Search

Mycobacterium tuberculosis can enter a dormant state enabling it to establish latent infections and to become tolerant to antibacterial drugs. Dormant bacteria’s physiology and the mechanism(s) used by bacteria to enter dormancy during infection are still unknown due to the lack of reliable animal models.

Motivation Recently, single-cell RNA-seq (scRNA-seq) data have been used to study cellular communication. Most bioinformatics methods infer only the intercellular signaling between groups of cells, mainly exploiting ligand–receptor expression levels. Only few methods consider the entire intercellular + intracellular signaling, mainly inferring lists/networks of signaling involved genes. Results Here, we present scSeqComm, a computational method to identify and quantify the evidence of ongoing intercellular and intracellular signaling from scRNA-seq data, and at the same time providing a functional characterization of the inferred cellular communication. The possibility to quantify the evidence of ongoing communication assists the prioritization of the results, while the combined evidence of both intercellular and intracellular signaling increase the reliability of inferred communication. The application to a scRNA-seq dataset of tumor microenvironment, the agreement with independent bioinformatics analysis, the validation using spatial transcriptomics data and the comparison with state-of-the-art intercellular scoring schemes confirmed the robustness and reliability of the proposed method. Availability and implementation scSeqComm R package is freely available at https://gitlab.com/sysbiobig/scseqcomm and https://sysbiobig.dei.unipd.it/software/#scSeqComm. Submitted software version and test data are available in Zenodo, at https://dx.doi.org/10.5281/zenodo.5833298. Supplementary information Supplementary data are available at Bioinformatics online.

Protein turnover rate is finely regulated through intracellular mechanisms and signals that are still incompletely understood but that are essential for the correct function of cellular processes. Indeed, a dysfunctional proteostasis often impacts the cell’s ability to remove unfolded, misfolded, degraded, non-functional, or damaged proteins. Thus, altered cellular mechanisms controlling protein turnover impinge on the pathophysiology of many diseases, making the study of protein synthesis and degradation rates an important step for a more comprehensive understanding of these pathologies. In this manuscript, we describe the application of a dynamic-SILAC approach to study the turnover rate and the abundance of proteins in a cellular model of diabetic nephropathy. We estimated protein half-lives and relative abundance for thousands of proteins, several of which are characterized by either an altered turnover rate or altered abundance between diabetic nephropathic subjects and diabetic controls. Many of these proteins were previously shown to be related to diabetic complications and represent therefore, possible biomarkers or therapeutic targets. Beside the aspects strictly related to the pathological condition, our data also represent a consistent compendium of protein half-lives in human fibroblasts and a rich source of important information related to basic cell biology.

Human cellular reprogramming to induced pluripotency is still an inefficient process and this has long hindered the study of the role of critical intermediate stages. We take advantage of high efficiency reprogramming in microfluidics and temporal multi-omics to identify and resolve distinct sub-populations and their interactions. The combination of secretome analysis and single-cell transcriptomics shows functional extrinsic pathways of protein communication between reprogramming sub-populations and the re-shaping of a permissive extracellular environment. We pinpointed the HGF/MET/STAT3 axis as a potent enhancer of reprogramming, which acts via HGF accumulation within the confined system of microfluidics, and in conventional dishes needs to be supplied exogenously to enhance efficiency. Our data integrate the notion of human cellular reprogramming as a transcription factor-driven process, with the concept that it is deeply dependent on extracellular context and cell population determinants.

Aim We aimed to compare cardiovascular outcomes of patients with type 2 diabetes (T2D) who initiated GLP-1 receptor agonists (GLP-1RA) or basal insulin (BI) under routine care. Methods We accessed the administrative claims database of the Veneto Region (Italy) to identify new users of GLP-1RA or BI in 2014–2018. Propensity score matching (PSM) was implemented to obtain two cohorts of patients with superimposable characteristics. The primary endpoint was the 3-point major adverse cardiovascular events (3P-MACE). Secondary endpoints included 3P-MACE components, hospitalization for heart failure, revascularizations, and adverse events. Results From a background population of 5,242,201 citizens, 330,193 were identified as having diabetes. PSM produced two very well matched cohorts of 4063 patients each, who initiated GLP-1RA or BI after an average of 2.5 other diabetes drug classes. Patients were 63-year-old and only 15% had a baseline history of cardiovascular disease. During a median follow-up of 24 months in the intention-to-treat analysis, 3P-MACE occurred less frequently in the GLP-1RA cohort (HR versus BI 0.59; 95% CI 0.50–0.71; p Conclusions Patients with T2D who initiated a GLP-1RA experienced far better cardiovascular outcomes than did matched patients who initiated a BI in the same healthcare system. These finding supports prioritization of GLP-1RA as the first injectable regimen for the management of T2D.

Background: The rapid growing proportion of elderly people on world’s population poses several challenges, especially related to public health. Among these, the increase in chronic diseases can be noted, which recent evidence attributes also to the lowering of the subjective assessment of wellbeing. Therefore, being able to predict changes in wellbeing could be key to manage this problem. In this study, we investigate for the first time the possibility of developing a purely predictive model of perceived wellbeing for population aged over 50. Methods: Data from the European SHARE project were used, specifically the demographic, health, social and financial variables of 9649 subjects. The subjective wellbeing was measured through the CASP-12 scale. Study outcome was defined binary, i.e., worsening/not worsening of the variation of CASP-12 in2 years. Logistic regression, logistic regression with LASSO regularization, and random forest were considered as candidate models. Performance was assessed in terms of accuracy in correctly predicting the outcome and area under the curve(AUC). Results: The best performing model was the logistic regression, achieving accuracy = 0.659 and AUC = 65.99%. All models proved to be able to generalize both across subjects and over time. The most predictive variables were CASP-12score at baseline, the presence of depression and financial difficulties. Conclusions: Predicting 2-year variations in wellbeing via modeling techniques ispossible, albeit with some limitations, probably originating from the subjectivenature of the outcome.

Objective To employ Artificial Intelligence to model, predict and simulate the amyotrophic lateral sclerosis (ALS) progression over time in terms of variable interactions, functional impairments, and survival. Methods We employed demographic and clinical variables, including functional scores and the utilisation of support interventions, of 3940 ALS patients from four Italian and two Israeli registers to develop a new approach based on Dynamic Bayesian Networks (DBNs) that models the ALS evolution over time, in two distinct scenarios of variable availability. The method allows to simulate patients’ disease trajectories and predict the probability of functional impairment and survival at different time points. Results DBNs explicitly represent the relationships between the variables and the pathways along which they influence the disease progression. Several notable inter-dependencies were identified and validated by comparison with literature. Moreover, the implemented tool allows the assessment of the effect of different markers on the disease course, reproducing the probabilistically expected clinical progressions. The tool shows high concordance in terms of predicted and real prognosis, assessed as time to functional impairments and survival (integral of the AU-ROC in the first 36 months between 0.80–0.93 and 0.84–0.89 for the two scenarios, respectively). Conclusions Provided only with measurements commonly collected during the first visit, our models can predict time to the loss of independence in walking, breathing, swallowing, communicating, and survival and it can be used to generate in silico patient cohorts with specific characteristics. Our tool provides a comprehensive framework to support physicians in treatment planning and clinical decision-making.

Fecal microbiota transplantation (FMT) represents a very promising approach to decreasing disease activity in canine chronic enteropathies (CE). However, the relationship between remission mechanisms and microbiome changes has not been elucidated yet. The main objective of this study was to report the clinical effects of oral freeze-dried FMT in CE dogs, comparing the fecal microbiomes of three groups: pre-FMT CE-affected dogs, post-FMT dogs, and healthy dogs. Diversity analysis, differential abundance analysis, and machine learning algorithms were applied to investigate the differences in microbiome composition between healthy and pre-FMT samples, while Canine Chronic Enteropathy Clinical Activity Index (CCECAI) changes and microbial diversity metrics were used to evaluate FMT effects. In the healthy/pre-FMT comparison, significant differences were noted in alpha and beta diversity and a list of differentially abundant taxa was identified, while machine learning algorithms predicted sample categories with 0.97 (random forest) and 0.87 (sPLS-DA) accuracy. Clinical signs of improvement were observed in 74% (20/27) of CE-affected dogs, together with a statistically significant decrease in CCECAI (median value from 5 to 2 median). Alpha and beta diversity variations between pre- and post-FMT were observed for each receiver, with a high heterogeneity in the response. This highlighted the necessity for further research on a larger dataset that could identify different healing patterns of microbiome changes.

Aims Glucagon like peptide-1 (GLP-1) receptor agonists (GLP-1RA) are effective to control type 2 diabetes (T2Ds) and can protect from adverse cardiovascular outcomes. GLP-1RA are based on the human GLP-1 or the exendin-4 sequence. We compared cardiovascular outcomes of patients with T2D who received human-based or exendin-based GLP-1RA in routine clinical practice. Methods and results We performed a retrospective study on the administrative database of T2D patients from the Veneto Region (North-East Italy). We identified patients who initiated a human-based or exendin-based GLP-1RA from 2011 to 2018. The primary outcome was occurrence of major adverse cardiovascular events (MACE). Secondary outcomes were individual MACE components, revascularization, hospitalization for heart failure, or for cardiovascular causes. From 330 193 patients with diabetes, 6620 were new users of GLP-1RA. After propensity score matching, we analysed 1098 patients in each group, who were on average 61 years old, 59.5% males, 13% with established cardiovascular disease, had an estimated diabetes duration of 8.4 years, and a baseline HbA1c of 7.9%. During a median follow-up of 18 months, patients treated with human-based GLP-1RA as compared to those treated with exendin-based GLP-1RA, showed lower rates of MACE [hazard ratio 0.61; 95% confidence interval (CI) 0.39–0.95], myocardial infarction (0.51; 95% CI 0.28–0.94), and hospitalization for cardiovascular causes (0.66; 95% CI 0.47–0.92). Conclusion We observed better cardiovascular outcomes among patients treated with human-based vs. exendin-based GLP-1RA under routine care. In the absence of comparative trials and in view of the limitations of retrospective studies, this finding provides a moderate level of evidence to guide clinical decision.

Background Clinical registers constitute an invaluable resource in the medical data-driven decision making context. Accurate machine learning and data mining approaches on these data can lead to faster diagnosis, definition of tailored interventions, and improved outcome prediction. A typical issue when implementing such approaches is the almost unavoidable presence of missing values in the collected data. In this work, we propose an imputation algorithm based on a mutual information-weighted k-nearest neighbours approach, able to handle the simultaneous presence of missing information in different types of variables. We developed and validated the method on a clinical register, constituted by the information collected over subsequent screening visits of a cohort of patients affected by amyotrophic lateral sclerosis. Methods For each subject with missing data to be imputed, we create a feature vector constituted by the information collected over his/her first three months of visits. This vector is used as sample in a k-nearest neighbours procedure, in order to select, among the other patients, the ones with the most similar temporal evolution of the disease over time. An ad hoc similarity metric was implemented for the sample comparison, capable of handling the different nature of the data, the presence of multiple missing values and include the cross-information among features captured by the mutual information statistic. Results We validated the proposed imputation method on an independent test set, comparing its performance with those of three state-of-the-art competitors, resulting in better performance. We further assessed the validity of our algorithm by comparing the performance of a survival classifier built on the data imputed with our method versus the one built on the data imputed with the best-performing competitor. Conclusions Imputation of missing data is a crucial –and often mandatory– step when working with real-world datasets. The algorithm proposed in this work could effectively impute an amyotrophic lateral sclerosis clinical dataset, by handling the temporal and the mixed-type nature of the data and by exploiting the cross-information among features. We also showed how the imputation quality can affect a machine learning task.

Because other coronaviruses enter the cells by binding to dipeptidyl-peptidase-4 (DPP-4), it has been speculated that DPP-4 inhibitors (DPP-4is) may exert an activity against severe acute respiratory syndrome coronavirus 2. In the absence of clinical trial results, we analysed epidemiological data to support or discard such a hypothesis. We retrieved information on exposure to DPP-4is among patients with type 2 diabetes (T2D) hospitalized for COVID-19 at an outbreak hospital in Italy. As a reference, we retrieved information on exposure to DPP-4is among matched patients with T2D in the same region. Of 403 hospitalized COVID-19 patients, 85 had T2D. The rate of exposure to DPP-4is was similar between T2D patients with COVID-19 (10.6%) and 14 857 matched patients in the region (8.8%), or 793 matched patients in the local outpatient clinic (15.4%), 8284 matched patients hospitalized for other reasons (8.5%), and when comparing 71 patients hospitalized for COVID-19 pneumonia (11.3%) with 351 matched patients with pneumonia of another aetiology (10.3%). T2D patients with COVID-19 who were on DPP-4is had a similar disease outcome as those who were not. In summary, we found no evidence that DPP-4is might affect hospitalization for COVID-19.

Amyotrophic lateral sclerosis (ALS) is a highly complex and heterogeneous neurodegenerative disease that affects motor neurons. Since life expectancy is relatively low, it is essential to promptly understand the course of the disease to better target the patient’s treatment. Predictive models for disease progression are thus of great interest. One of the most extensive and well-studied open-access data resources for ALS is the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) repository. In 2015, the DREAM-Phil Bowen ALS Prediction Prize4Life Challenge was held on PRO-ACT data, where competitors were asked to develop machine learning algorithms to predict disease progression measured through the slope of the ALSFRS score between 3 and 12 months. However, although it has already been successfully applied in several studies on ALS patients, to the best of our knowledge deep learning approaches still remain unexplored on the ALSFRS slope prediction in PRO-ACT cohort. Here, we investigate how deep learning models perform in predicting ALS progression using the PRO-ACT data. We developed three models based on different architectures that showed comparable or better performance with respect to the state-of-the-art models, thus representing a valid alternative to predict ALS disease progression.

Background Amyotrophic Lateral Sclerosis (ALS) is a neurodegenerative disease whose spreading and progression mechanisms are still unclear. The ability to predict ALS prognosis would improve the patients’ quality of life and support clinicians in planning treatments. In this paper, we investigate ALS evolution trajectories using Process Mining (PM) techniques enriched to both easily mine processes and automatically reveal how the pathways differentiate according to patients’ characteristics. Methods We consider data collected in two distinct data sources, namely the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) dataset and a real-world clinical register (ALS–BS) including data of patients followed up in two tertiary clinical centers of Brescia (Italy). With a focus on the functional abilities progressively impaired as the disease progresses, we use two Process Discovery methods, namely the Directly-Follows Graph and the CareFlow Miner, to mine the population disease trajectories on the PRO-ACT dataset. We characterize the impairment trajectories in terms of patterns, timing, and probabilities, and investigate the effect of some patients’ characteristics at onset on the followed paths. Finally, we perform a comparative study of the impairment trajectories mined in PRO-ACT versus ALS–BS. Results We delineate the progression pathways on PRO-ACT, identifying the predominant disabilities at different stages of the disease: for instance, 85% of patients enter the trials without disabilities, and 48% of them experience the impairment of Walking/Self-care abilities first. We then test how a spinal onset increases the risk of experiencing the loss of Walking/Self-care ability as first impairment (52% vs. 27% of patients develop it as the first impairment in the spinal vs. the bulbar cohorts, respectively), as well as how an older age at onset corresponds to a more rapid progression to death. When compared, the PRO-ACT and the ALS–BS patient populations present some similarities in terms of natural progression of the disease, as well as some differences in terms of observed trajectories plausibly due to the trial scheduling and recruitment criteria. Conclusions We exploited PM to provide an overview of the evolution scenarios of an ALS trial population and to preliminary compare it to the progression observed in a clinical cohort. Future work will focus on further improving the understanding of the disease progression mechanisms, by including additional real-world subjects as well as by extending the set of events considered in the impairment trajectories.

Motivation Recently, several computational modeling approaches, such as agent-based models, have been applied to study the interaction dynamics between immune and tumor cells in human cancer. However, each tumor is characterized by a specific and unique tumor microenvironment, emphasizing the need for specialized and personalized studies of each cancer scenario. Results We present MAST, a hybrid Multi-Agent Spatio-Temporal model which can be informed using a data-driven approach to simulate unique tumor subtypes and tumor–immune dynamics starting from high-throughput sequencing data. It captures essential components of the tumor microenvironment by coupling a discrete agent-based model with a continuous partial differential equations-based model. The application to real data of human colorectal cancer tissue investigating the spatio-temporal evolution and emergent properties of four simulated human colorectal cancer subtypes, along with their agreement with current biological knowledge of tumors and clinical outcome endpoints in a patient cohort, endorse the validity of our approach. Availability and implementation MAST, implemented in Python language, is freely available with an open-source license through GitLab (https://gitlab.com/sysbiobig/mast), and a Docker image is provided to ease its deployment. The submitted software version and test data are available in Zenodo at https://dx.doi.org/10.5281/zenodo.7267745. Supplementary information Supplementary data are available at Bioinformatics Advances online.

COVID-19 severity spans an entire clinical spectrum from asymptomatic to fatal. Most patients who require in-hospital care are admitted to non-intensive wards, but their clinical conditions can deteriorate suddenly and some eventually die. Clinical data from patients' case series have identified pre-hospital and in-hospital risk factors for adverse COVID-19 outcomes. However, most prior studies used static variables or dynamic changes of a few selected variables of interest. In this study, we aimed at integrating the analysis of time-varying multidimensional clinical-laboratory data to describe the pathways leading to COVID-19 outcomes among patients initially hospitalised in a non-intensive care setting.We collected the longitudinal retrospective data of 394 patients admitted to non-intensive care units at the University Hospital of Padova (Padova, Italy) due to COVID-19. We trained a dynamic Bayesian network (DBN) to encode the conditional probability relationships over time between death and all available demographics, pre-existing conditions, and clinical laboratory variables. We applied resampling, dynamic time warping, and prototyping to describe the typical trajectories of patients who died vs. those who survived.The DBN revealed that the trajectory linking demographics and pre-existing clinical conditions to death passed directly through kidney dysfunction or, more indirectly, through cardiac damage. As expected, admittance to the intensive care unit was linked to markers of respiratory function. Notably, death was linked to elevation in procalcitonin and D-dimer levels. Death was associated with persistently high levels of procalcitonin from admission and throughout the hospital stay, likely reflecting bacterial superinfection. A sudden raise in D-dimer levels 3-6 days after admission was also associated with subsequent death, possibly reflecting a worsening thrombotic microangiopathy.This innovative application of DBNs and prototyping to integrated data analysis enables visualising the patient's trajectories to COVID-19 outcomes and may instruct timely and appropriate clinical decisions.

Predicting the risk of cardiovascular complications, in particular heart failure hospitalisation (HHF), can improve the management of type 2 diabetes (T2D). Most predictive models proposed so far rely on clinical data not available at the higher Institutional level. Therefore, it is of interest to assess the risk of HHF in people with T2D using administrative claims data only, which are more easily obtainable and could allow public health systems to identify high-risk individuals. In this paper, the administrative claims of > 175,000 patients with T2D were used to develop a new risk score for HHF based on Cox regression. Internal validation on the administrative data cohort yielded satisfactory results in terms of discrimination (max AUROC = 0.792, C-index = 0.786) and calibration (Hosmer–Lemeshow test p value

Background Results of cardiovascular outcome trials enabled a shift from “treat-to-target” to “treat-to-benefit” paradigm in the management of type 2 diabetes (T2D). However, studies validating such approach are limited. Here, we examined whether treatment according to international recommendations for the pharmacological management of T2D had an impact on long-term outcomes. Methods This was an observational study conducted on outpatient data collected in 2008–2018 (i.e. prior to the “treat-to-benefit” shift). We defined 6 domains of treatment based on the ADA/EASD consensus covering all disease stages: first- and second-line treatment, intensification, use of insulin, cardioprotective, and weight-affecting drugs. At each visit, patients were included in Group 1 if at least one domain deviated from recommendation or in Group 2 if aligned with recommendations. We used Cox proportional hazard models with time-dependent co-variates or Cox marginal structural models (with inverse-probability of treatment weighing evaluated at each visit) to adjust for confounding factors and evaluate three outcomes: major adverse cardiovascular events (MACE), hospitalization for heart failure or cardiovascular mortality (HF-CVM), and all-cause mortality. Results We included 5419 patients, on average 66-year old, 41% women, with a baseline diabetes duration of 7.6 years. Only 11.7% had pre-existing cardiovascular disease. During a median follow-up of 7.3 years, patients were seen 12 times at the clinic, and we recorded 1325 MACE, 1593 HF-CVM, and 917 deaths. By the end of the study, each patient spent on average 63.6% of time in Group 1. In the fully adjusted model, being always in Group 2 was associated with a 45% lower risk of MACE (HR 0.55; 95% C.I. 0.46–0.66; p Conclusion Managing patients with T2D according to a “treat-to-benefit” approach based international standards was associated with a lower risk of MACE, heart failure, and mortality. These data provide ex-post validation of the ADA/EASD treatment algorithm.

Aim Treatment algorithms define lines of glucose lowering medications (GLM) for the management of type 2 diabetes (T2D), but whether therapeutic trajectories are associated with major adverse cardiovascular events (MACE) is unclear. We explored whether the temporal resolution of GLM usage discriminates patients who experienced a 4P-MACE (heart failure, myocardial infarction, stroke, death for all causes). Methods We used an administrative database (Veneto region, North-East Italy, 2011–2018) and implemented recurrent neural networks (RNN) with outcome-specific attention maps. The model input included age, sex, diabetes duration, and a matrix of GLM pattern before the 4P-MACE or censoring. Model output was discrimination, reported as area under receiver characteristic curve (AUROC). Attention maps were produced to show medications whose time-resolved trajectories were the most important for discrimination. Results The analysis was conducted on 147,135 patients for training and model selection and on 10,000 patients for validation. Collected data spanned a period of ~ 6 years. The RNN model efficiently discriminated temporal patterns of GLM ending in a 4P-MACE vs. those ending in an event-free censoring with an AUROC of 0.911 (95% C.I. 0.904–0.919). This excellent performance was significantly better than that of other models not incorporating time-resolved GLM trajectories: (i) a logistic regression on the bag-of-words encoding all GLM ever taken by the patient (AUROC 0.754; 95% C.I. 0.743–0.765); (ii) a model including the sequence of GLM without temporal relationships (AUROC 0.749; 95% C.I. 0.737–0.761); (iii) a RNN model with the same construction rules but including a time-inverted or randomised order of GLM. Attention maps identified the time-resolved pattern of most common first-line (metformin), second-line (sulphonylureas) GLM, and insulin (glargine) as those determining discrimination capacity. Conclusions The time-resolved pattern of GLM use identified patients with subsequent cardiovascular events better than the mere list or sequence of prescribed GLM. Thus, a patient’s therapeutic trajectory could determine disease outcomes.

The fifteen articles in this special section focus on data science used in smart healthcare applications. A shift toward a data-driven socio-economic health model is occurring. This is the result of the increased volume, velocity and variety of data collected from the public and private sector in healthcare, and biology in general. In the past five-years, there has been an impressive development of computational intelligence and informatics methods for application to health and biomedical science. However, the effective use of data to address the scale and scope of human health problems has yet to realize its full potential. The barriers limiting the impact of practical application of standard data mining and machine learning methods have been inherent to the characteristics of health data. Besides the volume of the data (‘big data’), these are challenging due to their heterogeneity, complexity, variability and dynamic nature. Finally, data management and interpretability of the results have been limited by practical challenges in implementing new and also existing standards across the different health providers and research institutions. The scope of this Special issue is to discuss some of these challenges and opportunities in health and biological data science, with particular focus on the infrastructure, software, methods and algorithms needed to analyze large datasets in biological and clinical research.

The presence of missing data is a common problem that affects almost all clinical datasets. Since most available data mining and machine learning algorithms require complete datasets, accurately imputing (i.e. “filling in”) the missing data is an essential step. This paper presents a methodology for the missing data imputation of longitudinal clinical data based on the integration of linear interpolation and a weighted K-Nearest Neighbours (KNN) algorithm. The Maximal Information Coefficient (MIC) values among features are employed as weights for the distance computation in the KNN algorithm in order to integrate intra- and inter-patient information. An interpolation-based imputation approach was also employed and tested both independently and in combination with the KNN algorithm. The final imputation is carried out by applying the best performing method for each feature. The methodology was validated on a dataset of clinical laboratory test results of 13 commonly measured analytes of patients in an intensive care unit (ICU) setting. The performance results are compared with those of 3D-MICE, a state-of-the-art imputation method for cross-sectional and longitudinal patient data. This work was presented in the context of the 2019 ICHI Data Analytics Challenge on Missing data Imputation (DACMI).

Background Amyotrophic lateral sclerosis (ALS) is an adult-onset neurodegenerative disease progressively affecting upper and lower motor neurons in the brain and spinal cord. Mean life expectancy is three to five years, with paralysis of muscles, respiratory failure and loss of vital functions being the common causes of death. Clinical manifestations of ALS are heterogeneous due to the mix of anatomic regions involvement and the variability in disease course; consequently, diagnosis and prognosis at the level of individual patient is really challenging. Prediction of ALS progression and stratification of patients into meaningful subgroups have been long-standing interests to clinical practice, research and drug development. Methods We developed a Dynamic Bayesian Network (DBN) model on more than 4500 ALS patients included in the Pooled Resource Open-Access ALS Clinical Trials Database (PRO-ACT), in order to detect probabilistic relationships among clinical variables and identify risk factors related to survival and loss of vital functions. Furthermore, the DBN was used to simulate the temporal evolution of an ALS cohort predicting survival and the time to impairment of vital functions (communication, swallowing, gait and respiration). A first attempt to stratify patients by risk factors and simulate the progression of ALS subgroups was also implemented. Results The DBN model provided the prediction of ALS most probable trajectories over time in terms of important clinical outcomes, including survival and loss of autonomy in functional domains. Furthermore, it allowed the identification of biomarkers related to patients’ clinical status as well as vital functions, and unrevealed their probabilistic relationships. For instance, DBN found that bicarbonate and calcium levels influence survival time; moreover, the model evidenced dependencies over time among phosphorus level, movement impairment and creatinine. Finally, our model provided a tool to stratify patients into subgroups of different prognosis studying the effect of specific variables, or combinations of them, on either survival time or time to loss of autonomy in specific functional domains. Conclusions The analysis of the risk factors and the simulation allowed by our DBN model might enable better support for ALS prognosis as well as a deeper insight into disease manifestations, in a context of a personalized medicine approach. Electronic supplementary material The online version of this article (10.1186/s12859-019-2692-x) contains supplementary material, which is available to authorized users.

In this letter, we provide a detailed analysis of a gene regulatory network exhibiting bistability within a certain region of parameter space. This network has been adopted in recent literature to describe cellular differentiation into two subpopulations. Biological and experimental evidence suggests that differentiation from stem cells into different tissues evolves through a cascade of similar stages characterized by differentiation into two subtypes. Each differentiation step is influenced by two mechanisms: the first one occurs within the cell and allows transition from an undifferentiated stage to a pluripotent stage where differentiation is possible and the second mechanism may be external to the cell and biases differentiation into a specific subpopulation. In this letter, we address the former mechanism for a gene circuit described by generalized Hill equations and endowed with mutual inhibitory feedback among two competing genes. The main contribution is twofold: 1) mutual inhibition is not sufficient to allow cell differentiation, but specific conditions on the generalized Hill equations’ parameters are required and 2) differentiation occurs only if the triggering gene expression level belongs to a well defined range. Theoretical analysis is complemented with numerical simulations.

Predicting the risk of cardiovascular complications, in particular heart failure hospitalisation (HHF), can improve the management of type 2 diabetes (T2D). Most predictive models proposed so far rely on clinical data not available at the higher Institutional level. Therefore, it is of interest to assess the risk of HHF in people with T2D using administrative claims data only, which are more easily obtainable and could allow public health systems to identify high-risk individuals. In this paper, the administrative claims of >175,000 patients with T2D were used to develop a new risk score for HHF based on Cox regression. Internal validation on the administrative data cohort yielded satisfactory results in terms of discrimination (max AUROC=0.792, C-index=0.786) and calibration (Hosmer-Lemeshow test p-value

When building a predictive model for predicting a clinical outcome using machine learning techniques, the model developers are often interested in ranking the features according to their predictive ability. A commonly used approach to obtain a robust variable ranking is to apply recursive feature elimination (RFE) on multiple resamplings of the training set and then to aggregate the ranking results using the Borda count method. However, the presence of highly correlated features in the training set can deteriorate the ranking performance. In this work, we propose a variant of the method based on RFE and Borda count that takes into account the correlation between variables during the ranking procedure in order to improve the ranking performance in the presence of highly correlated features. The proposed algorithm is tested on simulated datasets in which the true variable importance is known and compared to the standard RFE-Borda count method. According to the root mean square error between the estimated rank and the true (i.e., simulated) feature importance, the proposed algorithm overcomes the standard RFE-Borda count method. Finally, the proposed algorithm is applied to a case study related to the development of a predictive model of type 2 diabetes onset.

Simple Summary Chronic enteropathies are common gastrointestinal diseases in domestic dogs characterised by long-term duration, often impairing quality of life both for pets and owners. It has been demonstrated that the gut microbial community plays a central role in defining the host health status. Indeed, among a variety of biological functions, gut microbiota are involved in the metabolism of nutrients, in training the immune system and in preventing the gastrointestinal ecosystem from being colonised by pathogens. In chronic intestinal diseases, the equilibrium of the gut microbial population is largely impaired, as a consequence of both disease and therapy (e.g., antibiotic treatment). Faecal microbiota transplantation has the aim to restore a balanced microbial population in the patient by simply implanting a healthy gut microbiota derived from a healthy donor to a diseased animal. In doing so, the eubiotic community—and the extensive network of beneficial cross-feeding interactions—are transferred to the receiver’s gut as a whole, favouring the patient to renew a healthy intestinal ecosystem. In this work, we report the encouraging results of a faecal transplantation on a 9-year-old dog suffering from chronic enteropathy for the last 3 years. After the treatment, the dog’s appetite, body weight and vitality were restored, with complete disappearance of gastrointestinal and systemic symptoms. Abstract Chronic enteropathies (CE) are gastrointestinal diseases that afflict about one in five dogs in Europe. Conventional therapeutic approaches include dietary intervention, pharmacological treatment and probiotic supplements. The patient response can be highly variable and the interventions are often not resolutive. Moreover, the therapeutic strategy is usually planned (and gradually corrected) based on the patient’s response to empirical treatment, with few indirect gut health indicators useful to drive clinicians’ decisions. The ever-diminishing cost of high-throughput sequencing (HTS) allows clinicians to directly follow and characterise the evolution of the whole gut microbial community in order to highlight possible weaknesses. In this framework, faecal microbiome transplantation (FMT) is emerging as a feasible solution to CE, based on the implant of a balanced, eubiotic microbial community from a healthy donor to a dysbiotic patient. In this study, we report the promising results of FMT carried out in a 9-year-old dog suffering from CE for the last 3 years. The patient underwent a two-cycle oral treatment of FMT and the microbiota evolution was monitored by 16S rRNA gene sequencing both prior to FMT and after the two administrations. We evaluated the variation of microbial composition by calculating three different alpha diversity indices and compared the patient and donor data to a healthy control population of 94 dogs. After FMT, the patient’s microbiome and clinical parameters gradually shifted to values similar to those observed in healthy dogs. Symptoms disappeared during a follow-up period of six months after the second FMT. We believe that this study opens the door for potential applications of FMT in clinical veterinary practice and highlights the need to improve our knowledge on this relevant topic.

The development of increasingly efficient and cost-effective high throughput DNA sequencing techniques has enhanced the possibility of studying complex microbial systems. Recently, researchers have shown great interest in studying the microorganisms that characterise different ecological niches. Differential abundance analysis aims to find the differences in the abundance of each taxa between two classes of subjects or samples, assigning a significance value to each comparison. Several bioinformatic methods have been specifically developed, taking into account the challenges of microbiome data, such as sparsity, the different sequencing depth constraint between samples and compositionality. Differential abundance analysis has led to important conclusions in different fields, from health to the environment. However, the lack of a known biological truth makes it difficult to validate the results obtained. In this work we exploit metaSPARSim, a microbial sequencing count data simulator, to simulate data with differential abundance features between experimental groups. We perform a complete comparison of recently developed and established methods on a common benchmark with great effort to the reliability of both the simulated scenarios and the evaluation metrics. The performance overview includes the investigation of numerous scenarios, studying the effect on methods’ results on the main covariates such as sample size, percentage of differentially abundant features, sequencing depth, feature variability, normalisation approach and ecological niches. Mainly, we find that methods show a good control of the type I error and, generally, also of the false discovery rate at high sample size, while recall seem to depend on the dataset and sample size.

People with diabetes require lifelong access to healthcare services to delay the onset of complications. Their disease management processes generate great volumes of data across several domains, from clinical to administrative. Difficulties in accessing and processing these data hinder their secondary use in an institutional setting, even for highly desirable applications, such as the prediction of cardiovascular disease, the main driver of excess mortality in diabetes. Hence, in the present work, we propose a deep learning model for the prediction of major adverse cardiovascular events (MACE), developed and validated using the administrative claims of 214,676 diabetic patients of the Veneto region, in North East Italy. Specifically, we use a year of pharmacy and hospitalisation claims, together with basic patient's information, to predict the 4P-MACE composite endpoint, i.e., the first occurrence of death, heart failure, myocardial infarction, or stroke, with a variable prediction horizon of 1 to 5 years. Adapting to the time-to-event nature of this task, we cast our problem as a multi-outcome (4P-MACE and components), multi-label (1 to 5 years) classification task with a custom loss to account for the effect of censoring. Our model, purposefully specified to minimise data preparation costs, exhibits satisfactory performance in predicting 4P-MACE at all prediction horizons: AUROC from 0.812 (C.I.: 0.797 – 0.827) to 0.792 (C.I.: 0.781 – 0.802); C-index from 0.802 (C.I.: 0.788 – 0.816) to 0.770 (C.I.: 0.761 – 0.779). Components’ prediction performance is also adequate, ranging from death's 0.877 1-year AUROC to stroke's 0.689 5-year AUROC.

Interferon-α (IFN-α) comprises a family of 13 cytokines involved in the modulation of antiviral, immune, and anticancer responses by orchestrating a complex transcriptional network. The activation of IFN-α signaling pathway in endothelial cells results in decreased proliferation and migration, ultimately leading to suppression of angiogenesis. In this study, we knocked-down the expression of seven established or candidate modulators of IFN-α response in endothelial cells to reconstruct a gene regulatory network and to investigate the antiangiogenic activity of IFN-α. This genetic perturbation approach, along with the analysis of interferon-induced gene expression dynamics, highlighted a complex and highly interconnected network, in which the angiostatic chemokine C-X-C Motif Chemokine Ligand 10 (CXCL10) was a central node targeted by multiple modulators. IFN-α-induced secretion of CXCL10 protein by endothelial cells was blunted by the silencing of Signal Transducer and Activator of Transcription 1 (STAT1) and of Interferon Regulatory Factor 1 (IRF1) and it was exacerbated by the silencing of Ubiquitin Specific Peptidase 18 (USP18).

Aims We compared cardiovascular outcomes of patients with type 2 diabetes (T2D) receiving sodium glucose cotransporter-2 inhibitors (SGLT2i) or dipeptidyl peptidase-4 inhibitors (DPP4i) under routine care. Methods From an administrative claims database of >5.2M citizen, we identified patients with T2D who initiated SGLT2i or DPP4i from 2014 to 2018. Patients were matched by propensity scores. The primary outcome was the 3-point major adverse cardiovascular events (3P-MACE). Results After matching, we included 3216 patients/group, with mean age of 63 years, diabetes duration of 8.7 years, and 20% had cardiovascular disease. During a median follow-up of 18 months, the rate of 3P-MACE was lower among patients who initiated SGLT2i versus DPP4i (HR 0.74; 95 %C.I. 0.58–0.94). Initiators of SGLT2i also showed significantly lower rates of myocardial infarction (HR 0.75; 95 %C.I. 0.56–1.00), hospitalization for heart failure (HR 0.44; 95 %C.I. 0.25–0.95) or cardiovascular causes (HR 0.72; 95 %C.I. 0.60–0.87), and all-cause death (HR 0.49; 95 %C.I. 0.25–0.95). Renal failure was less common with SGLT2i than with DPP4i. Results were consistent to those obtained in a meta-analysis of 10 observational studies on ~1.5M patients. Conclusions Patients with T2D who initiated SGLT2i under routine care had better cardio-renal outcomes and lower all-cause mortality than similar patients who initiated DPP4i.

Newly differentiated pancreatic β cells lack proper insulin secretion profiles of mature functional β cells. The global gene expression differences between paired immature and mature β cells have been studied, but the dynamics of transcriptional events, correlating with temporal development of glucose-stimulated insulin secretion (GSIS), remain to be fully defined. This aspect is important to identify which genes and pathways are necessary for β-cell development or for maturation, as defective insulin secretion is linked with diseases such as diabetes. In this study, we assayed through RNA sequencing the global gene expression across six β-cell developmental stages in mice, spanning from β-cell progenitor to mature β cells. A computational pipeline then selected genes differentially expressed with respect to progenitors and clustered them into groups with distinct temporal patterns associated with biological functions and pathways. These patterns were finally correlated with experimental GSIS, calcium influx, and insulin granule formation data. Gene expression temporal profiling revealed the timing of important biological processes across β-cell maturation, such as the deregulation of β-cell developmental pathways and the activation of molecular machineries for vesicle biosynthesis and transport, signal transduction of transmembrane receptors, and glucose-induced Ca2+ influx, which were established over a week before β-cell maturation completes. In particular, β cells developed robust insulin secretion at high glucose several days after birth, coincident with the establishment of glucose-induced calcium influx. Yet the neonatal β cells displayed high basal insulin secretion, which decreased to the low levels found in mature β cells only a week later. Different genes associated with calcium-mediated processes, whose alterations are linked with insulin resistance and deregulation of glucose homeostasis, showed increased expression across β-cell stages, in accordance with the temporal acquisition of proper GSIS. Our temporal gene expression pattern analysis provided a comprehensive database of the underlying molecular components and biological mechanisms driving β-cell maturation at different temporal stages, which are fundamental for better control of the in vitro production of functional β cells from human embryonic stem/induced pluripotent cell for transplantation-based type 1 diabetes therapy.

Hypertension is a critical condition that represents a leading risk factor for mortality. The identification of subjects at risk of developing hypertension is important to improve life expectancy and reduce the burden of healthcare systems. Available models to predict hypertension onset in some years in the future mainly include blood pressure (BP) measurements as well as blood test and lifestyle variables. However, systolic and diastolic BP are inevitably strong predictors of the disease and their presence in such models may hide a possible key role of other covariates. The aim of this work is to develop predictive models of hypertension onset both with and without the use of BP measurements to investigate if and how BP variables influence the feature selection process. By involving a large dataset on individuals socio-economic status, demographics, wellbeing, lifestyle, medical history and blood exams, logistic regression models (w/ and w/o BP) have been trained using a stepwise selection procedure to select only highly predictive variables. The model with systolic and diastolic BP selected as important variables HDL cholesterol, hemoglobin, marital status, depression scale and alcohol drinking, achieving an area under the receiver-operating characteristic curve (AU-ROC) of 0.80. The model without BP variables exploits heart rate, waist, age and marital status, and achieves AU-ROC=0.74. As expected, the model employing BP measurements performs better than the one that does not consider them. However, also without BP, it was possible to develop a model with satisfactory performance involving only easily accessible information that do not require laboratory tests.

Chronic Myeloid Leukemia is a clonal disorder characterized by the presence of the Ph-chromosome and the BCR-ABL tyrosine-kinase (TK). Target-therapy with Imatinib has greatly improved its outcome. Deeper and faster responses are reported with the second-generation TKI Nilotinib. Sustained responses may enable TKI discontinuation. However, even in a complete molecular response, some patients experience disease recurrence possibly due to persistence of quiescent leukemic CD34+/lin−Ph+ stem cells (LSCs). Degree and mechanisms of LSCs clearance during TKI treatment are not clearly established. The PhilosoPhi34 study was designed to verify the in-vivo activity and timecourse of first-line Nilotinib therapy on BM CD34+/lin−Ph+ cells clearance. Eighty-seven CP-CML patients were enrolled. BM cells were collected and tested for Ph+ residual cells, at diagnosis, 3, 6 and 12 months of treatment. FISH analysis of unstimulated CD34+/lin− cells in CCyR patients were positive in 8/65 (12.3%), 5/71 (7%), 0/69 (0%) evaluable tests, respectively. Per-Protocol analysis response rates were as follows: CCyR 95% at 12 months, MR4.5 31% and 46% at 12 and 36 months, respectively. An exploratory Gene Expression Profiling (GEP) study of CD34+/lin− cells was performed on 30 patients at diagnosis and after, on 79 patients at diagnosis vs 12 months of nilotinib treatment vs 10 healthy subjects. Data demonstrated some genes significantly different expressed: NFKBIA, many cell cycle genes, ABC transporters, JAK-STAT signaling pathway (JAK2). In addition, a correlation between different expression of some genes (JAK2, OLFM4, ICAM1, NFKBIA) among patients at diagnosis and their achievement of an early and deeper MR was observed.

People affected by diabetes are at a high risk of developing diabetic nephropathy, which, in turn, is the leading cause of end-stage chronic kidney disease worldwide. Predicting the onset of renal complications as early as possible, when kidney function is still intact, is of paramount importance for therapy selection due to existence of a class of antidiabetic agents (SGLT2 inhibitors) with known nephroprotective properties.

This work presents a tool based on a Dynamic Bayesian Network (DBN) model to simulate the progression to type 2 diabetes (T2D) onset in the ageing population. Including longitudinally collected features characterizing different aspects of the ageing process, we dynamically model the relationships among the variables and the outcome over time, obtaining a network that shows a direct joined effect of glycated hemoglobin and body mass index (BMI) on the T2D onset. Remarkably, DBNs present a broad interpretability regardless of their complexity. We also employ the model to assess the impact of modifiable risk factors on developing the disease, showing how an increased BMI leads to an augmented T2D risk.

Background The ability to rapidly adapt to adverse environmental conditions represents the key of success of many pathogens and, in particular, of Mycobacterium tuberculosis. Upon exposition to heat shock, antibiotics or other sources of stress, appropriate responses in terms of genes transcription and proteins activity are activated leading part of a genetically identical bacterial population to express a different phenotype, namely to develop persistence. When the stress response network is mathematically described by an ordinary differential equations model, development of persistence in the bacterial population is associated with bistability of the model, since different emerging phenotypes are represented by different stable steady states. Results In this work, we develop a mathematical model of SigE stress response network that incorporates interactions not considered in mathematical models currently available in the literature. We provide, through involved analytical computations, accurate approximations of the system’s nullclines, and exploit the obtained expressions to determine, in a reliable though computationally efficient way, the number of equilibrium points of the system. Conclusions Theoretical analysis and perturbation experiments point out the crucial role played by the degradation pathway involving RseA, the anti-sigma factor of SigE, for coexistence of two stable equilibria and the emergence of bistability. Our results also indicate that a fine control on RseA concentration is a necessary requirement in order for the system to exhibit bistability.

Background 16S rRNA-gene sequencing is a valuable approach to characterize the taxonomic content of the whole bacterial population inhabiting a metabolic and spatial niche, providing an important opportunity to study bacteria and their role in many health and environmental mechanisms. The analysis of data produced by amplicon sequencing, however, brings very specific methodological issues that need to be properly addressed to obtain reliable biological conclusions. Among these, 16S count data tend to be very sparse, with many null values reflecting species that are present but got unobserved due to the multiplexing constraints. However, current data workflows do not consider a step in which the information about unobserved species is recovered. Results In this work, we evaluate for the first time the effects of introducing in the 16S data workflow a new preprocessing step, zero-imputation, to recover this lost information. Due to the lack of published zero-imputation methods specifically designed for 16S count data, we considered a set of zero-imputation strategies available for other frameworks, and benchmarked them using in silico 16S count data reflecting different experimental designs. Additionally, we assessed the effect of combining zero-imputation and normalization, i.e. the only preprocessing step in current 16S workflow. Overall, we benchmarked 35 16S preprocessing pipelines assessing their ability to handle data sparsity, identify species presence/absence, recovery sample proportional abundance distributions, and improve typical downstream analyses such as computation of alpha and beta diversity indices and differential abundance analysis. Conclusions The results clearly show that 16S data analysis greatly benefits from a properly-performed zero-imputation step, despite the choice of the right zero-imputation method having a pivotal role. In addition, we identify a set of best-performing pipelines that could be a valuable indication for data analysts.

In the current research landscape, microbiota composition studies are of extreme interest, since it has been widely shown that resident microorganisms affect and shape the ecological niche they inhabit. This complex micro-world is characterized by different types of interactions. Understanding these relationships provides a useful tool for decoding the causes and effects of communities’ organizations. Next-Generation Sequencing technologies allow to reconstruct the internal composition of the whole microbial community present in a sample. Sequencing data can then be investigated through statistical and computational method coming from network theory to infer the network of interactions among microbial species. : Since there are several network inference approaches in the literature, in this paper we tried to shed light on their main characteristics and challenges, providing a useful tool not only to those interested in using the methods, but also to those who want to develop new ones. In addition, we focused on the frameworks used to produce synthetic data, starting from the simulation of network structures up to their integration with abundance models, with the aim of clarifying the key points of the entire generative process.

Background Cardiovascular outcome trials in high-risk patients showed that some GLP-1 receptor agonists (GLP-1RA), but not dipeptidyl-peptidase-4 inhibitors (DPP-4i), can prevent cardiovascular events in type 2 diabetes (T2D). Since no trial has directly compared these two classes of drugs, we performed a comparative outcome analysis using real-world data. Methods From a database of ~ 5 million people from North-East Italy, we retrospectively identified initiators of GLP-1RA or DPP-4i from 2011 to 2018. We obtained two balanced cohorts by 1:1 propensity score matching. The primary outcome was the 3-point major adverse cardiovascular events (3P-MACE; a composite of death, myocardial infarction, or stroke). 3P-MACE components and hospitalization for heart failure were secondary outcomes. Results From 330,193 individuals with T2D, we extracted two matched cohorts of 2807 GLP-1RA and 2807 DPP-4i initiators, followed for a median of 18 months. On average, patients were 63 years old, 60% male; 15% had pre-existing cardiovascular disease. The rate of 3P-MACE was lower in patients treated with GLP-1RA compared to DPP4i (23.5 vs. 34.9 events per 1000 person-years; HR: 0.67; 95% C.I. 0.53–0.86; p = 0.002). Rates of myocardial infarction (HR 0.67; 95% C.I. 0.50–0.91; p = 0.011) and all-cause death (HR 0.58; 95% C.I. 0.35–0.96; p = 0.034) were lower among GLP-1RA initiators. The as-treated and intention-to-treat approaches yielded similar results. Conclusions Patients initiating a GLP-1RA in clinical practice had better cardiovascular outcomes than similar patients who initiated a DPP-4i. These data strongly confirm findings from cardiovascular outcome trials in a lower risk population.

Diabetes is a chronic disease characterised by high blood glucose levels, resulting in reduced life expectancy and requiring lifelong treatment. Because of their frequent necessity to seek healthcare services, diabetic patients generate huge volumes of administrative claims, detailing their healthcare encounters and prescription history. In the Veneto region (NE Italy), as in the rest of the country, these data are collected systematically and cover all eligible beneficiaries (~5 million in Veneto). Since cardiovascular complications are the main drivers of excess mortality in diabetes, forecasting their onset is desirable for patients’ care and policymaking. Hence, in the present work, we investigate the possibility of predicting 3-year cardiovascular outcomes following a 1-year baseline period of pharmacy data collection. We implement an approach based on recurrent neural networks that combines the chronologically-ordered sequence of prescriptions filled by a diabetic patient and basic biographical information (age, gender, estimated diabetes duration) to determine whether he or she will experience a 4P-MACE (4-point major adverse cardiovascular event; defined as death, myocardial infarction, stroke, or heart failure) endpoint. We develop our model with the data of 97,466 known diabetic patients identified using a validated claims-based algorithm. Independent performance tests on 4,873 subjects yield an area under the receiver-operating characteristic curve of 0.791 and a concordance index of 0.765 for the 4P-MACE primary outcome. We find death to be the easiest 4P-MACE component to predict (AUROC = 0.846), followed by heart failure (0.796), stroke (0.714), and, finally, myocardial infarction (0.708). Secondary, stratified experiments highlight independence of performance from gender, age, and number of prescriptions filled in a year with respect to the primary outcome. To the best of our knowledge, this is the first large-scale prediction model of cardiovascular complications in medication-treated diabetes solely based on sequences of filled prescriptions collected from administrative claims.

IntroductionMany predictive models for incident type 2 diabetes (T2D) exist, but these models are not used frequently for public health management. Barriers to their application include (1) the problem of model choice (some models are applicable only to certain ethnic groups), (2) missing input variables, and (3) the lack of calibration. While (1) and (2) drives to missing predictions, (3) causes inaccurate incidence predictions. In this paper, a combined T2D risk model for public health management that addresses these three issues is developed.Research design and methodsThe combined T2D risk model combines eight existing predictive models by weighted average to overcome the problem of missing incidence predictions. Moreover, the combined model implements a simple recalibration strategy in which the risk scores are rescaled based on the T2D incidence in the target population. The performance of the combined model was compared with that of the eight existing models using data from two test datasets extracted from the Multi-Ethnic Study of Atherosclerosis (MESA; n=1031) and the English Longitudinal Study of Ageing (ELSA; n=4820). Metrics of discrimination, calibration, and missing incidence predictions were used for the assessment.ResultsThe combined T2D model performed well in terms of both discrimination (concordance index: 0.83 on MESA; 0.77 on ELSA) and calibration (expected to observed event ratio: 1.00 on MESA; 1.17 on ELSA), similarly to the best-performing existing models. However, while the existing models yielded a large percentage of missing predictions (17%–45% on MESA; 63%–64% on ELSA), this was negligible with the combined model (0% on MESA, 4% on ELSA).ConclusionsLeveraging on existing literature T2D predictive models, a simple approach based on risk score rescaling and averaging was shown to provide accurate and robust incidence predictions, overcoming the problem of recalibration and missing predictions in practical application of predictive models.