Search

There has been an exponential growth of artificial intelligence (AI) and machine learning (ML) publications aimed at advancing our understanding of atrial fibrillation (AF), which has been mainly driven by the confluence of two factors: the advances in deep neural networks (DeepNNs) and the availability of large, open access databases. It is observed that most of the attention has centred on applying ML for dvsetecting AF, particularly using electrocardiograms (ECGs) as the main data modality. Nearly a third of them used DeepNNs to minimize or eliminate the need for transforming the ECGs to extract features prior to ML modelling; however, we did not observe a significant advantage in following this approach. We also found a fraction of studies using other data modalities, and others centred in aims, such as risk prediction, AF management, and others. From the clinical perspective, AI/ML can help expand the utility of AF detection and risk prediction, especially for patients with additional comorbidities. The use of AI/ML for detection and risk prediction into applications and smart mobile health (mHealth) technology would enable ‘real time’ dynamic assessments. AI/ML could also adapt to treatment changes over time, as well as incident risk factors. Incorporation of a dynamic AI/ML model into mHealth technology would facilitate ‘real time’ assessment of stroke risk, facilitating mitigation of modifiable risk factors (e.g. blood pressure control). Overall, this would lead to an improvement in clinical care for patients with AF., Graphical Abstract

Significance Machine learning (ML) is the branch of artificial intelligence (AI) that develops computational systems that learn from experience. In supervised ML, the ML system generalizes from labelled examples to learn a model that can predict the labels of unseen examples. Examples are generally represented using features that directly describe the examples. For instance, in drug design, ML uses features that describe molecular shape and so on. In cases where there are multiple related ML problems, it is possible to use a different type of feature: predictions made about the examples by ML models learned on other problems. We call this transformational ML. We show that this results in better predictions and improved understanding when applied to scientific problems., Almost all machine learning (ML) is based on representing examples using intrinsic features. When there are multiple related ML problems (tasks), it is possible to transform these features into extrinsic features by first training ML models on other tasks and letting them each make predictions for each example of the new task, yielding a novel representation. We call this transformational ML (TML). TML is very closely related to, and synergistic with, transfer learning, multitask learning, and stacking. TML is applicable to improving any nonlinear ML method. We tested TML using the most important classes of nonlinear ML: random forests, gradient boosting machines, support vector machines, k-nearest neighbors, and neural networks. To ensure the generality and robustness of the evaluation, we utilized thousands of ML problems from three scientific domains: drug design, predicting gene expression, and ML algorithm selection. We found that TML significantly improved the predictive performance of all the ML methods in all the domains (4 to 50% average improvements) and that TML features generally outperformed intrinsic features. Use of TML also enhances scientific understanding through explainable ML. In drug design, we found that TML provided insight into drug target specificity, the relationships between drugs, and the relationships between target proteins. TML leads to an ecosystem-based approach to ML, where new tasks, examples, predictions, and so on synergistically interact to improve performance. To contribute to this ecosystem, all our data, code, and our ∼50,000 ML models have been fully annotated with metadata, linked, and openly published using Findability, Accessibility, Interoperability, and Reusability principles (∼100 Gbytes).

© The Author(s) 2019. The goal of quantitative structure activity relationship (QSAR) learning is to learn a function that, given the structure of a small molecule (a potential drug), outputs the predicted activity of the compound. We employed multi-task learning (MTL) to exploit commonalities in drug targets and assays. We used datasets containing curated records about the activity of specific compounds on drug targets provided by ChEMBL. Totally, 1091 assays have been analysed. As a baseline, a single task learning approach that trains random forest to predict drug activity for each drug target individually was considered. We then carried out feature-based and instance-based MTL to predict drug activities. We introduced a natural metric of evolutionary distance between drug targets as a measure of tasks relatedness. Instance-based MTL significantly outperformed both, feature-based MTL and the base learner, on 741 drug targets out of 1091. Feature-based MTL won on 179 occasions and the base learner performed best on 171 drug targets. We conclude that MTL QSAR is improved by incorporating the evolutionary distance between targets. These results indicate that QSAR learning can be performed effectively, even if little data is available for specific drug targets, by leveraging what is known about similar drug targets. This research was funded by the Engineering and Physical Sciences Research Council (EPSRC) grant EP/K030469/1. NS would like to thank the EU PhenoM-eNal project (Horizon 2020, 654241)

This paper discusses the concepts of interpretability and explainability and outlines desiderata for robust interpretability. It then describes a neural network model that meets all criteria, with the addition of global faithfulness.

Neural networks are frequently applied to medical data. We describe how complex and imbalanced data can be modelled with simple but accurate neural networks that are transparent to the user. In the case of a data set on cervical cancer with 753 observations excluding, missing values, and 32 covariates, with a prevalence of 73 cases (9.69%), we explain how model selection can be applied to the Multi-Layer Perceptron (MLP) by deriving a representation using a General Additive Neural Network.

Being able to predict the activity of chemical compounds against a drug target (e.g. a protein) in the preliminary stages of drug development is critical. In drug discovery, this is known as Quantitative Structure Activity Relationships (QSARs). Datasets for QSARs are often ill-posed for traditional machine learning to provide meaningful insights (e.g. very high dimensionality). Here, we propose a multi-task learning (MTL) approach to enrich the original QSAR datasets with the hope of improving overall QSAR performance. The proposed approach, henceforth named MTL-AT, increases the size of the useable data by the use of an assistant task: a supplementary dataset formed by compounds automatically extracted from other possibly related tasks. The main novelty in our MTL-AT approach is the addition of control for data leakage. We tested MTL-AT in two drug discovery scenarios: 1) using 100 unrelated QSAR datasets, and 2) using 20 QSAR datasets that are related to the same protein family. Results were compared against equivalent single-task approach (STL). MTL-AT outperformed STL in 45 tasks of scenario 1, and in 12 tasks of scenario 2. The best overall method appears to be MTL-AT on both scenarios, with the small datasets yielded the best performance improvement from using multi-task learning. These results show that implementing multi-task learning with QSAR data has promise, but more investigation is required to test its applicability to certain features in datasets to make it suitable for widespread use in the drug discovery area. To the best of our knowledge, this is the first study that benchmarks the use of MTL on a large number of small datasets, which represents a more realistic scenario in drug development.

© 2020, Springer Nature Switzerland AG. The aim of this paper is to build a computer based clinical decision support tool using a semi-supervised framework, the Fisher Information Network (FIN), for visualization of a set of mammographic images. The FIN organizes the images into a similarity network from which, for any new image, reference images that are closely related can be identified. This enables clinicians to review not just the reference images but also ancillary information e.g. about response to therapy. The Fisher information metric defines a Riemannian space where distances reflect similarity with respect to a given probability distribution. This metric is informed about generative properties of data, and hence assesses the importance of directions in space of parameters. It automatically performs feature relevance detection. This approach focusses on the interpretability of the model from the standpoint of the clinical user. Model predictions were validated using the prevalence of classes in each of the clusters identified by the FIN.

Accurate and reliable urban water demand prediction is imperative for providing the basis to design, operate, and manage water system, especially under the scarcity of the natural water resources. A new methodology combining discrete wavelet transform (DWT) with an adaptive neuro-fuzzy inference system (ANFIS) is proposed to predict monthly urban water demand based on several intervals of historical water consumption. This ANFIS model is evaluated against a hybrid crow search algorithm and artificial neural network (CSA-ANN), since these methods have been successfully used recently to tackle a range of engineering optimization problems. The study outcomes reveal that 1) data preprocessing is essential for denoising raw time series and choosing the model inputs to render the highest model performance, 2) both methodologies, ANFIS and CSA-ANN, are statistically equivalent and capable of accurately predicting monthly urban water demand with high accuracy based on several statistical metric measures such as coefficient of efficiency (0.974, 0.971, respectively). This study could help policymakers to manage extensions of urban water system in response to the increasing demand with low risk related to a decision.

This case study benchmarks a range of statistical and machine learning methods relevant to computer-based decision support in clinical medicine, focusing on the diagnosis of knee osteoarthritis at first presentation. The methods, comprising logistic regression, Multilayer Perceptron (MLP), Chi-square Automatic Interaction Detector (CHAID) and Classification and Regression Trees (CART), are applied to a public domain database, the Osteoarthritis Initiative (OAI), a 10 year longitudinal study starting in 2002 (n = 4,796). In this real-world application, it is shown that logistic regression is comparable with the neural networks and decision trees for discrimination of positive diagnosis on this data set. This is likely because of weak non-linearities among high levels of noise. After comparing the explanations provided by the different methods, it is concluded that the interpretability of the risk score index provided by logistic regression is expressed in a form that most naturally integrates with clinical reasoning. The reason for this is that it gives a statistical assessment of the weight of evidence for making the diagnosis, so providing a direction for future research to improve explanation of generic non-linear models.

We investigate the learning of quantitative structure activity relationships (QSARs) as a case-study of meta-learning. This application area is of the highest societal importance, as it is a key step in the development of new medicines. The standard QSAR learning problem is: given a target (usually a protein) and a set of chemical compounds (small molecules) with associated bioactivities (e.g. inhibition of the target), learn a predictive mapping from molecular representation to activity. Although almost every type of machine learning method has been applied to QSAR learning there is no agreed single best way of learning QSARs, and therefore the problem area is well-suited to meta-learning. We first carried out the most comprehensive ever comparison of machine learning methods for QSAR learning: 18 regression methods, 6 molecular representations, applied to more than 2,700 QSAR problems. (These results have been made publicly available on OpenML and represent a valuable resource for testing novel meta-learning methods.) We then investigated the utility of algorithm selection for QSAR problems. We found that this meta-learning approach outperformed the best individual QSAR learning method (random forests using a molecular fingerprint representation) by up to 13%, on average. We conclude that meta-learning outperforms base-learning methods for QSAR learning, and as this investigation is one of the most extensive ever comparisons of base and meta-learning methods ever made, it provides evidence for the general effectiveness of meta-learning over base-learning., Comment: 33 pages and 15 figures. Manuscript accepted for publication in Machine Learning Journal. This is the author's pre-print version

[EN] Automatic brain tumor classification by MRS has been under development for more than a decade. Nonetheless, to our knowledge, there are no published evaluations of predictive models with unseen cases that are subsequently acquired in different centers. The multicenter eTUMOUR project (2004-2009), which builds upon previous expertise from the INTERPRET project (2000-2002) has allowed such an evaluation to take place. A total of 253 pairwise classifiers for glioblastoma, meningioma, metastasis, and low-grade glial diagnosis were inferred based on 211 SV short TE INTERPRET MR spectra obtained at 1.5 T (PRESS or STEAM, 20-32 ms) and automatically pre-processed. Afterwards, the classifiers were tested with 97 spectra, which were subsequently compiled during eTUMOUR. In our results based on subsequently acquired spectra, accuracies of around 90% were achieved for most of the pairwise discrimination problems. The exception was for the glioblastoma versus metastasis discrimination, which was below 78%. A more clear definition of metastases may be obtained by other approaches, such as MRSI + MRI. The prediction of the tumor type of in-vivo MRS is possible using classifiers developed from previously acquired data, in different hospitals with different instrumentation under the same acquisition protocols. This methodology may find application for assisting in the diagnosis of new brain tumor cases and for the quality control of multicenter MRS databases., We would like to thank the INTERPRET and eTUMOUR partners for providing data, particularly, Carles Majos (IDI-Bellvitge), John Griffiths and Franklyn Howe (SGUL), Arend Heerschap (RU), Witold Gajewicz (MUL), Jorge Calvar (FLENI), and Antoni Capdevila (H. de Sant Joan de Deu). This work was partially funded by the European Commission: eTUMOUR (contract no. FP62002-LIFESCIHEALTH 503094), the HEALTHAGENTS EC project (HEALTHAGENTS) (contract no. FP6-2005-IST 027213), BIOPATTERN (contract no. FP6-2002-IST 508803). The authors appreciate the suggestions from the reviewers that have improved the discussion presented in this work. We also thank the following for their contributions: Programa de Apoyo a la Investigacion y Desarrollo, PAID-00-06 UPV; Research Council KUL: GOA-AMBioRICS, Centers-of-excellence optimisation; Belgian Federal Government: DWTC, IUAPV P6/04 (DYSCO 2007-2011); the following participants acknowledge the following: JVR acknowledges to Programa Torres Quevedo from Ministerio de Educacion y Ciencia, co-founded by the European Social Fund (PTQ05-02-03386). JL is a PhD student supported by an IWT grant. DM is supported by the Ministerio de Educacion y Ciencia del Gobierno de Espa a for a Ramon y Cajal 2006 Contract. BC and CA gratefully acknowledge the Ministerio de Educacion y Ciencia del Gobierno de Espa a (BC: SAF2004-06297 and SAF2007-6547; CA: SAF2005-03650). CIBER-BBN is an initiative of the "Instituto de Salud Carlos III" (ISCIII), Spain

Background SpectraClassifier (SC) is a Java solution for designing and implementing Magnetic Resonance Spectroscopy (MRS)-based classifiers. The main goal of SC is to allow users with minimum background knowledge of multivariate statistics to perform a fully automated pattern recognition analysis. SC incorporates feature selection (greedy stepwise approach, either forward or backward), and feature extraction (PCA). Fisher Linear Discriminant Analysis is the method of choice for classification. Classifier evaluation is performed through various methods: display of the confusion matrix of the training and testing datasets; K-fold cross-validation, leave-one-out and bootstrapping as well as Receiver Operating Characteristic (ROC) curves. Results SC is composed of the following modules: Classifier design, Data exploration, Data visualisation, Classifier evaluation, Reports, and Classifier history. It is able to read low resolution in-vivo MRS (single-voxel and multi-voxel) and high resolution tissue MRS (HRMAS), processed with existing tools (jMRUI, INTERPRET, 3DiCSI or TopSpin). In addition, to facilitate exchanging data between applications, a standard format capable of storing all the information needed for a dataset was developed. Each functionality of SC has been specifically validated with real data with the purpose of bug-testing and methods validation. Data from the INTERPRET project was used. Conclusions SC is a user-friendly software designed to fulfil the needs of potential users in the MRS community. It accepts all kinds of pre-processed MRS data types and classifies them semi-automatically, allowing spectroscopists to concentrate on interpretation of results with the use of its visualisation tools.

Generative Topographic Mapping (GTM) is a latent variable model that, in its original version, was conceived to provide clustering and visualization of multivariate, realvalued, i.i.d. data. It was also extended to deal with noni-i.i.d. data such as multivariate time series in a variant called GTM Through Time (GTM-TT), defined as a constrained Hidden Markov Model (HMM). In this paper, we provide the theoretical foundations of the reformulation of GTM-TT within the Variational Bayesian framework and provide an illustrative example of its application. This approach handles the presence of noise in the time series, helping to avert the problem of data overfitting.

Generative Topographic Mapping (GTM) is a manifold learning model for the simultaneous visualization and clustering of multivariate data. It was originally formulated as a constrained mixture of distributions, for which the adaptive parameters were determined by Maximum Likelihood (ML), using the Expectation-Maximization (EM) algorithm. In this formulation, GTM is prone to data overfitting unless a regularization mechanism is included. The theoretical principles of Variational GTM, an approximate method that provides a full Bayesian treatment to a Gaussian Process (GP)-based variation of the GTM, were recently introduced as alternative way to control data overfitting. In this paper we assess in some detail the generalization capabilities of Variational GTM and compare them with those of alternative regularization approaches in terms of test log-likelihood, using several artificial and real datasets.

Most of the existing research on multivariate time series concerns supervised forecasting problems. In comparison, little research has been devoted to their exploration through unsupervised clustering and visualization. In this paper, the capabilities of Generative Topographic Mapping Through Time, a model with foundations in probability theory, that performs simultaneous time series clustering and visualization, are assessed in detail. Focus is placed on the visualization of the evolution of signal regimes and the exploration of sudden transitions, for which a novel identification index is defined. The interpretability of time series clustering results may become extremely difficult, even in exploratory visualization, for high dimensional datasets. Here, we define and test an unsupervised time series relevance determination method, fully integrated in the Generative Topographic Mapping Through Time model, that can be used as a basis for time series selection. This method should ease the interpretation of time series clustering results.

Most of the existing research on multivariate time series concerns supervised forecasting problems. In comparison, little research has been devoted to unsupervised methods for the visual exploration of this type of data. The interpretability of time series clustering results may be difficult, even in exploratory visualization, for high dimensional datasets. In this paper, we define and test an unsupervised time series relevance determination method for Generative Topographic Mapping Through Time, a topology-constrained Hidden Markov Model that performs simultaneous time series data clustering and visualization. This relevance determination method can be used as a basis for time series selection, and should ease the interpretation of the time series clustering results.

The incompleteness of data is a most common source of uncertainty in real-world Data Mining applications. The management of this uncertainty is, therefore, a task of paramount importance for the data analyst. Many methods have been developed for missing data imputation, but few of them approach the problem of imputation as part of a general data density estimation scheme. Amongst the latter, a method for imputing and visualizing multivariate missing data using Generative Topographic Mapping was recently presented. This model and some of its extensions are tested under different experimental conditions. Its performance is compared to that of other missing data imputation techniques, thus assessing its relative capabilities and limitations.

Time-dependent natural phenomena and artificial processes can often be quantitatively expressed as multivariate time series (MTS). As in any other process of knowledge extraction from data, the analyst can benefit from the exploration of the characteristics of MTS through data visualization. This visualization often becomes difficult to interpret when MTS are modelled using nonlinear techniques. Despite their flexibility, nonlinear models can be rendered useless if such interpretability is lacking. In this brief paper, we model MTS using Variational Bayesian Generative Topographic Mapping Through Time (VB-GTM-TT), a variational Bayesian variant of a constrained hidden Markov model of the manifold learning family defined for MTS visualization. We aim to increase its interpretability by taking advantage of two results of the probabilistic definition of the model: the explicit estimation of probabilities of transition between states described in the visualization space and the quantification of the nonlinear mapping distortion.

Background Magnetic resonance spectroscopy provides metabolic information about living tissues in a non-invasive way. However, there are only few multi-centre clinical studies, mostly performed on a single scanner model or data format, as there is no flexible way of documenting and exchanging processed magnetic resonance spectroscopy data in digital format. This is because the DICOM standard for spectroscopy deals with unprocessed data. This paper proposes a plugin tool developed for jMRUI, namely jMRUI2XML, to tackle the latter limitation. jMRUI is a software tool for magnetic resonance spectroscopy data processing that is widely used in the magnetic resonance spectroscopy community and has evolved into a plugin platform allowing for implementation of novel features. Results jMRUI2XML is a Java solution that facilitates common preprocessing of magnetic resonance spectroscopy data across multiple scanners. Its main characteristics are: 1) it automates magnetic resonance spectroscopy preprocessing, and 2) it can be a platform for outputting exchangeable magnetic resonance spectroscopy data. The plugin works with any kind of data that can be opened by jMRUI and outputs in extensible markup language format. Data processing templates can be generated and saved for later use. The output format opens the way for easy data sharing- due to the documentation of the preprocessing parameters and the intrinsic anonymization - for example for performing pattern recognition analysis on multicentre/multi-manufacturer magnetic resonance spectroscopy data. Conclusions jMRUI2XML provides a self-contained and self-descriptive format accounting for the most relevant information needed for exchanging magnetic resonance spectroscopy data in digital form, as well as for automating its processing. This allows for tracking the procedures the data has undergone, which makes the proposed tool especially useful when performing pattern recognition analysis. Moreover, this work constitutes a first proposal for a minimum amount of information that should accompany any magnetic resonance processed spectrum, towards the goal of achieving better transferability of magnetic resonance spectroscopy studies. Electronic supplementary material The online version of this article (doi:10.1186/s12859-015-0796-5) contains supplementary material, which is available to authorized users.