Optimization of Large Vessel Occlusion Detection in Acute Ischemic Stroke Using Machine Learning Methods.

The early detection of large-vessel occlusion (LVO) strokes is increasingly important as these patients are potential candidates for endovascular therapy, the availability of which is limited. Prehospital LVO detection scales mainly contain symptom variables only; however, recent studies revealed that other types of variables could be useful as well. Our aim was to comprehensively assess the predictive ability of several clinical variables for LVO prediction and to develop an optimal combination of them using machine learning tools. We have retrospectively analysed data from a prospectively collected multi-centre stroke registry. Data on 41 variables were collected and divided into four groups (baseline vital parameters/demographic data, medical history, laboratory values, and symptoms). Following the univariate analysis, the LASSO method was used for feature selection to select an optimal combination of variables, and various machine learning methods (random forest (RF), logistic regression (LR), elastic net method (ENM), and simple neural network (SNN)) were applied to optimize the performance of the model. A total of 526 patients were included. Several neurological symptoms were more common and more severe in the group of LVO patients. Atrial fibrillation (AF) was more common, and serum white blood cell (WBC) counts were higher in the LVO group, while systolic blood pressure (SBP) was lower among LVO patients. Using the LASSO method, nine variables were selected for modelling (six symptom variables, AF, chronic heart failure, and WBC count). When applying machine learning methods and 10-fold cross validation using the selected variables, all models proved to have an AUC between 0.736 (RF) and 0.775 (LR), similar to the performance of National Institutes of Health Stroke Scale (AUC: 0.790). Our study highlights that, although certain neurological symptoms have the best ability to predict an LVO, other variables (such as AF and CHF in medical history and white blood cell counts) should also be included in multivariate models to optimize their efficiency.

Paradigm changes for diagnosis: using big data for prediction.

Due to profound changes occurring in biomedical knowledge and in health systems worldwide, an entirely new health and social care scenario is emerging. Moreover, the enormous technological potential developed over the last years is increasingly influencing life sciences and driving changes toward personalized medicine and value-based healthcare. However, the current slow progression of adoption, limiting the generation of healthcare efficiencies through technological innovation, can be realistically overcome by fostering convergence between a systems medicine approach and the principles governing Integrated Care. Implicit with this strategy is the multidisciplinary active collaboration of all stakeholders involved in the change, namely: citizens, professionals with different profiles, academia, policy makers, industry and payers. The article describes the key building blocks of an open and collaborative hub currently being developed in Catalonia (Spain) aiming at generation, deployment and evaluation of a personalized medicine program addressing highly prevalent chronic conditions that often show co-occurrence, namely: cardiovascular disorders, chronic obstructive pulmonary disease, type 2 diabetes mellitus; metabolic syndrome and associated mental disturbances (anxiety-depression and altered behavioral patterns leading to unhealthy life styles).

Risk and temporal order of disease diagnosis of comorbidities in patients with COPD: a population health perspective.

INTRODUCTION: Comorbidities in patients with chronic obstructive pulmonary disease (COPD) generate a major burden on healthcare. Identification of cost-effective strategies aiming at preventing and enhancing management of comorbid conditions in patients with COPD requires deeper knowledge on epidemiological patterns and on shared biological pathways explaining co-occurrence of diseases. METHODS: The study assesses the co-occurrence of several chronic conditions in patients with COPD using two different datasets: Catalan Healthcare Surveillance System (CHSS) (ES, 1.4 million registries) and Medicare (USA, 13 million registries). Temporal order of disease diagnosis was analysed in the CHSS dataset. RESULTS: The results demonstrate higher prevalence of most of the diseases, as comorbid conditions, in elderly (>65) patients with COPD compared with non-COPD subjects, an effect observed in both CHSS and Medicare datasets. Analysis of temporal order of disease diagnosis showed that comorbid conditions in elderly patients with COPD tend to appear after the diagnosis of the obstructive disease, rather than before it. CONCLUSION: The results provide a population health perspective of the comorbidity challenge in patients with COPD, indicating the increased risk of developing comorbid conditions in these patients. The research reinforces the need for novel approaches in the prevention and management of comorbidities in patients with COPD to effectively reduce the overall burden of the disease on these patients.

Population-based analysis of patients with COPD in Catalonia: a cohort study with implications for clinical management.

BACKGROUND: Clinical management of patients with chronic obstructive pulmonary disease (COPD) shows potential for improvement provided that patients' heterogeneities are better understood. The study addresses the impact of comorbidities and its role in health risk assessment. OBJECTIVE: To explore the potential of health registry information to enhance clinical risk assessment and stratification. DESIGN: Fixed cohort study including all registered patients with COPD in Catalonia (Spain) (7.5 million citizens) at 31 December 2014 with 1-year (2015) follow-up. METHODS: A total of 264 830 patients with COPD diagnosis, based on the International Classification of Diseases (Ninth Revision) coding, were assessed. Performance of multiple logistic regression models for the six main dependent variables of the study: mortality, hospitalisations (patients with one or more admissions; all cases and COPD-related), multiple hospitalisations (patients with at least two admissions; all causes and COPD-related) and users with high healthcare costs. Neither clinical nor forced spirometry data were available. RESULTS: Multimorbidity, assessed with the adjusted morbidity grouper, was the covariate with the highest impact in the predictive models, which in turn showed high performance measured by the C-statistics: (1) mortality (0.83), (2 and 3) hospitalisations (all causes: 0.77; COPD-related: 0.81), (4 and 5) multiple hospitalisations (all causes: 0.80; COPD-related: 0.87) and (6) users with high healthcare costs (0.76). Fifteen per cent of individuals with highest healthcare costs to year ratio represented 59% of the overall costs of patients with COPD. CONCLUSIONS: The results stress the impact of assessing multimorbidity with the adjusted morbidity grouper on considered health indicators, which has implications for enhanced COPD staging and clinical management. TRIAL REGISTRATION NUMBER: NCT02956395.

Network modules uncover mechanisms of skeletal muscle dysfunction in COPD patients.

BACKGROUND: Chronic obstructive pulmonary disease (COPD) patients often show skeletal muscle dysfunction that has a prominent negative impact on prognosis. The study aims to further explore underlying mechanisms of skeletal muscle dysfunction as a characteristic systemic effect of COPD, potentially modifiable with preventive interventions (i.e. muscle training). The research analyzes network module associated pathways and evaluates the findings using independent measurements. METHODS: We characterized the transcriptionally active network modules of interacting proteins in the vastus lateralis of COPD patients (n = 15, FEV1 46 ± 12% pred, age 68 ± 7 years) and healthy sedentary controls (n = 12, age 65 ± 9  years), at rest and after an 8-week endurance training program. Network modules were functionally evaluated using experimental data derived from the same study groups. RESULTS: At baseline, we identified four COPD specific network modules indicating abnormalities in creatinine metabolism, calcium homeostasis, oxidative stress and inflammatory responses, showing statistically significant associations with exercise capacity (VO2 peak, Watts peak, BODE index and blood lactate levels) (P < 0.05 each), but not with lung function (FEV1). Training-induced network modules displayed marked differences between COPD and controls. Healthy subjects specific training adaptations were significantly associated with cell bioenergetics (P < 0.05) which, in turn, showed strong relationships with training-induced plasma metabolomic changes; whereas, effects of training in COPD were constrained to muscle remodeling. CONCLUSION: In summary, altered muscle bioenergetics appears as the most striking finding, potentially driving other abnormal skeletal muscle responses. Trial registration The study was based on a retrospectively registered trial (May 2017), ClinicalTrials.gov identifier: NCT03169270.

ChainRank, a chain prioritisation method for contextualisation of biological networks.

BACKGROUND: Advances in high throughput technologies and growth of biomedical knowledge have contributed to an exponential increase in associative data. These data can be represented in the form of complex networks of biological associations, which are suitable for systems analyses. However, these networks usually lack both, context specificity in time and space as well as the distinctive borders, which are usually assigned in the classical pathway view of molecular events (e.g. signal transduction). This complexity and high interconnectedness call for automated techniques that can identify smaller targeted subnetworks specific to a given research context (e.g. a disease scenario). RESULTS: Our method, named ChainRank, finds relevant subnetworks by identifying and scoring chains of interactions that link specific network components. Scores can be generated from integrating multiple general and context specific measures (e.g. experimental molecular data from expression to proteomics and metabolomics, literature evidence, network topology). The performance of the novel ChainRank method was evaluated on recreating selected signalling pathways from a human protein interaction network. Specifically, we recreated skeletal muscle specific signaling networks in healthy and chronic obstructive pulmonary disease (COPD) contexts. The analysis showed that ChainRank can identify main mediators of context specific molecular signalling. An improvement of up to factor 2.5 was shown in the precision of finding proteins of the recreated pathways compared to random simulation. CONCLUSIONS: ChainRank provides a framework, which can integrate several user-defined scores and evaluate their combined effect on ranking interaction chains linking input data sets. It can be used to contextualise networks, identify signaling and regulatory path amongst targeted genes or to analyse synthetic lethality in the context of anticancer therapy. ChainRank is implemented in R programming language and freely available at https://github.com/atenyi/ChainRank.

Systems Medicine: from molecular features and models to the clinic in COPD.

BACKGROUND AND HYPOTHESIS: Chronic Obstructive Pulmonary Disease (COPD) patients are characterized by heterogeneous clinical manifestations and patterns of disease progression. Two major factors that can be used to identify COPD subtypes are muscle dysfunction/wasting and co-morbidity patterns. We hypothesized that COPD heterogeneity is in part the result of complex interactions between several genes and pathways. We explored the possibility of using a Systems Medicine approach to identify such pathways, as well as to generate predictive computational models that may be used in clinic practice. OBJECTIVE AND METHOD: Our overarching goal is to generate clinically applicable predictive models that characterize COPD heterogeneity through a Systems Medicine approach. To this end we have developed a general framework, consisting of three steps/objectives: (1) feature identification, (2) model generation and statistical validation, and (3) application and validation of the predictive models in the clinical scenario. We used muscle dysfunction and co-morbidity as test cases for this framework. RESULTS: In the study of muscle wasting we identified relevant features (genes) by a network analysis and generated predictive models that integrate mechanistic and probabilistic models. This allowed us to characterize muscle wasting as a general de-regulation of pathway interactions. In the co-morbidity analysis we identified relevant features (genes/pathways) by the integration of gene-disease and disease-disease associations. We further present a detailed characterization of co-morbidities in COPD patients that was implemented into a predictive model. In both use cases we were able to achieve predictive modeling but we also identified several key challenges, the most pressing being the validation and implementation into actual clinical practice. CONCLUSIONS: The results confirm the potential of the Systems Medicine approach to study complex diseases and generate clinically relevant predictive models. Our study also highlights important obstacles and bottlenecks for such approaches (e.g. data availability and normalization of frameworks among others) and suggests specific proposals to overcome them.

The COPD Knowledge Base: enabling data analysis and computational simulation in translational COPD research.

BACKGROUND: Previously we generated a chronic obstructive pulmonary disease (COPD) specific knowledge base (http://www.copdknowledgebase.eu) from clinical and experimental data, text-mining results and public databases. This knowledge base allowed the retrieval of specific molecular networks together with integrated clinical and experimental data. RESULTS: The COPDKB has now been extended to integrate over 40 public data sources on functional interaction (e.g. signal transduction, transcriptional regulation, protein-protein interaction, gene-disease association). In addition we integrated COPD-specific expression and co-morbidity networks connecting over 6 000 genes/proteins with physiological parameters and disease states. Three mathematical models describing different aspects of systemic effects of COPD were connected to clinical and experimental data. We have completely redesigned the technical architecture of the user interface and now provide html and web browser-based access and form-based searches. A network search enables the use of interconnecting information and the generation of disease-specific sub-networks from general knowledge. Integration with the Synergy-COPD Simulation Environment enables multi-scale integrated simulation of individual computational models while integration with a Clinical Decision Support System allows delivery into clinical practice. CONCLUSIONS: The COPD Knowledge Base is the only publicly available knowledge resource dedicated to COPD and combining genetic information with molecular, physiological and clinical data as well as mathematical modelling. Its integrated analysis functions provide overviews about clinical trends and connections while its semantically mapped content enables complex analysis approaches. We plan to further extend the COPDKB by offering it as a repository to publish and semantically integrate data from relevant clinical trials. The COPDKB is freely available after registration at http://www.copdknowledgebase.eu.