Transfer Learning for Mortality Prediction in Non-Small Cell Lung Cancer with Low-Resolution Histopathology Slide Snapshots.

High-resolution whole slide image scans of histopathology slides have been widely used in recent years for prediction in cancer. However, in some cases, clinical informatics practitioners may only have access to low-resolution snapshots of histopathology slides, not high-resolution scans. We evaluated strategies for training neural network prognostic models in non-small cell lung cancer (NSCLC) based on low-resolution snapshots, using data from the Veterans Affairs Precision Oncology Data Repository. We compared strategies without transfer learning, with transfer learning from general domain images, and with transfer learning from publicly available high-resolution histopathology scans. We found transfer learning from high-resolution scans achieved significantly better performance than other strategies. Our contribution provides a foundation for future development of prognostic models in NSCLC that incorporate data from low-resolution pathology slide snapshots alongside known clinical predictors.

The MT1 receptor as the target of ramelteon neuroprotection in ischemic stroke.

Stroke is the leading cause of death and disability worldwide. Novel and effective therapies for ischemic stroke are urgently needed. Here, we report that melatonin receptor 1A (MT1) agonist ramelteon is a neuroprotective drug candidate as demonstrated by comprehensive experimental models of ischemic stroke, including a middle cerebral artery occlusion (MCAO) mouse model of cerebral ischemia in vivo, organotypic hippocampal slice cultures ex vivo, and cultured neurons in vitro; the neuroprotective effects of ramelteon are diminished in MT1-knockout (KO) mice and MT1-KO cultured neurons. For the first time, we report that the MT1 receptor is significantly depleted in the brain of MCAO mice, and ramelteon treatment significantly recovers the brain MT1 losses in MCAO mice, which is further explained by the Connectivity Map L1000 bioinformatic analysis that shows gene-expression signatures of MCAO mice are negatively connected to melatonin receptor agonist like Ramelteon. We demonstrate that ramelteon improves the cerebral blood flow signals in ischemic stroke that is potentially mediated, at least, partly by mechanisms of activating endothelial nitric oxide synthase. Our results also show that the neuroprotection of ramelteon counteracts reactive oxygen species-induced oxidative stress and activates the nuclear factor erythroid 2-related factor 2/heme oxygenase-1 pathway. Ramelteon inhibits the mitochondrial and autophagic death pathways in MCAO mice and cultured neurons, consistent with gene set enrichment analysis from a bioinformatics perspective angle. Our data suggest that Ramelteon is a potential neuroprotective drug candidate, and MT1 is the neuroprotective target for ischemic stroke, which provides new insights into stroke therapy. MT1-KO mice and cultured neurons may provide animal and cellular models of accelerated ischemic damage and neuronal cell death.

Predicting ward transfer mortality with machine learning.

In order to address a long standing challenge for internal medicine physicians we developed artificial intelligence (AI) models to identify patients at risk of increased mortality. After querying 2,425 records of patients transferred from non-intensive care units to intensive care units from the Veteran Affairs Corporate Data Warehouse (CDW), we created two datasets. The former used 22 independent variables that included "Length of Hospital Stay" and "Days to Intensive Care Transfer," and the latter lacked these two variables. Since these two variables are unknown at the time of admission, the second set is more clinically relevant. We trained 16 machine learning models using both datasets. The best-performing models were fine-tuned and evaluated. The LightGBM model achieved the best results for both datasets. The model trained with 22 variables achieved a Receiver Operating Characteristics Curve-Area Under the Curve (ROC-AUC) of 0.89 and an accuracy of 0.72, with a sensitivity of 0.97 and a specificity of 0.68. The model trained with 20 variables achieved a ROC-AUC of 0.86 and an accuracy of 0.71, with a sensitivity of 0.94 and a specificity of 0.67. The top features for the former model included "Total length of Stay," "Admit to ICU Transfer Days," and "Lymphocyte Next Lab Value." For the latter model, the top features included "Lymphocyte First Lab Value," "Hemoglobin First Lab Value," and "Hemoglobin Next Lab Value." Our clinically relevant predictive mortality model can assist providers in optimizing resource utilization when managing large caseloads, particularly during shift changes.

Introducing HL7 FHIR Genomics Operations: a developer-friendly approach to genomics-EHR integration.

OBJECTIVE: Enabling clinicians to formulate individualized clinical management strategies from the sea of molecular data remains a fundamentally important but daunting task. Here, we describe efforts towards a new paradigm in genomics-electronic health record (HER) integration, using a standardized suite of FHIR Genomics Operations that encapsulates the complexity of molecular data so that precision medicine solution developers can focus on building applications. MATERIALS AND METHODS: FHIR Genomics Operations essentially "wrap" a genomics data repository, presenting a uniform interface to applications. More importantly, operations encapsulate the complexity of data within a repository and normalize redundant data representations-particularly relevant in genomics, where a tremendous amount of raw data exists in often-complex non-FHIR formats. RESULTS: Fifteen FHIR Genomics Operations have been developed, designed to support a wide range of clinical scenarios, such as variant discovery; clinical trial matching; hereditary condition and pharmacogenomic screening; and variant reanalysis. Operations are being matured through the HL7 balloting process, connectathons, pilots, and the HL7 FHIR Accelerator program. DISCUSSION: Next-generation sequencing can identify thousands to millions of variants, whose clinical significance can change over time as our knowledge evolves. To manage such a large volume of dynamic and complex data, new models of genomics-EHR integration are needed. Qualitative observations to date suggest that freeing application developers from the need to understand the nuances of genomic data, and instead base applications on standardized APIs can not only accelerate integration but also dramatically expand the applications of Omic data in driving precision care at scale for all.

The effects of department of Veterans Affairs medical centers on socio-economic outcomes: Evidence from the Paycheck Protection Program.

Do medical facilities also help advance improvements in socio-economic outcomes? We focus on Veterans, a vulnerable group over the COVID-19 pandemic who have access to a comprehensive healthcare network, and the receipt of funds from the Paycheck Protection Program (PPP) between April and June as a source of variation. First, we find that Veterans received 3.5% more loans and 6.8% larger loans than their counterparts (p < 0.01), controlling for a wide array of zipcode characteristics. Second, we develop models to predict the number of PPP loans awarded to Veterans, finding that the inclusion of local VA medical center characteristics adds almost as much explanatory power as the industry and occupational composition in an area and even more than the education, race, and age distribution combined. Our results suggest that VA medical centers can play an important role in helping Veterans thrive even beyond addressing their direct medical needs.

Characterizing Thrombotic Complication Risk Factors Associated With COVID-19 via Heterogeneous Patient Data: Retrospective Observational Study.

BACKGROUND: COVID-19 has been observed to be associated with venous and arterial thrombosis. The inflammatory disease prolongs hospitalization, and preexisting comorbidities can intensity the thrombotic burden in patients with COVID-19. However, venous thromboembolism, arterial thrombosis, and other vascular complications may go unnoticed in critical care settings. Early risk stratification is paramount in the COVID-19 patient population for proactive monitoring of thrombotic complications. OBJECTIVE: The aim of this exploratory research was to characterize thrombotic complication risk factors associated with COVID-19 using information from electronic health record (EHR) and insurance claims databases. The goal is to develop an approach for analysis using real-world data evidence that can be generalized to characterize thrombotic complications and additional conditions in other clinical settings as well, such as pneumonia or acute respiratory distress syndrome in COVID-19 patients or in the intensive care unit. METHODS: We extracted deidentified patient data from the insurance claims database IBM MarketScan, and formulated hypotheses on thrombotic complications in patients with COVID-19 with respect to patient demographic and clinical factors using logistic regression. The hypotheses were then verified with analysis of deidentified patient data from the Research Patient Data Registry (RPDR) Mass General Brigham (MGB) patient EHR database. Data were analyzed according to odds ratios, 95% CIs, and P values. RESULTS: The analysis identified significant predictors (P<.001) for thrombotic complications in 184,831 COVID-19 patients out of the millions of records from IBM MarketScan and the MGB RPDR. With respect to age groups, patients 60 years and older had higher odds (4.866 in MarketScan and 6.357 in RPDR) to have thrombotic complications than those under 60 years old. In terms of gender, men were more likely (odds ratio of 1.245 in MarketScan and 1.693 in RPDR) to have thrombotic complications than women. Among the preexisting comorbidities, patients with heart disease, cerebrovascular diseases, hypertension, and personal history of thrombosis all had significantly higher odds of developing a thrombotic complication. Cancer and obesity were also associated with odds>1. The results from RPDR validated the IBM MarketScan findings, as they were largely consistent and afford mutual enrichment. CONCLUSIONS: The analysis approach adopted in this study can work across heterogeneous databases from diverse organizations and thus facilitates collaboration. Searching through millions of patient records, the analysis helped to identify factors influencing a phenotype. Use of thrombotic complications in COVID-19 patients represents only a case study; however, the same design can be used across other disease areas by extracting corresponding disease-specific patient data from available databases.

Developing and Implementing Predictive Models in a Learning Healthcare System: Traditional and Artificial Intelligence Approaches in the Veterans Health Administration.

Predicting clinical risk is an important part of healthcare and can inform decisions about treatments, preventive interventions, and provision of extra services. The field of predictive models has been revolutionized over the past two decades by electronic health record data; the ability to link such data with other demographic, socioeconomic, and geographic information; the availability of high-capacity computing; and new machine learning and artificial intelligence methods for extracting insights from complex datasets. These advances have produced a new generation of computerized predictive models, but debate continues about their development, reporting, validation, evaluation, and implementation. In this review we reflect on more than 10 years of experience at the Veterans Health Administration, the largest integrated healthcare system in the United States, in developing, testing, and implementing such models at scale. We report lessons from the implementation of national risk prediction models and suggest an agenda for research.

A research agenda to support the development and implementation of genomics-based clinical informatics tools and resources.

OBJECTIVE: The Genomic Medicine Working Group of the National Advisory Council for Human Genome Research virtually hosted its 13th genomic medicine meeting titled "Developing a Clinical Genomic Informatics Research Agenda". The meeting's goal was to articulate a research strategy to develop Genomics-based Clinical Informatics Tools and Resources (GCIT) to improve the detection, treatment, and reporting of genetic disorders in clinical settings. MATERIALS AND METHODS: Experts from government agencies, the private sector, and academia in genomic medicine and clinical informatics were invited to address the meeting's goals. Invitees were also asked to complete a survey to assess important considerations needed to develop a genomic-based clinical informatics research strategy. RESULTS: Outcomes from the meeting included identifying short-term research needs, such as designing and implementing standards-based interfaces between laboratory information systems and electronic health records, as well as long-term projects, such as identifying and addressing barriers related to the establishment and implementation of genomic data exchange systems that, in turn, the research community could help address. DISCUSSION: Discussions centered on identifying gaps and barriers that impede the use of GCIT in genomic medicine. Emergent themes from the meeting included developing an implementation science framework, defining a value proposition for all stakeholders, fostering engagement with patients and partners to develop applications under patient control, promoting the use of relevant clinical workflows in research, and lowering related barriers to regulatory processes. Another key theme was recognizing pervasive biases in data and information systems, algorithms, access, value, and knowledge repositories and identifying ways to resolve them.

SMART COVID Navigator, a Clinical Decision Support Tool for COVID-19 Treatment: Design and Development Study.

BACKGROUND: COVID-19 caused by SARS-CoV-2 has infected 219 million individuals at the time of writing of this paper. A large volume of research findings from observational studies about disease interactions with COVID-19 is being produced almost daily, making it difficult for physicians to keep track of the latest information on COVID-19's effect on patients with certain pre-existing conditions. OBJECTIVE: In this paper, we describe the creation of a clinical decision support tool, the SMART COVID Navigator, a web application to assist clinicians in treating patients with COVID-19. Our application allows clinicians to access a patient's electronic health records and identify disease interactions from a large set of observational research studies that affect the severity and fatality due to COVID-19. METHODS: The SMART COVID Navigator takes a 2-pronged approach to clinical decision support. The first part is a connection to electronic health record servers, allowing the application to access a patient's medical conditions. The second is accessing data sets with information from various observational studies to determine the latest research findings about COVID-19 outcomes for patients with certain medical conditions. By connecting these 2 data sources, users can see how a patient's medical history will affect their COVID-19 outcomes. RESULTS: The SMART COVID Navigator aggregates patient health information from multiple Fast Healthcare Interoperability Resources-enabled electronic health record systems. This allows physicians to see a comprehensive view of patient health records. The application accesses 2 data sets of over 1100 research studies to provide information on the fatality and severity of COVID-19 for several pre-existing conditions. We also analyzed the results of the collected studies to determine which medical conditions result in an increased chance of severity and fatality of COVID-19 progression. We found that certain conditions result in a higher likelihood of severity and fatality probabilities. We also analyze various cancer tissues and find that the probabilities for fatality vary greatly depending on the tissue being examined. CONCLUSIONS: The SMART COVID Navigator allows physicians to predict the fatality and severity of COVID-19 progression given a particular patient's medical conditions. This can allow physicians to determine how aggressively to treat patients infected with COVID-19 and to prioritize different patients for treatment considering their prior medical conditions.

Computational prediction and validation of specific EmbR binding site on PknH.

Tuberculosis drug resistance continues to threaten global health but the underline molecular mechanisms are not clear. Ethambutol (EMB), one of the well-known first - line drugs in tuberculosis treatment is, unfortunately, not free from drug resistance problems. Genomic studies have shown that some genetic mutations in Mycobacterium tuberculosis (Mtb) EmbR, and EmbC/A/B genes cause EMB resistance. EmbR-PknH pair controls embC/A/B operon, which encodes EmbC/A/B genes, and EMB interacts with EmbA/B proteins. However, the EmbR binding site on PknH was unknown. We conducted molecular simulation on the EmbR- peptides binding structures and discovered phosphorylated PknH 273-280 (N'-HEALSPDPD-C') makes ß strand with the EmbR FHA domain, as ß-MoRF (MoRF; molecular recognition feature) does at its binding site. Hydrogen bond number analysis also supported the peptides' ß-MoRF forming activity at the EmbR FHA domain. Also, we discovered that previously known phosphorylation residues might have their chronological order according to the phosphorylation status. The discovery validated that Mtb PknH 273-280 (N'-HEALSDPD-C') has reliable EmbR binding affinity. This approach is revolutionary in the computer-aided drug discovery field, because it is the first trial to discover the protein-protein interaction site, and find binding partner in nature from this site.

Interoperable genetic lab test reports: mapping key data elements to HL7 FHIR specifications and professional reporting guidelines.

OBJECTIVE: In many cases, genetic testing labs provide their test reports as portable document format files or scanned images, which limits the availability of the contained information to advanced informatics solutions, such as automated clinical decision support systems. One of the promising standards that aims to address this limitation is Health Level Seven International (HL7) Fast Healthcare Interoperability Resources Clinical Genomics Implementation Guide-Release 1 (FHIR CG IG STU1). This study aims to identify various data content of some genetic lab test reports and map them to FHIR CG IG specification to assess its coverage and to provide some suggestions for standard development and implementation. MATERIALS AND METHODS: We analyzed sample reports of 4 genetic tests and relevant professional reporting guidelines to identify their key data elements (KDEs) that were then mapped to FHIR CG IG. RESULTS: We identified 36 common KDEs among the analyzed genetic test reports, in addition to other unique KDEs for each genetic test. Relevant suggestions were made to guide the standard implementation and development. DISCUSSION AND CONCLUSION: The FHIR CG IG covers the majority of the identified KDEs. However, we suggested some FHIR extensions that might better represent some KDEs. These extensions may be relevant to FHIR implementations or future FHIR updates.The FHIR CG IG is an excellent step toward the interoperability of genetic lab test reports. However, it is a work-in-progress that needs informative and continuous input from the clinical genetics' community, specifically professional organizations, systems implementers, and genetic knowledgebase providers.

Hyper-Synergistic Antifungal Activity of Rapamycin and Peptide-Like Compounds against Candida albicans Orthogonally via Tor1 Kinase.

Candida albicans is a life-threatening, opportunistic fungal pathogen with a high mortality rate, especially within the immunocompromised populations. Multidrug resistance combined with limited antifungal drugs even worsens the situation. Given the facts that the current drug discovery strategies fail to deliver sufficient antifungals for the emerging multidrug resistance, we urgently need to develop novel approaches. By systematically investigating what caused the different antifungal activity of rapamycin in RPMI 1640 and YPD, we discovered that peptide-like compounds can generate a hyper-synergistic antifungal effect with rapamycin on both azole-resistant and sensitive clinical C. albicans isolates. The minimum inhibitory concentration (MIC) of rapamycin reaches as low as 2.14 nM (2-9 µg/mL), distinguishing this drug combination as a hyper-synergism by having a fractional inhibitory concentration (FIC) index ≤ 0.05 from the traditional defined synergism with an FIC index < 0.5. Further studies reveal that this hyper-synergism orthogonally targets the protein Tor1 and affects the TOR signaling pathway in C. albicans, very likely without crosstalk to the stress response, Ras/cAMP/PKA, or calcineurin signaling pathways. These results lead to a novel strategy of controlling drug resistant C. albicans infection in the immunocompromised populations. Instead of prophylactically administering other antifungals with undesirable side-effects for extended durations, we now only need to coadminister some nontoxic peptide additives. The novel antifungal strategy approached in this study not only provides a new therapeutic method to control fungal infections in rapamycin-taking immunocompromised patients but also mitigates the immunosuppressive side-effects of rapamycin, repurposing rapamycin as an antifungal agent with wide applications.

Ethical Applications of Artificial Intelligence: Evidence From Health Research on Veterans.

BACKGROUND: Despite widespread agreement that artificial intelligence (AI) offers significant benefits for individuals and society at large, there are also serious challenges to overcome with respect to its governance. Recent policymaking has focused on establishing principles for the trustworthy use of AI. Adhering to these principles is especially important for ensuring that the development and application of AI raises economic and social welfare, including among vulnerable groups and veterans. OBJECTIVE: We explore the newly developed principles around trustworthy AI and how they can be readily applied at scale to vulnerable groups that are potentially less likely to benefit from technological advances. METHODS: Using the US Department of Veterans Affairs as a case study, we explore the principles of trustworthy AI that are of particular interest for vulnerable groups and veterans. RESULTS: We focus on three principles: (1) designing, developing, acquiring, and using AI so that the benefits of its use significantly outweigh the risks and the risks are assessed and managed; (2) ensuring that the application of AI occurs in well-defined domains and is accurate, effective, and fit for the intended purposes; and (3) ensuring that the operations and outcomes of AI applications are sufficiently interpretable and understandable by all subject matter experts, users, and others. CONCLUSIONS: These principles and applications apply more generally to vulnerable groups, and adherence to them can allow the VA and other organizations to continue modernizing their technology governance, leveraging the gains of AI while simultaneously managing its risks.

Team Science in Precision Medicine: Study of Coleadership and Coauthorship Across Health Organizations.

BACKGROUND: Interdisciplinary collaborations bring lots of benefits to researchers in multiple areas, including precision medicine. OBJECTIVE: This viewpoint aims at studying how cross-institution team science would affect the development of precision medicine. METHODS: Publications of organizations on the eHealth Catalogue of Activities were collected in 2015 and 2017. The significance of the correlation between coleadership and coauthorship among different organizations was calculated using the Pearson chi-square test of independence. Other nonparametric tests examined whether organizations with coleaders publish more and better papers than organizations without coleaders. RESULTS: A total of 374 publications from 69 organizations were analyzed in 2015, and 7064 papers from 87 organizations were analyzed in 2017. Organizations with coleadership published more papers (P<.001, 2015 and 2017), which received higher citations (Z=-13.547, P<.001, 2017), compared to those without coleadership. Organizations with coleaders tended to publish papers together (P<.001, 2015 and 2017). CONCLUSIONS: Our findings suggest that organizations in the field of precision medicine could greatly benefit from institutional-level team science. As a result, stronger collaboration is recommended.

Designing COVID-19 mortality predictions to advance clinical outcomes: Evidence from the Department of Veterans Affairs.

Using administrative data on all Veterans who enter Department of Veterans Affairs (VA) medical centres throughout the USA, this paper uses artificial intelligence (AI) to predict mortality rates for patients with COVID-19 between March and August 2020. First, using comprehensive data on over 10 000 Veterans' medical history, demographics and lab results, we estimate five AI models. Our XGBoost model performs the best, producing an area under the receive operator characteristics curve (AUROC) and area under the precision-recall curve of 0.87 and 0.41, respectively. We show how focusing on the performance of the AUROC alone can lead to unreliable models. Second, through a unique collaboration with the Washington D.C. VA medical centre, we develop a dashboard that incorporates these risk factors and the contributing sources of risk, which we deploy across local VA medical centres throughout the country. Our results provide a concrete example of how AI recommendations can be made explainable and practical for clinicians and their interactions with patients.

Leveraging machine learning to characterize the role of socio-economic determinants on physical health and well-being among veterans.

INTRODUCTION: We investigate the contribution of demographic, socio-economic, and geographic characteristics as determinants of physical health and well-being to guide public health policies and preventative behavior interventions (e.g., countering coronavirus). METHODS: We use machine learning to build predictive models of overall well-being and physical health among veterans as a function of these three sets of characteristics. We link Gallup's U.S. Daily Poll between 2014 and 2017 over a range of demographic and socio-economic characteristics with zipcode characteristics from the Census Bureau to build predictive models of overall and physical well-being. RESULTS: Although the predictive models of overall well-being have weak performance, our classification of low levels of physical well-being performed better. Gradient boosting delivered the best results (80.2% precision, 82.4% recall, and 80.4% AUROC) with perceptions of purpose in the workplace and financial anxiety as the most predictive features. Our results suggest that additional measures of socio-economic characteristics are required to better predict physical well-being, particularly among vulnerable groups, like veterans. CONCLUSION: Socio-economic characteristics explain large differences in physical and overall well-being. Effective predictive models that incorporate socio-economic data will provide opportunities to create real-time and personalized feedback to help individuals improve their quality of life.

vcf2fhir: a utility to convert VCF files into HL7 FHIR format for genomics-EHR integration.

BACKGROUND: VCF formatted files are the lingua franca of next-generation sequencing, whereas HL7 FHIR is emerging as a standard language for electronic health record interoperability. A growing number of FHIR-based clinical genomics applications are emerging. Here, we describe an open source utility for converting variants from VCF format into HL7 FHIR format. RESULTS: vcf2fhir converts VCF variants into a FHIR Genomics Diagnostic Report. Conversion translates each VCF row into a corresponding FHIR-formatted variant in the generated report. In scope are simple variants (SNVs, MNVs, Indels), along with zygosity and phase relationships, for autosomes, sex chromosomes, and mitochondrial DNA. Input parameters include VCF file and genome build ('GRCh37' or 'GRCh38'); and optionally a conversion region that indicates the region(s) to convert, a studied region that lists genomic regions studied by the lab, and a non-callable region that lists studied regions deemed uncallable by the lab. Conversion can be limited to a subset of VCF by supplying genomic coordinates of the conversion region(s). If studied and non-callable regions are also supplied, the output FHIR report will include 'region-studied' observations that detail which portions of the conversion region were studied, and of those studied regions, which portions were deemed uncallable. We illustrate the vcf2fhir utility via two case studies. The first, 'SMART Cancer Navigator', is a web application that offers clinical decision support by linking patient EHR information to cancerous gene variants. The second, 'Precision Genomics Integration Platform', intersects a patient's FHIR-formatted clinical and genomic data with knowledge bases in order to provide on-demand delivery of contextually relevant genomic findings and recommendations to the EHR. CONCLUSIONS: Experience to date shows that the vcf2fhir utility can be effectively woven into clinically useful genomic-EHR integration pipelines. Additional testing will be a critical step towards the clinical validation of this utility, enabling it to be integrated in a variety of real world data flow scenarios. For now, we propose the use of this utility primarily to accelerate FHIR Genomics understanding and to facilitate experimentation with further integration of genomics data into the EHR.

SARS-CoV-2 Drug Discovery based on Intrinsically Disordered Regions.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a close relative of SARS-CoV-1, causes coronavirus disease 2019 (COVID-19), which, at the time of writing, has spread to over 19.9 million people worldwide. In this work, we aim to discover drugs capable of inhibiting SARS-CoV-2 through interaction modeling and statistical methods. Currently, many drug discovery approaches follow the typical protein structure-function paradigm, designing drugs to bind to fixed three-dimensional structures. However, in recent years such approaches have failed to address drug resistance and limit the set of possible drug targets and candidates. For these reasons we instead focus on targeting protein regions that lack a stable structure, known as intrinsically disordered regions (IDRs). Such regions are essential to numerous biological pathways that contribute to the virulence of various viruses. In this work, we discover eleven new SARS-CoV-2 drug candidates targeting IDRs and provide further evidence for the involvement of IDRs in viral processes such as enzymatic peptide cleavage while demonstrating the efficacy of our unique docking approach.

How Much Does the (Social) Environment Matter? Using Artificial Intelligence to Predict COVID-19 Outcomes with Socio-demographic Data.

While the coronavirus pandemic has affected all demographic brackets and geographies, certain areas have been more adversely affected than others. This paper focuses on Veterans as a potentially vulnerable group that might be systematically more exposed to infection than others because of their co-morbidities, i.e., greater incidence of physical and mental health challenges. Using data on 122 Veteran Healthcare Systems (HCS), this paper tests three machine learning models for predictive analysis. The combined LASSO and ridge regression with five-fold cross validation performs the best. We find that socio-demographic features are highly predictive of both cases and deaths-even more important than any hospital-specific characteristics. These results suggest that socio-demographic and social capital characteristics are important determinants of public health outcomes, especially for vulnerable groups, like Veterans, and they should be investigated further.