site for biopsy. Aim: Develop a bimodal multispectral imaging system (BMIS) combining tissue autofluorescence and diffuse reflectance (DR) for mapping changes in oxygenated hemoglobin (HbO2) absorption in the oral mucosa, quantifying tissue abnormalities, and guiding biopsies. Approach: The hand-held widefield BMIS consisting of LEDs emitting at 405, 545, 575, and 610 nm, 5MPx monochrome camera, and proprietary Windows-based software was developed for image capture, processing, and analytics. The DR image ratio (R610/R545) was compared with pathologic classification to develop a CBML algorithm for real-time assessment of tissue status at the POC. Results: Sensitivity of 97.5% and specificity of 92.5% were achieved for discrimination of OPML from patient normal in 40 sites, whereas 82% sensitivity and 96.6% specificity were obtained for discrimination of abnormal (OPML + SCC) in 89 sites. Site-specific algorithms derived for buccal mucosa (27 sites) showed improved sensitivity and specificity of 96.3% for discrimination of OPML from normal. Conclusions: Assessment of oral cancer risk is possible by mapping of HbO2 absorption in tissues, and the BMIS system developed appears to be suitable for biopsy guidance and early detection of oral cancers.

BackgroundDigital phenotyping and machine learning are currently being used to augment or even replace traditional analytic procedures in many domains, including health care. Given the heavy reliance on smartphones and mobile devices around the world, this readily available source of data is an important and highly underutilized source that has the potential to improve mental health risk prediction and prevention and advance mental health globally.ObjectiveThis study aimed to apply machine learning in an acute mental health setting for suicide risk prediction. This study uses a nascent approach, adding to existing knowledge by using data collected through a smartphone in place of clinical data, which have typically been collected from health care records.MethodsWe created a smartphone app called Strength Within Me, which was linked to Fitbit, Apple Health kit, and Facebook, to collect salient clinical information such as sleep behavior and mood, step frequency and count, and engagement patterns with the phone from a cohort of inpatients with acute mental health (n=66). In addition, clinical research interviews were used to assess mood, sleep, and suicide risk. Multiple machine learning algorithms were tested to determine the best fit.ResultsK-nearest neighbors (KNN; k=2) with uniform weighting and the Euclidean distance metric emerged as the most promising algorithm, with 68% mean accuracy (averaged over 10,000 simulations of splitting the training and testing data via 10-fold cross-validation) and an average area under the curve of 0.65. We applied a combined 5×2 F test to test the model performance of KNN against the baseline classifier that guesses training majority, random forest, support vector machine and logistic regression, and achieved F statistics of 10.7 (P=.009) and 17.6 (P=.003) for training majority and random forest, respectively, rejecting the null of performance being the same. Therefore, we have taken the first steps in prototyping a system that could continuously and accurately assess the risk of suicide via mobile devices.ConclusionsPredicting for suicidality is an underaddressed area of research to which this paper makes a useful contribution. This is part of the first generation of studies to suggest that it is feasible to utilize smartphone-generated user input and passive sensor data to generate a risk algorithm among inpatients at suicide risk. The model reveals fair concordance between phone-derived and research-generated clinical data, and with iterative development, it has the potential for accurate discriminant risk prediction. However, although full automation and independence of clinical judgment or input would be a worthy development for those individuals who are less likely to access specialist mental health services, and for providing a timely response in a crisis situation, the ethical and legal implications of such advances in the field of psychiatry need to be acknowledged.

Background The number of patients with neurological diseases is currently increasing annually, which presents tremendous challenges for both patients and doctors. With the advent of advanced information technology, digital medical care is gradually changing the medical ecology. Numerous people are exploring new ways to receive a consultation, track their diseases, and receive rehabilitation training in more convenient and efficient ways. In this paper, we explore the use of facial expression recognition via artificial intelligence to diagnose a typical neurological system disease, Parkinson disease (PD). Objective This study proposes methods to diagnose PD through facial expression recognition. Methods We collected videos of facial expressions of people with PD and matched controls. We used relative coordinates and positional jitter to extract facial expression features (facial expression amplitude and shaking of small facial muscle groups) from the key points returned by Face++. Algorithms from traditional machine learning and advanced deep learning were utilized to diagnose PD. Results The experimental results showed our models can achieve outstanding facial expression recognition ability for PD diagnosis. Applying a long short-term model neural network to the positions of the key features, precision and F1 values of 86% and 75%, respectively, can be reached. Further, utilizing a support vector machine algorithm for the facial expression amplitude features and shaking of the small facial muscle groups, an F1 value of 99% can be achieved. Conclusions This study contributes to the digital diagnosis of PD based on facial expression recognition. The disease diagnosis model was validated through our experiment. The results can help doctors understand the real-time dynamics of the disease and even conduct remote diagnosis.

©Andrea Ferrario, Burcu Demiray, Kristina Yordanova, Minxia Luo, Mike Martin. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 15.09.2020. BACKGROUND: Reminiscence is the act of thinking or talking about personal experiences that occurred in the past. It is a central task of old age that is essential for healthy aging, and it serves multiple functions, such as decision-making and introspection, transmitting life lessons, and bonding with others. The study of social reminiscence behavior in everyday life can be used to generate data and detect reminiscence from general conversations. OBJECTIVE: The aims of this original paper are to (1) preprocess coded transcripts of conversations in German of older adults with natural language processing (NLP), and (2) implement and evaluate learning strategies using different NLP features and machine learning algorithms to detect reminiscence in a corpus of transcripts. METHODS: The methods in this study comprise (1) collecting and coding of transcripts of older adults' conversations in German, (2) preprocessing transcripts to generate NLP features (bag-of-words models, part-of-speech tags, pretrained German word embeddings), and (3) training machine learning models to detect reminiscence using random forests, support vector machines, and adaptive and extreme gradient boosting algorithms. The data set comprises 2214 transcripts, including 109 transcripts with reminiscence. Due to class imbalance in the data, we introduced three learning strategies: (1) class-weighted learning, (2) a meta-classifier consisting of a voting ensemble, and (3) data augmentation with the Synthetic Minority Oversampling Technique (SMOTE) algorithm. For each learning strategy, we performed cross-validation on a random sample of the training data set of transcripts. We computed the area under the curve (AUC), the average precision (AP), precision, recall, as well as F1 score and specificity measures on the test data, for all combinations of NLP features, algorithms, and learning strategies. RESULTS: Class-weighted support vector machines on bag-of-words features outperformed all other classifiers (AUC=0.91, AP=0.56, precision=0.5, recall=0.45, F1=0.48, specificity=0.98), followed by support vector machines on SMOTE-augmented data and word embeddings features (AUC=0.89, AP=0.54, precision=0.35, recall=0.59, F1=0.44, specificity=0.94). For the meta-classifier strategy, adaptive and extreme gradient boosting algorithms trained on word embeddings and bag-of-words outperformed all other classifiers and NLP features; however, the performance of the meta-classifier learning strategy was lower compared to other strategies, with highly imbalanced precision-recall trade-offs. CONCLUSIONS: This study provides evidence of the applicability of NLP and machine learning pipelines for the automated detection of reminiscence in older adults' everyday conversations in German. The methods and findings of this study could be relevant for designing unobtrusive computer systems for the real-time detection of social reminiscence in the everyday life of older adults and classifying their functions. With further improvements, these systems could be deployed in health interventions aimed at improving older adults' well-being by promoting self-reflection and suggesting coping strategies to be used in the case of dysfunctional reminiscence cases, which can undermine physical and mental health., Journal of Medical Internet Research, 22 (9), ISSN:1438-8871

Messenger RNA (mRNA) vaccines have recently emerged as a new type of vaccine technology, showing strong potential to combat the COVID-19 pandemic. In addition to SARS-CoV-2 which caused the pandemic, mRNA vaccines have been developed and tested to prevent infectious diseases caused by other viruses such as Zika virus, the dengue virus, the respiratory syncytial virus, influenza H7N9 and Flavivirus. Interestingly, mRNA vaccines may also be useful for preventing non-infectious diseases such as diabetes and cancer. This review summarises the current progresses of mRNA vaccines designed for a range of diseases including COVID-19. As epitope study is a primary component in the in silico design of mRNA vaccines, we also survey on advanced bioinformatics and machine learning algorithms which have been used for epitope prediction, and review on user-friendly software tools available for this purpose. Finally, we discuss some of the unanswered concerns about mRNA vaccines, such as unknown long-term side effects, and present with our perspectives on future developments in this exciting area.

Motivation Validation of variable selection and predictive performance is crucial in construction of robust multivariate models that generalize well, minimize overfitting and facilitate interpretation of results. Inappropriate variable selection leads instead to selection bias, thereby increasing the risk of model overfitting and false positive discoveries. Although several algorithms exist to identify a minimal set of most informative variables (i.e. the minimal-optimal problem), few can select all variables related to the research question (i.e. the all-relevant problem). Robust algorithms combining identification of both minimal-optimal and all-relevant variables with proper cross-validation are urgently needed. Results We developed the MUVR algorithm to improve predictive performance and minimize overfitting and false positives in multivariate analysis. In the MUVR algorithm, minimal variable selection is achieved by performing recursive variable elimination in a repeated double cross-validation (rdCV) procedure. The algorithm supports partial least squares and random forest modelling, and simultaneously identifies minimal-optimal and all-relevant variable sets for regression, classification and multilevel analyses. Using three authentic omics datasets, MUVR yielded parsimonious models with minimal overfitting and improved model performance compared with state-of-the-art rdCV. Moreover, MUVR showed advantages over other variable selection algorithms, i.e. Boruta and VSURF, including simultaneous variable selection and validation scheme and wider applicability. Availability and implementation Algorithms, data, scripts and tutorial are open source and available as an R package (‘MUVR’) at https://gitlab.com/CarlBrunius/MUVR.git. Supplementary information Supplementary data are available at Bioinformatics online.

MotivationRecent sequence-based analyses have identified a lot of gene variants that may contribute to neurogenetic disorders such as autism spectrum disorder and schizophrenia. Several state-of-the-art network-based analyses have been proposed for mechanical understanding of genetic variants in neurogenetic disorders. However, these methods were mainly designed for modeling and analyzing single networks that do not interact with or depend on other networks, and thus cannot capture the properties between interdependent systems in brain-specific tissues, circuits, and regions which are connected each other and affect behavior and cognitive processes.ResultsWe introduce a novel and efficient framework, called a “Network of Networks” (NoN) approach, to infer the interconnectivity structure between multiple networks where the response and the predictor variables are topological information matrices of given networks. We also propose Graph-Oriented SParsE Learning (GOSPEL), a new sparse structural learning algorithm for network graph data to identify a subset of the topological information matrices of the predictors related to the response. We demonstrate on simulated data that GOSPEL outperforms existing kernel-based algorithms in terms of F-measure. On real data from human brain region-specific functional networks associated with the autism risk genes, we show that the NoN model provides insights on the autism-associated interconnectivity structure between functional interaction networks and a comprehensive understanding of the genetic basis of autism across diverse regions of the brain.AvailabilityOur software is available from https://github.com/infinite-point/GOSPEL.Contactkawakubo@med.nagoya-u.ac.jp, shimamura@med.nagoya-u.ac.jpSupplementary informationSupplementary data are available at Bioinformatics online.

This study presents an application of machine learning (ML) methods for detecting the presence of stenoses and aneurysms in the human arterial system. Four major forms of arterial disease -- carotid artery stenosis (CAS), subclavian artery stenosis (SAC), peripheral arterial disease (PAD), and abdominal aortic aneurysms (AAA) -- are considered. The ML methods are trained and tested on a physiologically realistic virtual patient database (VPD) containing 28,868 healthy subjects, which is adapted from the authors previous work and augmented to include the four disease forms. Six ML methods -- Naive Bayes, Logistic Regression, Support Vector Machine, Multi-layer Perceptron, Random Forests, and Gradient Boosting -- are compared with respect to classification accuracies and it is found that the tree-based methods of Random Forest and Gradient Boosting outperform other approaches. The performance of ML methods is quantified through the F1 score and computation of sensitivities and specificities. When using all the six measurements, it is found that maximum F1 scores larger than 0.9 are achieved for CAS and PAD, larger than 0.85 for SAS, and larger than 0.98 for both low- and high-severity AAAs. Corresponding sensitivities and specificities are larger than 90% for CAS and PAD, larger than 85% for SAS, and larger than 98% for both low- and high-severity AAAs. When reducing the number of measurements, it is found that the performance is degraded by less than 5% when three measurements are used, and less than 10% when only two measurements are used for classification. For AAA, it is shown that F1 scores larger than 0.85 and corresponding sensitivities and specificities larger than 85% are achievable when using only a single measurement. The results are encouraging to pursue AAA monitoring and screening through wearable devices which can reliably measure pressure or flow-rates, Changelog: Added a bullet point in the discussion (end of section 3.3)

Background The dataset from genes used to predict hepatitis C virus outcome was evaluated in a previous study using a conventional statistical methodology. Objective The aim of this study was to reanalyze this same dataset using the data mining approach in order to find models that improve the classification accuracy of the genes studied. Methods We built predictive models using different subsets of factors, selected according to their importance in predicting patient classification. We then evaluated each independent model and also a combination of them, leading to a better predictive model. Results Our data mining approach identified genetic patterns that escaped detection using conventional statistics. More specifically, the partial decision trees and ensemble models increased the classification accuracy of hepatitis C virus outcome compared with conventional methods. Conclusions Data mining can be used more extensively in biomedicine, facilitating knowledge building and management of human diseases.

Journal of medical internet research : JMIR 23(3), e26646 (2021). doi:10.2196/26646, Published by Healthcare World, Richmond, Va.

Background Consumer-oriented mobile self-diagnosis apps have been developed using undisclosed algorithms, presumably based on machine learning and other artificial intelligence (AI) technologies. The US Food and Drug Administration now discerns apps with learning AI algorithms from those with stable ones and treats the former as medical devices. To the author’s knowledge, no self-diagnosis app testing has been performed in the field of ophthalmology so far. Objective The objective of this study was to test apps that were previously mentioned in the scientific literature on a set of diagnoses in a deliberate time interval, comparing the results and looking for differences that hint at “nonlocked” learning algorithms. Methods Four apps from the literature were chosen (Ada, Babylon, Buoy, and Your.MD). A set of three ophthalmology diagnoses (glaucoma, retinal tear, dry eye syndrome) representing three levels of urgency was used to simultaneously test the apps’ diagnostic efficiency and treatment recommendations in this specialty. Two years was the chosen time interval between the tests (2018 and 2020). Scores were awarded by one evaluating physician using a defined scheme. Results Two apps (Ada and Your.MD) received significantly higher scores than the other two. All apps either worsened in their results between 2018 and 2020 or remained unchanged at a low level. The variation in the results over time indicates “nonlocked” learning algorithms using AI technologies. None of the apps provided correct diagnoses and treatment recommendations for all three diagnoses in 2020. Two apps (Babylon and Your.MD) asked significantly fewer questions than the other two (P Conclusions “Nonlocked” algorithms are used by self-diagnosis apps. The diagnostic efficiency of the tested apps seems to worsen over time, with some apps being more capable than others. Systematic studies on a wider scale are necessary for health care providers and patients to correctly assess the safety and efficacy of such apps and for correct classification by health care regulating authorities.

Background Performing systematic reviews is a time-consuming and resource-intensive process. Objective We investigated whether a machine learning system could perform systematic reviews more efficiently. Methods All systematic reviews and meta-analyses of interventional randomized controlled trials cited in recent clinical guidelines from the American Diabetes Association, American College of Cardiology, American Heart Association (2 guidelines), and American Stroke Association were assessed. After reproducing the primary screening data set according to the published search strategy of each, we extracted correct articles (those actually reviewed) and incorrect articles (those not reviewed) from the data set. These 2 sets of articles were used to train a neural network–based artificial intelligence engine (Concept Encoder, Fronteo Inc). The primary endpoint was work saved over sampling at 95% recall (WSS@95%). Results Among 145 candidate reviews of randomized controlled trials, 8 reviews fulfilled the inclusion criteria. For these 8 reviews, the machine learning system significantly reduced the literature screening workload by at least 6-fold versus that of manual screening based on WSS@95%. When machine learning was initiated using 2 correct articles that were randomly selected by a researcher, a 10-fold reduction in workload was achieved versus that of manual screening based on the WSS@95% value, with high sensitivity for eligible studies. The area under the receiver operating characteristic curve increased dramatically every time the algorithm learned a correct article. Conclusions Concept Encoder achieved a 10-fold reduction of the screening workload for systematic review after learning from 2 randomly selected studies on the target topic. However, few meta-analyses of randomized controlled trials were included. Concept Encoder could facilitate the acquisition of evidence for clinical guidelines.

Background Cancer of unknown primary (CUP), representing approximately 3-5% of all malignancies, is defined as metastatic cancer where a primary site of origin cannot be found despite a standard diagnostic workup. Because knowledge of a patient's primary cancer remains fundamental to their treatment, CUP patients are significantly disadvantaged and most have a poor survival outcome. Developing robust and accessible diagnostic methods for resolving cancer tissue of origin, therefore, has significant value for CUP patients. Methods We developed an RNA-based classifier called CUP-AI-Dx that utilizes a 1D Inception convolutional neural network (1D-Inception) model to infer a tumor's primary tissue of origin. CUP-AI-Dx was trained using the transcriptional profiles of 18,217 primary tumours representing 32 cancer types from The Cancer Genome Atlas project (TCGA) and International Cancer Genome Consortium (ICGC). Gene expression data was ordered by gene chromosomal coordinates as input to the 1D-CNN model, and the model utilizes multiple convolutional kernels with different configurations simultaneously to improve generality. The model was optimized through extensive hyperparameter tuning, including different max-pooling layers and dropout settings. For 11 tumour types, we also developed a random forest model that can classify the tumour's molecular subtype according to prior TCGA studies. The optimised CUP-AI-Dx tissue of origin classifier was tested on 394 metastatic samples from 11 tumour types from TCGA and 92 formalin-fixed paraffin-embedded (FFPE) samples representing 18 cancer types from two clinical laboratories. The CUP-AI-Dx molecular subtype was also independently tested on independent ovarian and breast cancer microarray datasets Findings CUP-AI-Dx identifies the primary site with an overall top-1-accuracy of 98.54% in cross-validation and 96.70% on a test dataset. When applied to two independent clinical-grade RNA-seq datasets generated from two different institutes from the US and Australia, our model predicted the primary site with a top-1-accuracy of 86.96% and 72.46% respectively. Interpretation The CUP-AI-Dx predicts tumour primary site and molecular subtype with high accuracy and therefore can be used to assist the diagnostic work-up of cancers of unknown primary or uncertain origin using a common and accessible genomics platform. Funding NIH R35 GM133562, NCI P30 CA034196, Victorian Cancer Agency Australia.

Significance: Melanoma is a deadly cancer that physicians struggle to diagnose early because they lack the knowledge to differentiate benign from malignant lesions. Deep machine learning approaches to image analysis offer promise but lack the transparency to be widely adopted as stand-alone diagnostics. Aim: We aimed to create a transparent machine learning technology (i.e., not deep learning) to discriminate melanomas from nevi in dermoscopy images and an interface for sensory cue integration. Approach: Imaging biomarker cues (IBCs) fed ensemble machine learning classifier (Eclass) training while raw images fed deep learning classifier training. We compared the areas under the diagnostic receiver operator curves. Results: Our interpretable machine learning algorithm outperformed the leading deep-learning approach 75% of the time. The user interface displayed only the diagnostic imaging biomarkers as IBCs. Conclusions: From a translational perspective, Eclass is better than convolutional machine learning diagnosis in that physicians can embrace it faster than black box outputs. Imaging biomarkers cues may be used during sensory cue integration in clinical screening. Our method may be applied to other image-based diagnostic analyses, including pathology and radiology.

Background Fractures as a result of osteoporosis and low bone mass are common and give rise to significant clinical, personal, and economic burden. Even after a fracture occurs, high fracture risk remains widely underdiagnosed and undertreated. Common fracture risk assessment tools utilize a subset of clinical risk factors for prediction, and often require manual data entry. Furthermore, these tools predict risk over the long term and do not explicitly provide short-term risk estimates necessary to identify patients likely to experience a fracture in the next 1-2 years. Objective The goal of this study was to develop and evaluate an algorithm for the identification of patients at risk of fracture in a subsequent 1- to 2-year period. In order to address the aforementioned limitations of current prediction tools, this approach focused on a short-term timeframe, automated data entry, and the use of longitudinal data to inform the predictions. Methods Using retrospective electronic health record data from over 1,000,000 patients, we developed Crystal Bone, an algorithm that applies machine learning techniques from natural language processing to the temporal nature of patient histories to generate short-term fracture risk predictions. Similar to how language models predict the next word in a given sentence or the topic of a document, Crystal Bone predicts whether a patient’s future trajectory might contain a fracture event, or whether the signature of the patient’s journey is similar to that of a typical future fracture patient. A holdout set with 192,590 patients was used to validate accuracy. Experimental baseline models and human-level performance were used for comparison. Results The model accurately predicted 1- to 2-year fracture risk for patients aged over 50 years (area under the receiver operating characteristics curve [AUROC] 0.81). These algorithms outperformed the experimental baselines (AUROC 0.67) and showed meaningful improvements when compared to retrospective approximation of human-level performance by correctly identifying 9649 of 13,765 (70%) at-risk patients who did not receive any preventative bone-health-related medical interventions from their physicians. Conclusions These findings indicate that it is possible to use a patient’s unique medical history as it changes over time to predict the risk of short-term fracture. Validating and applying such a tool within the health care system could enable automated and widespread prediction of this risk and may help with identification of patients at very high risk of fracture.

Background The rapid spread of COVID-19 means that government and health services providers have little time to plan and design effective response policies. It is therefore important to quickly provide accurate predictions of how vulnerable geographic regions such as counties are to the spread of this virus. Objective The aim of this study is to develop county-level prediction around near future disease movement for COVID-19 occurrences using publicly available data. Methods We estimated county-level COVID-19 occurrences for the period March 14 to 31, 2020, based on data fused from multiple publicly available sources inclusive of health statistics, demographics, and geographical features. We developed a three-stage model using XGBoost, a machine learning algorithm, to quantify the probability of COVID-19 occurrence and estimate the number of potential occurrences for unaffected counties. Finally, these results were combined to predict the county-level risk. This risk was then used as an estimated after-five-day-vulnerability of the county. Results The model predictions showed a sensitivity over 71% and specificity over 94% for models built using data from March 14 to 31, 2020. We found that population, population density, percentage of people aged >70 years, and prevalence of comorbidities play an important role in predicting COVID-19 occurrences. We observed a positive association at the county level between urbanicity and vulnerability to COVID-19. Conclusions The developed model can be used for identification of vulnerable counties and potential data discrepancies. Limited testing facilities and delayed results introduce significant variation in reported cases, which produces a bias in the model.

Purpose 3T magnetic resonance scanners have boosted clinical application of 1H‐MR spectroscopy (MRS) by offering an improved signal‐to‐noise ratio and increased spectral resolution, thereby identifying more metabolites and extending the range of metabolic information. Spectroscopic data from clinical 1.5T MR scanners has been shown to discriminate between pediatric brain tumors by applying machine learning techniques to further aid diagnosis. The purpose of this multi‐center study was to investigate the discriminative potential of metabolite profiles obtained from 3T scanners in classifying pediatric brain tumors. Methods A total of 41 pediatric patients with brain tumors (17 medulloblastomas, 20 pilocytic astrocytomas, and 4 ependymomas) were scanned across four different hospitals. Raw spectroscopy data were processed using TARQUIN. Borderline synthetic minority oversampling technique was used to correct for the data skewness. Different classifiers were trained using linear discriminative analysis, support vector machine, and random forest techniques. Results Support vector machine had the highest balanced accuracy for discriminating the three tumor types. The balanced accuracy achieved was higher than the balanced accuracy previously reported for similar multi‐center dataset from 1.5T magnets with echo time 20 to 32 ms alone. Conclusion This study showed that 3T MRS can detect key differences in metabolite profiles for the main types of childhood tumors. Magn Reson Med 79:2359–2366, 2018. © 2017 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc. on behalf of International Society for Magnetic Resonance in Medicine. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Background Semisupervised and unsupervised anomaly detection methods have been widely used in various applications to detect anomalous objects from a given data set. Specifically, these methods are popular in the medical domain because of their suitability for applications where there is a lack of a sufficient data set for the other classes. Infection incidence often brings prolonged hyperglycemia and frequent insulin injections in people with type 1 diabetes, which are significant anomalies. Despite these potentials, there have been very few studies that focused on detecting infection incidences in individuals with type 1 diabetes using a dedicated personalized health model. Objective This study aims to develop a personalized health model that can automatically detect the incidence of infection in people with type 1 diabetes using blood glucose levels and insulin-to-carbohydrate ratio as input variables. The model is expected to detect deviations from the norm because of infection incidences considering elevated blood glucose levels coupled with unusual changes in the insulin-to-carbohydrate ratio. Methods Three groups of one-class classifiers were trained on target data sets (regular days) and tested on a data set containing both the target and the nontarget (infection days). For comparison, two unsupervised models were also tested. The data set consists of high-precision self-recorded data collected from three real subjects with type 1 diabetes incorporating blood glucose, insulin, diet, and events of infection. The models were evaluated on two groups of data: raw and filtered data and compared based on their performance, computational time, and number of samples required. Results The one-class classifiers achieved excellent performance. In comparison, the unsupervised models suffered from performance degradation mainly because of the atypical nature of the data. Among the one-class classifiers, the boundary and domain-based method produced a better description of the data. Regarding the computational time, nearest neighbor, support vector data description, and self-organizing map took considerable training time, which typically increased as the sample size increased, and only local outlier factor and connectivity-based outlier factor took considerable testing time. Conclusions We demonstrated the applicability of one-class classifiers and unsupervised models for the detection of infection incidence in people with type 1 diabetes. In this patient group, detecting infection can provide an opportunity to devise tailored services and also to detect potential public health threats. The proposed approaches achieved excellent performance; in particular, the boundary and domain-based method performed better. Among the respective groups, particular models such as one-class support vector machine, K-nearest neighbor, and K-means achieved excellent performance in all the sample sizes and infection cases. Overall, we foresee that the results could encourage researchers to examine beyond the presented features into other additional features of the self-recorded data, for example, continuous glucose monitoring features and physical activity data, on a large scale.

Background As the need for sharing genomic data grows, privacy issues and concerns, such as the ethics surrounding data sharing and disclosure of personal information, are raised. Objective The main purpose of this study was to verify whether genomic data is sufficient to predict a patient's personal information. Methods RNA expression data and matched patient personal information were collected from 9538 patients in The Cancer Genome Atlas program. Five personal information variables (age, gender, race, cancer type, and cancer stage) were recorded for each patient. Four different machine learning algorithms (support vector machine, decision tree, random forest, and artificial neural network) were used to determine whether a patient's personal information could be accurately predicted from RNA expression data. Performance measurement of the prediction models was based on the accuracy and area under the receiver operating characteristic curve. We selected five cancer types (breast carcinoma, kidney renal clear cell carcinoma, head and neck squamous cell carcinoma, low-grade glioma, and lung adenocarcinoma) with large samples sizes to verify whether predictive accuracy would differ between them. We also validated the efficacy of our four machine learning models in analyzing normal samples from 593 cancer patients. Results In most samples, personal information with high genetic relevance, such as gender and cancer type, could be predicted from RNA expression data alone. The prediction accuracies for gender and cancer type, which were the best models, were 0.93-0.99 and 0.78-0.94, respectively. Other aspects of personal information, such as age, race, and cancer stage, were difficult to predict from RNA expression data, with accuracies ranging from 0.0026-0.29, 0.76-0.96, and 0.45-0.79, respectively. Among the tested machine learning methods, the highest predictive accuracy was obtained using the support vector machine algorithm (mean accuracy 0.77), while the lowest accuracy was obtained using the random forest method (mean accuracy 0.65). Gender and race were predicted more accurately than other variables in the samples. On average, the accuracy of cancer stage prediction ranged between 0.71-0.67, while the age prediction accuracy ranged between 0.18-0.23 for the five cancer types. Conclusions We attempted to predict patient information using RNA expression data. We found that some identifiers could be predicted, but most others could not. This study showed that personal information available from RNA expression data is limited and this information cannot be used to identify specific patients.

Significance: We introduce an application of machine learning trained on optical phase features of epithelial and mesenchymal cells to grade cancer cells’ morphologies, relevant to evaluation of cancer phenotype in screening assays and clinical biopsies. Aim: Our objective was to determine quantitative epithelial and mesenchymal qualities of breast cancer cells through an unbiased, generalizable, and linear score covering the range of observed morphologies. Approach: Digital holographic microscopy was used to generate phase height maps of noncancerous epithelial (Gie-No3B11) and fibroblast (human gingival) cell lines, as well as MDA-MB-231 and MCF-7 breast cancer cell lines. Several machine learning algorithms were evaluated as binary classifiers of the noncancerous cells that graded the cancer cells by transfer learning. Results: Epithelial and mesenchymal cells were classified with 96% to 100% accuracy. Breast cancer cells had scores in between the noncancer scores, indicating both epithelial and mesenchymal morphological qualities. The MCF-7 cells skewed toward epithelial scores, while MDA-MB-231 cells skewed toward mesenchymal scores. Linear support vector machines (SVMs) produced the most distinct score distributions for each cell line. Conclusions: The proposed epithelial–mesenchymal score, derived from linear SVM learning, is a sensitive and quantitative approach for detecting epithelial and mesenchymal characteristics of unknown cells based on well-characterized cell lines. We establish a framework for rapid and accurate morphological evaluation of single cells and subtle phenotypic shifts in imaged cell populations.

Summary For a continuous treatment, the generalised propensity score (GPS) is defined as the conditional density of the treatment, given covariates. GPS adjustment may be implemented by including it as a covariate in an outcome regression. Here, the unbiased estimation of the dose–response function assumes correct specification of both the GPS and the outcome‐treatment relationship. This paper introduces a machine learning method, the ‘Super Learner’, to address model selection in this context. In the two‐stage estimation approach proposed, the Super Learner selects a GPS and then a dose–response function conditional on the GPS, as the convex combination of candidate prediction algorithms. We compare this approach with parametric implementations of the GPS and to regression methods. We contrast the methods in the Risk Adjustment in Neurocritical care cohort study, in which we estimate the marginal effects of increasing transfer time from emergency departments to specialised neuroscience centres, for patients with acute traumatic brain injury. With parametric models for the outcome, we find that dose–response curves differ according to choice of specification. With the Super Learner approach to both regression and the GPS, we find that transfer time does not have a statistically significant marginal effect on the outcomes. © 2015 The Authors. Health Economics Published by John Wiley & Sons Ltd.

Summary Genomic sequences are assembled into a variable, but large number of contigs that should be scaffolded (ordered and oriented) for facilitating comparative or functional analysis. Finding scaffolding is computationally challenging due to misassemblies, inconsistent coverage across the genome and long repeats. An accurate assessment of scaffolding tools should take into account multiple locations of the same contig on the reference scaffolding rather than matching a repeat to a single best location. This makes mapping of inferred scaffoldings onto the reference a computationally challenging problem. This paper formulates the repeat-aware scaffolding evaluation problem, which is to find a mapping of the inferred scaffolding onto the reference maximizing number of correct links and proposes a scalable algorithm capable of handling large whole-genome datasets. Our novel scaffolding validation framework has been applied to assess the most of state-of-the-art scaffolding tools on the representative subset of Genome Assembly Golden-Standard Evaluations (GAGE) datasets and some novel simulated datasets. Availability and implementation The source code of this evaluation framework is available at https://github.com/mandricigor/repeat-aware. The documentation is hosted at https://mandricigor.github.io/repeat-aware. Supplementary information Supplementary data are available at Bioinformatics online.

Background: The rapid deterioration observed in the condition of some hospitalized patients can be attributed to either disease progression or imperfect triage and level of care assignment after their admission. An early warning system (EWS) to identify patients at high risk of subsequent intrahospital death can be an effective tool for ensuring patient safety and quality of care and reducing avoidable harm and costs. Objective: The aim of this study was to prospectively validate a real-time EWS designed to predict patients at high risk of inpatient mortality during their hospital episodes. Methods: Data were collected from the system-wide electronic medical record (EMR) of two acute Berkshire Health System hospitals, comprising 54,246 inpatient admissions from January 1, 2015, to September 30, 2017, of which 2.30% (1248/54,246) resulted in intrahospital deaths. Multiple machine learning methods (linear and nonlinear) were explored and compared. The tree-based random forest method was selected to develop the predictive application for the intrahospital mortality assessment. After constructing the model, we prospectively validated the algorithms as a real-time inpatient EWS for mortality. Results: The EWS algorithm scored patients’ daily and long-term risk of inpatient mortality probability after admission and stratified them into distinct risk groups. In the prospective validation, the EWS prospectively attained a c-statistic of 0.884, where 99 encounters were captured in the highest risk group, 69% (68/99) of whom died during the episodes. It accurately predicted the possibility of death for the top 13.3% (34/255) of the patients at least 40.8 hours before death. Important clinical utilization features, together with coded diagnoses, vital signs, and laboratory test results were recognized as impactful predictors in the final EWS. Conclusions: In this study, we prospectively demonstrated the capability of the newly-designed EWS to monitor and alert clinicians about patients at high risk of in-hospital death in real time, thereby providing opportunities for timely interventions. This real-time EWS is able to assist clinical decision making and enable more actionable and effective individualized care for patients’ better health outcomes in target medical facilities.

Background Metabolic rewiring allows cancer cells to sustain high proliferation rates. Thus, targeting only the cancer-specific cellular metabolism will safeguard healthy tissues. Methods We developed the very efficient FASTCORMICS RNA-seq workflow (rFASTCORMICS) to build 10,005 high-resolution metabolic models from the TCGA dataset to capture metabolic rewiring strategies in cancer cells. Colorectal cancer (CRC) was used as a test case for a repurposing workflow based on rFASTCORMICS. Findings Alternative pathways that are not required for proliferation or survival tend to be shut down and, therefore, tumours display cancer-specific essential genes that are significantly enriched for known drug targets. We identified naftifine, ketoconazole, and mimosine as new potential CRC drugs, which were experimentally validated. Interpretation The here presented rFASTCORMICS workflow successfully reconstructs a metabolic model based on RNA-seq data and successfully predicted drug targets and drugs not yet indicted for colorectal cancer. Fund This study was supported by the University of Luxembourg (IRP grant scheme; R-AGR-0755-12), the Luxembourg National Research Fund (FNR PRIDE PRIDE15/10675146/CANBIO), the Fondation Cancer (Luxembourg), the European Union‘s Horizon2020 research and innovation programme under the Marie Sklodowska- Curie grant agreement No 642295 (MEL-PLEX), and the German Federal Ministry of Education and Research (BMBF) within the project MelanomSensitivity (BMBF/BM/7643621).

Since lots of miRNA-disease associations have been verified, it is meaningful to discover more miRNA-disease associations for serving disease diagnosis and prevention of human complex diseases. However, it is not practical to identify potential associations using traditional biological experimental methods since the process is expensive and time consuming. Therefore, it is necessary to develop efficient computational methods to accomplish this task. In this work, we introduced a matrix completion model with dual Laplacian regularization (DLRMC) to infer unknown miRNA-disease associations in heterogeneous omics data. Specifically, DLRMC transformed the task of miRNA-disease association prediction into a matrix completion problem, in which the potential missing entries of the miRNA-disease association matrix were calculated, the missing association can be obtained based on the prediction scores after the completion procedure. Meanwhile, the miRNA functional similarity and the disease semantic similarity were fully exploited to serve the miRNA-disease association matrix completion by using a dual Laplacian regularization term. In the experiments, we conducted global and local Leave-One-Out Cross Validation (LOOCV) and case studies to evaluate the efficacy of DLRMC on the Human miRNA-disease associations dataset obtained from the HMDDv2.0 database. As a result, the AUCs of DLRMC is 0.9174 and 0.8289 in global LOOCV and local LOOCV, respectively, which significantly outperform a variety of previous methods. In addition, in the case studies on four significant diseases related to human health including Colon Neoplasms, Kidney neoplasms, Lymphoma and Prostate neoplasms, 90%, 92%, 92% and 94% out of the top 50 predicted miRNAs has been confirmed, respectively.

Background: Prevention and management of chronic diseases are the main goals of national health maintenance programs. Previously widely used screening tools, such as Health Risk Appraisal, are restricted in their achievement this goal due to their limitations, such as static characteristics, accessibility, and generalizability. Hypertension is one of the most important chronic diseases requiring management via the nationwide health maintenance program, and health care providers should inform patients about their risks of a complication caused by hypertension. Objective: Our goal was to develop and compare machine learning models predicting high-risk vascular diseases for hypertensive patients so that they can manage their blood pressure based on their risk level. Methods: We used a 12-year longitudinal dataset of the nationwide sample cohort, which contains the data of 514,866 patients and allows tracking of patients’ medical history across all health care providers in Korea (N=51,920). To ensure the generalizability of our models, we conducted an external validation using another national sample cohort dataset, comprising one million different patients, published by the National Health Insurance Service. From each dataset, we obtained the data of 74,535 and 59,738 patients with essential hypertension and developed machine learning models for predicting cardiovascular and cerebrovascular events. Six machine learning models were developed and compared for evaluating performances based on validation metrics. Results: Machine learning algorithms enabled us to detect high-risk patients based on their medical history. The long short-term memory-based algorithm outperformed in the within test (F1-score=.772, external test F1-score=.613), and the random forest-based algorithm of risk prediction showed better performance over other machine learning algorithms concerning generalization (within test F1-score=.757, external test F1-score=.705). Concerning the number of features, in the within test, the long short-term memory-based algorithms outperformed regardless of the number of features. However, in the external test, the random forest-based algorithm was the best, irrespective of the number of features it encountered. Conclusions: We developed and compared machine learning models predicting high-risk vascular diseases in hypertensive patients so that they may manage their blood pressure based on their risk level. By relying on the prediction model, a government can predict high-risk patients at the nationwide level and establish health care policies in advance.

Motivation: Human immunodeficiency virus (HIV) and cancer require personalized therapies owing to their inherent heterogeneous nature. For both diseases, large-scale pharmacogenomic screens of molecularly characterized samples have been generated with the hope of identifying genetic predictors of drug susceptibility. Thus, computational algorithms capable of inferring robust predictors of drug responses from genomic information are of great practical importance. Most of the existing computational studies that consider drug susceptibility prediction against a panel of drugs formulate a separate learning problem for each drug, which cannot make use of commonalities between subsets of drugs. Results: In this study, we propose to solve the problem of drug susceptibility prediction against a panel of drugs in a multitask learning framework by formulating a novel Bayesian algorithm that combines kernel-based non-linear dimensionality reduction and binary classification (or regression). The main novelty of our method is the joint Bayesian formulation of projecting data points into a shared subspace and learning predictive models for all drugs in this subspace, which helps us to eliminate off-target effects and drug-specific experimental noise. Another novelty of our method is the ability of handling missing phenotype values owing to experimental conditions and quality control reasons. We demonstrate the performance of our algorithm via cross-validation experiments on two benchmark drug susceptibility datasets of HIV and cancer. Our method obtains statistically significantly better predictive performance on most of the drugs compared with baseline single-task algorithms that learn drug-specific models. These results show that predicting drug susceptibility against a panel of drugs simultaneously within a multitask learning framework improves overall predictive performance over single-task learning approaches. Availability and implementation: Our Matlab implementations for binary classification and regression are available at https://github.com/mehmetgonen/kbmtl. Contact: mehmet.gonen@sagebase.org Supplementary Information: Supplementary data are available at Bioinformatics online.

Motivation: Somatic copy number aberrations (SCNAs) are frequent in cancer genomes, but many of these are random, passenger events. A common strategy to distinguish functional aberrations from passengers is to identify those aberrations that are recurrent across multiple samples. However, the extensive variability in the length and position of SCNAs makes the problem of identifying recurrent aberrations notoriously difficult. Results: We introduce a combinatorial approach to the problem of identifying independent and recurrent SCNAs, focusing on the key challenging of separating the overlaps in aberrations across individuals into independent events. We derive independent and recurrent SCNAs as maximal cliques in an interval graph constructed from overlaps between aberrations. We efficiently enumerate all such cliques, and derive a dynamic programming algorithm to find an optimal selection of non-overlapping cliques, resulting in a very fast algorithm, which we call RAIG (Recurrent Aberrations from Interval Graphs). We show that RAIG outperforms other methods on simulated data and also performs well on data from three cancer types from The Cancer Genome Atlas (TCGA). In contrast to existing approaches that employ various heuristics to select independent aberrations, RAIG optimizes a well-defined objective function. We show that this allows RAIG to identify rare aberrations that are likely functional, but are obscured by overlaps with larger passenger aberrations. Availability: http://compbio.cs.brown.edu/software. Contact: braphael@brown.edu Supplementary information: Supplementary data are available at Bioinformatics online.

Motivation: Alternative splicing (AS) is a regulated process that directs the generation of different transcripts from single genes. A computational model that can accurately predict splicing patterns based on genomic features and cellular context is highly desirable, both in understanding this widespread phenomenon, and in exploring the effects of genetic variations on AS. Methods: Using a deep neural network, we developed a model inferred from mouse RNA-Seq data that can predict splicing patterns in individual tissues and differences in splicing patterns across tissues. Our architecture uses hidden variables that jointly represent features in genomic sequences and tissue types when making predictions. A graphics processing unit was used to greatly reduce the training time of our models with millions of parameters. Results: We show that the deep architecture surpasses the performance of the previous Bayesian method for predicting AS patterns. With the proper optimization procedure and selection of hyperparameters, we demonstrate that deep architectures can be beneficial, even with a moderately sparse dataset. An analysis of what the model has learned in terms of the genomic features is presented. Contact: frey@psi.toronto.edu Supplementary information: Supplementary data are available at Bioinformatics online.

// Xing Chen 1 , Zhu-Hong You 2 , Gui-Ying Yan 3, 4 , Dun-Wei Gong 1 1 School of Information and Electrical Engineering, China University of Mining and Technology, Xuzhou, 221116, China 2 School of Computer Science and Technology, China University of Mining and Technology, Xuzhou, 221116, China 3 Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing, 100190, China 4 National Center for Mathematics and Interdisciplinary Sciences, Chinese Academy of Sciences, Beijing, 100190, China Correspondence to: Xing Chen, email: xingchen@amss.ac.cn Zhu-Hong You, email: zhuhongyou@gmail.com Keywords: lncRNAs, disease, cancer, random walk with restart Received: January 20, 2016 Accepted: July 06, 2016 Published: August 09, 2016 ABSTRACT In recent years, accumulating evidences have shown that the dysregulations of lncRNAs are associated with a wide range of human diseases. It is necessary and feasible to analyze known lncRNA-disease associations, predict potential lncRNA-disease associations, and provide the most possible lncRNA-disease pairs for experimental validation. Considering the limitations of traditional Random Walk with Restart (RWR), the model of Improved Random Walk with Restart for LncRNA-Disease Association prediction (IRWRLDA) was developed to predict novel lncRNA-disease associations by integrating known lncRNA-disease associations, disease semantic similarity, and various lncRNA similarity measures. The novelty of IRWRLDA lies in the incorporation of lncRNA expression similarity and disease semantic similarity to set the initial probability vector of the RWR. Therefore, IRWRLDA could be applied to diseases without any known related lncRNAs. IRWRLDA significantly improved previous classical models with reliable AUCs of 0.7242 and 0.7872 in two known lncRNA-disease association datasets downloaded from the lncRNADisease database, respectively. Further case studies of colon cancer and leukemia were implemented for IRWRLDA and 60% of lncRNAs in the top 10 prediction lists have been confirmed by recent experimental reports.

Motivation Molecular analyses suggest that myeloma is composed of distinct sub-types that have different molecular pathologies and various response rates to certain treatments. Drug responses in multiple myeloma (MM) are usually recorded as a multi-level ordinal outcome. One of the goals of drug response studies is to predict which response category any patients belong to with high probability based on their clinical and molecular features. However, as most of genes have small effects, gene-based models may provide limited predictive accuracy. In that case, methods for predicting multi-level ordinal drug responses by incorporating biological pathways are desired but have not been developed yet. Results We propose a pathway-structured method for predicting multi-level ordinal responses using a two-stage approach. We first develop hierarchical ordinal logistic models and an efficient quasi-Newton algorithm for jointly analyzing numerous correlated variables. Our two-stage approach first obtains the linear predictor (called the pathway score) for each pathway by fitting all predictors within each pathway using the hierarchical ordinal logistic approach, and then combines the pathway scores as new predictors to build a predictive model. We applied the proposed method to two publicly available datasets for predicting multi-level ordinal drug responses in MM using large-scale gene expression data and pathway information. Our results show that our approach not only significantly improved the predictive performance compared with the corresponding gene-based model but also allowed us to identify biologically relevant pathways. Availability and implementation The proposed approach has been implemented in our R package BhGLM, which is freely available from the public GitHub repository https://github.com/abbyyan3/BhGLM.

Motivation: Computational methods for the prediction of peptide-MHC binding have become an integral and essential component for candidate selection in experimental T cell epitope discovery studies. The sheer amount of published prediction methods—and often discordant reports on their performance—poses a considerable quandary to the experimentalist who needs to choose the best tool for their research. Results: With the goal to provide an unbiased, transparent evaluation of the state-of-the-art in the field, we created an automated platform to benchmark peptide-MHC class II binding prediction tools. The platform evaluates the absolute and relative predictive performance of all participating tools on data newly entered into the Immune Epitope Database (IEDB) before they are made public, thereby providing a frequent, unbiased assessment of available prediction tools. The benchmark runs on a weekly basis, is fully automated, and displays up-to-date results on a publicly accessible website. The initial benchmark described here included six commonly used prediction servers, but other tools are encouraged to join with a simple sign-up procedure. Performance evaluation on 59 data sets composed of over 10 000 binding affinity measurements suggested that NetMHCIIpan is currently the most accurate tool, followed by NN-align and the IEDB consensus method. Fil: Andreatta, Massimo. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - La Plata. Instituto de Investigaciones Biotecnológicas. Universidad Nacional de San Martín. Instituto de Investigaciones Biotecnológicas; Argentina Fil: Trolle, Thomas. Evaxion Biotech; Dinamarca Fil: Yan, Zhen. La Jolla Institute for Allergy and Immunology; Estados Unidos Fil: Greenbaum, Jason A. La Jolla Institute for Allergy and Immunology; Estados Unidos Fil: Peters, Bjoern. La Jolla Institute for Allergy and Immunology; Estados Unidos Fil: Nielsen, Morten. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - La Plata. Instituto de Investigaciones Biotecnológicas. Universidad Nacional de San Martín. Instituto de Investigaciones Biotecnológicas; Argentina. Technical University of Denmark; Dinamarca

Background: Researchers are developing methods to automatically extract clinically relevant and useful patient characteristics from raw healthcare datasets. These characteristics, often capturing essential properties of patients with common medical conditions, are called computational phenotypes. Being generated by automated or semiautomated, data-driven methods, such potential phenotypes need to be validated as clinically meaningful (or not) before they are acceptable for use in decision making. Objective: The objective of this study was to present Phenotype Instance Verification and Evaluation Tool (PIVET), a framework that uses co-occurrence analysis on an online corpus of publically available medical journal articles to build clinical relevance evidence sets for user-supplied phenotypes. PIVET adopts a conceptual framework similar to the pioneering prototype tool PheKnow-Cloud that was developed for the phenotype validation task. PIVET completely refactors each part of the PheKnow-Cloud pipeline to deliver vast improvements in speed without sacrificing the quality of the insights PheKnow-Cloud achieved. Methods: PIVET leverages indexing in NoSQL databases to efficiently generate evidence sets. Specifically, PIVET uses a succinct representation of the phenotypes that corresponds to the index on the corpus database and an optimized co-occurrence algorithm inspired by the Aho-Corasick algorithm. We compare PIVET’s phenotype representation with PheKnow-Cloud’s by using PheKnow-Cloud’s experimental setup. In PIVET’s framework, we also introduce a statistical model trained on domain expert–verified phenotypes to automatically classify phenotypes as clinically relevant or not. Additionally, we show how the classification model can be used to examine user-supplied phenotypes in an online, rather than batch, manner. Results: PIVET maintains the discriminative power of PheKnow-Cloud in terms of identifying clinically relevant phenotypes for the same corpus with which PheKnow-Cloud was originally developed, but PIVET’s analysis is an order of magnitude faster than that of PheKnow-Cloud. Not only is PIVET much faster, it can be scaled to a larger corpus and still retain speed. We evaluated multiple classification models on top of the PIVET framework and found ridge regression to perform best, realizing an average F1 score of 0.91 when predicting clinically relevant phenotypes. Conclusions: Our study shows that PIVET improves on the most notable existing computational tool for phenotype validation in terms of speed and automation and is comparable in terms of accuracy.

While novel technologies such as high-throughput screening have advanced together with significant investment by pharmaceutical companies during the past decades, the success rate for drug development has not yet been improved prompting researchers looking for new strategies of drug discovery. Drug repositioning is a potential approach to solve this dilemma. However, experimental identification and validation of potential drug targets encoded by the human genome is both costly and time-consuming. Therefore, effective computational approaches have been proposed to facilitate drug repositioning, which have proved to be successful in drug discovery. Doubtlessly, the availability of open-accessible data from basic chemical biology research and the success of human genome sequencing are crucial to develop effective in silico drug repositioning methods allowing the identification of potential targets for existing drugs. In this work, we review several chemogenomic data-driven computational algorithms with source codes publicly accessible for predicting drug–target interactions (DTIs). We organize these algorithms by model properties and model evolutionary relationships. We re-implemented five representative algorithms in R programming language, and compared these algorithms by means of mean percentile ranking, a new recall-based evaluation metric in the DTI prediction research field. We anticipate that this review will be objective and helpful to researchers who would like to further improve existing algorithms or need to choose appropriate algorithms to infer potential DTIs in the projects. The source codes for DTI predictions are available at: https://github.com/minghao2016/chemogenomicAlg4DTIpred.

OBJECTIVE: To train a generic deep learning software (DLS) to classify breast cancer on ultrasound images and to compare its performance to human readers with variable breast imaging experience. METHODS: In this retrospective study, all breast ultrasound examinations from January 1, 2014 to December 31, 2014 at our institution were reviewed. Patients with post-surgical scars, initially indeterminate, or malignant lesions with histological diagnoses or 2-year follow-up were included. The DLS was trained with 70% of the images, and the remaining 30% were used to validate the performance. Three readers with variable expertise also evaluated the validation set (radiologist, resident, medical student). Diagnostic accuracy was assessed with a receiver operating characteristic analysis. RESULTS: 82 patients with malignant and 550 with benign lesions were included. Time needed for training was 7 min (DLS). Evaluation time for the test data set were 3.7 s (DLS) and 28, 22 and 25 min for human readers (decreasing experience). Receiver operating characteristic analysis revealed non-significant differences (p-values 0.45–0.47) in the area under the curve of 0.84 (DLS), 0.88 (experienced and intermediate readers) and 0.79 (inexperienced reader). CONCLUSION: DLS may aid diagnosing cancer on breast ultrasound images with an accuracy comparable to radiologists, and learns better and faster than a human reader with no prior experience. Further clinical trials with dedicated algorithms are warranted. ADVANCES IN KNOWLEDGE: DLS can be trained classify cancer on breast ultrasound images high accuracy even with comparably few training cases. The fast evaluation speed makes real-time image analysis feasible.

Background Automatic early detection of acromegaly is theoretically possible from facial photographs, which can lessen the prevalence and increase the cure probability. Methods In this study, several popular machine learning algorithms were used to train a retrospective development dataset consisting of 527 acromegaly patients and 596 normal subjects. We firstly used OpenCV to detect the face bounding rectangle box, and then cropped and resized it to the same pixel dimensions. From the detected faces, locations of facial landmarks which were the potential clinical indicators were extracted. Frontalization was then adopted to synthesize frontal facing views to improve the performance. Several popular machine learning methods including LM, KNN, SVM, RT, CNN, and EM were used to automatically identify acromegaly from the detected facial photographs, extracted facial landmarks, and synthesized frontal faces. The trained models were evaluated using a separate dataset, of which half were diagnosed as acromegaly by growth hormone suppression test. Results The best result of our proposed methods showed a PPV of 96%, a NPV of 95%, a sensitivity of 96% and a specificity of 96%. Conclusions Artificial intelligence can automatically early detect acromegaly with a high sensitivity and specificity., Highlights • An automatic and handy diagnosis system for acromegaly was developed. • Several popular machine learning methods including LM, KNN, SVM, RT, CNN, and EM were used to automatically identify acromegaly from the detected facial photographs, extracted facial landmarks, and synthesized frontal faces. • The algorithm allows medical practitioners and patients to proactively track face changes and detect acromegaly earlier and automatically, thus to facilitate cures and increase the likelihood of preventing irreversible complications of excessive secretion of growth hormone. We developed an automatic and handy diagnosis system for acromegaly, which can allow medical practitioners and patients to proactively track face changes and detect acromegaly earlier and automatically, thus to facilitate cures and increase the likelihood of preventing irreversible complications of excessive secretion of growth hormone.

Motivation: An important aspect of infectious disease research involves understanding the differences and commonalities in the infection mechanisms underlying various diseases. Systems biology-based approaches study infectious diseases by analyzing the interactions between the host species and the pathogen organisms. This work aims to combine the knowledge from experimental studies of host–pathogen interactions in several diseases to build stronger predictive models. Our approach is based on a formalism from machine learning called ‘multitask learning’, which considers the problem of building models across tasks that are related to each other. A ‘task’ in our scenario is the set of host–pathogen protein interactions involved in one disease. To integrate interactions from several tasks (i.e. diseases), our method exploits the similarity in the infection process across the diseases. In particular, we use the biological hypothesis that similar pathogens target the same critical biological processes in the host, in defining a common structure across the tasks. Results: Our current work on host–pathogen protein interaction prediction focuses on human as the host, and four bacterial species as pathogens. The multitask learning technique we develop uses a task-based regularization approach. We find that the resulting optimization problem is a difference of convex (DC) functions. To optimize, we implement a Convex–Concave procedure-based algorithm. We compare our integrative approach to baseline methods that build models on a single host–pathogen protein interaction dataset. Our results show that our approach outperforms the baselines on the training data. We further analyze the protein interaction predictions generated by the models, and find some interesting insights. Availability: The predictions and code are available at: http://www.cs.cmu.edu/∼mkshirsa/ismb2013_paper320.html Contact: j.klein-seetharaman@warwick.ac.uk Supplementary information: Supplementary data are available at Bioinformatics online.

Purpose Classification of pediatric brain tumors from 1H‐magnetic resonance spectroscopy (MRS) can aid diagnosis and management of brain tumors. However, varied incidence of the different tumor types leads to imbalanced class sizes and introduces difficulties in classifying rare tumor groups. This study assessed different imbalanced multiclass learning techniques and compared the use of complete spectra and quantified metabolite profiles for classification of three main childhood brain tumor types. Methods Single‐voxel, Short echo time MRS data were collected from 90 patients with pilocytic astrocytoma (n = 42), medulloblastoma (n = 38), or ependymoma (n = 10). Both spectra and metabolite profiles were used to develop the learning algorithms. The borderline synthetic minority oversampling technique and AdaboostM1 were used to correct for the skewed distribution. Classifiers were trained using five different pattern recognition algorithms. Results Use of imbalanced learning techniques improved the balanced accuracy rate (BAR) of all classification methods (average BAR over all classification methods for spectra: oversampled data = 0.81, original = 0.63, P

Motivation Genomic networks represent a complex map of molecular interactions which are descriptive of the biological processes occurring in living cells. Identifying the small over-represented circuitry patterns in these networks helps generate hypotheses about the functional basis of such complex processes. Network motif discovery is a systematic way of achieving this goal. However, a reliable network motif discovery outcome requires generating random background networks which are the result of a uniform and independent graph sampling method. To date, there has been no method to numerically evaluate whether any network motif discovery algorithm performs as intended on realistically sized datasets—thus it was not possible to assess the validity of resulting network motifs. Results In this work, we present IndeCut, the first method to date that characterizes network motif finding algorithm performance in terms of uniform sampling on realistically sized networks. We demonstrate that it is critical to use IndeCut prior to running any network motif finder for two reasons. First, IndeCut indicates the number of samples needed for a tool to produce an outcome that is both reproducible and accurate. Second, IndeCut allows users to choose the tool that generates samples in the most independent fashion for their network of interest among many available options. Availability and implementation The open source software package is available at https://github.com/megrawlab/IndeCut. Supplementary information Supplementary data are available at Bioinformatics online.

Speech understanding in noisy environments is still one of the major challenges for cochlear implant (CI) users in everyday life. We evaluated a speech enhancement algorithm based on neural networks (NNSE) for improving speech intelligibility in noise for CI users. The algorithm decomposes the noisy speech signal into time-frequency units, extracts a set of auditory-inspired features and feeds them to the neural network to produce an estimation of which frequency channels contain more perceptually important information (higher signal-to-noise ratio, SNR). This estimate is used to attenuate noise-dominated and retain speech-dominated CI channels for electrical stimulation, as in traditional n-of-m CI coding strategies. The proposed algorithm was evaluated by measuring the speech-in-noise performance of 14 CI users using three types of background noise. Two NNSE algorithms were compared: a speaker-dependent algorithm, that was trained on the target speaker used for testing, and a speaker-independent algorithm, that was trained on different speakers. Significant improvements in the intelligibility of speech in stationary and fluctuating noises were found relative to the unprocessed condition for the speaker-dependent algorithm in all noise types and for the speaker-independent algorithm in 2 out of 3 noise types. The NNSE algorithms used noise-specific neural networks that generalized to novel segments of the same noise type and worked over a range of SNRs. The proposed algorithm has the potential to improve the intelligibility of speech in noise for CI users while meeting the requirements of low computational complexity and processing delay for application in CI devices., Highlights • An algorithm for improving speech understanding in noise for cochlear implant users is evaluated. • Significant improvements were found for stationary and non-stationary noise types. • It generalizes to a novel speaker and works over a range of signal-to-noise ratios. • The small algorithmic delay makes it suitable for real-time application.

Motivation: Time-to-event regression models are a critical tool for associating survival time outcomes with molecular data. Despite mounting evidence that genetic subgroups of the same clinical disease exist, little attention has been given to exploring how this heterogeneity affects time-to-event model building and how to accommodate it. Methods able to diagnose and model heterogeneity should be valuable additions to the biomarker discovery toolset. Results: We propose a mixture of survival functions that classifies subjects with similar relationships to a time-to-event response. This model incorporates multivariate regression and model selection and can be fit with an expectation maximization algorithm, we call Cox-assisted clustering. We illustrate a likely manifestation of genetic heterogeneity and demonstrate how it may affect survival models with little warning. An application to gene expression in ovarian cancer DNA repair pathways illustrates how the model may be used to learn new genetic subsets for risk stratification. We explore the implications of this model for censored observations and the effect on genomic predictors and diagnostic analysis. Availability and implementation: R implementation of CAC using standard packages is available at https://gist.github.com/programeng/8620b85146b14b6edf8f Data used in the analysis are publicly available. Contact: kevin.eng@roswellpark.org Supplementary information: Supplementary data are available at Bioinformatics online.