Previous genome-wide association studies (GWASs) of stroke — the second leading cause of death worldwide — were conducted predominantly in populations of European ancestry^1,2. Here, in cross-ancestry GWAS meta-analyses of 110,182 patients who have had a stroke (five ancestries, 33% non-European) and 1,503,898 control individuals, we identify association signals for stroke and its subtypes at 89 (61 new) independent loci: 60 in primary inverse-variance-weighted analyses and 29 in secondary meta-regression and multitrait analyses. On the basis of internal cross-ancestry validation and an independent follow-up in 89,084 additional cases of stroke (30% non-European) and 1,013,843 control individuals, 87% of the primary stroke risk loci and 60% of the secondary stroke risk loci were replicated (P < 0.05). Effect sizes were highly correlated across ancestries. Cross-ancestry fine-mapping, in silico mutagenesis analysis³, and transcriptome-wide and proteome-wide association analyses revealed putative causal genes (such as SH3PXD2A and FURIN) and variants (such as at GRK5 and NOS3). Using a three-pronged approach⁴, we provide genetic evidence for putative drug effects, highlighting F11, KLKB1, PROC, GP1BA, LAMC2 and VCAM1 as possible targets, with drugs already under investigation for stroke for F11 and PROC. A polygenic score integrating cross-ancestry and ancestry-specific stroke GWASs with vascular-risk factor GWASs (integrative polygenic scores) strongly predicted ischaemic stroke in populations of European, East Asian and African ancestry⁵. Stroke genetic risk scores were predictive of ischaemic stroke independent of clinical risk factors in 52,600 clinical-trial participants with cardiometabolic disease. Our results provide insights to inform biology, reveal potential drug targets and derive genetic risk prediction tools across ancestries.

Laboratory data from Electronic Health Records (EHR) are often used in prediction models where estimation bias and model performance from missingness can be mitigated using imputation methods. We demonstrate the utility of imputation in two real-world EHR-derived cohorts of ischemic stroke from Geisinger and of heart failure from Sutter Health to: (1) characterize the patterns of missingness in laboratory variables; (2) simulate two missing mechanisms, arbitrary and monotone; (3) compare cross-sectional and multi-level multivariate missing imputation algorithms applied to laboratory data; (4) assess whether incorporation of latent information, derived from comorbidity data, can improve the performance of the algorithms. The latter was based on a case study of hemoglobin A1c under a univariate missing imputation framework. Overall, the pattern of missingness in EHR laboratory variables wasnot at randomand was highly associated with patients’ comorbidity data; and the multi-level imputation algorithm showed smaller imputation error than the cross-sectional method.

Background: The long-term risk of recurrent ischemic stroke, estimated to be between 17% and 30%, cannot be reliably assessed at an individual level. Our goal was to study whether machine-learning can be trained to predict stroke recurrence and identify key clinical variables and assess whether performance metrics can be optimized. Methods: We used patient-level data from electronic health records, six interpretable algorithms (Logistic Regression, Extreme Gradient Boosting, Gradient Boosting Machine, Random Forest, Support Vector Machine, Decision Tree), four feature selection strategies, five prediction windows, and two sampling strategies to develop 288 models for up to 5-year stroke recurrence prediction. We further identified important clinical features and different optimization strategies. Results: We included 2091 ischemic stroke patients. Model area under the receiver operating characteristic (AUROC) curve was stable for prediction windows of 1, 2, 3, 4, and 5 years, with the highest score for the 1-year (0.79) and the lowest score for the 5-year prediction window (0.69). A total of 21 (7%) models reached an AUROC above 0.73 while 110 (38%) models reached an AUROC greater than 0.7. Among the 53 features analyzed, age, body mass index, and laboratory-based features (such as high-density lipoprotein, hemoglobin A1c, and creatinine) had the highest overall importance scores. The balance between specificity and sensitivity improved through sampling strategies. Conclusion: All of the selected six algorithms could be trained to predict the long-term stroke recurrence and laboratory-based variables were highly associated with stroke recurrence. The latter could be targeted for personalized interventions. Model performance metrics could be optimized, and models can be implemented in the same healthcare system as intelligent decision support for targeted intervention.

Background: Developing a decision support system based on advances in machine learning is one area for strategic innovation in healthcare. Predicting a patient’s progression to septic shock is an active field of translational research. The goal of this study was to develop a working model of a clinical decision support system for predicting septic shock in an acute care setting for up to 6 h from the time of admission in an integrated healthcare setting. Method: Clinical data from Electronic Health Record (EHR), at encounter level, were used to build a predictive model for progression from sepsis to septic shock up to 6 h from the time of admission; that is, T = 1, 3, and 6 h from admission. Eight different machine learning algorithms (Random Forest, XGBoost, C5.0, Decision Trees, Boosted Logistic Regression, Support Vector Machine, Logistic Regression, Regularized Logistic, and Bayes Generalized Linear Model) were used for model development. Two adaptive sampling strategies were used to address the class imbalance. Data from two sources (clinical and billing codes) were used to define the case definition (septic shock) using the Centers for Medicare & Medicaid Services (CMS) Sepsis criteria. The model assessment was performed using Area under Receiving Operator Characteristics (AUROC), sensitivity, and specificity. Model predictions for each feature window (1, 3 and 6 h from admission) were consolidated. Results: Retrospective data from April 2005 to September 2018 were extracted from the EHR, Insurance Claims, Billing, and Laboratory Systems to create a dataset for septic shock detection. The clinical criteria and billing information were used to label patients into two classes-septic shock patients and sepsis patients at three different time points from admission, creating two different case-control cohorts. Data from 45,425 unique in-patient visits were used to build 96 prediction models comparing clinical-based definition versus billing-based information as the gold standard. Of the 24 consolidated models (based on eight machine learning algorithms and three feature windows), four models reached an AUROC greater than 0.9. Overall, all the consolidated models reached an AUROC of at least 0.8820 or higher. Based on the AUROC of 0.9483, the best model was based on Random Forest, with a sensitivity of 83.9% and specificity of 88.1%. The sepsis detection window at 6 h outperformed the 1 and 3-h windows. The sepsis definition based on clinical variables had improved performance when compared to the sepsis definition based on only billing information. Conclusion: This study corroborated that machine learning models can be developed to predict septic shock using clinical and administrative data. However, the use of clinical information to define septic shock outperformed models developed based on only administrative data. Intelligent decision support tools can be developed and integrated into the EHR and improve clinical outcomes and facilitate the optimization of resources in real-time.

We examined whether common variants from the extended major histocompatibility complex (xMHC) region contribute to the response to antipsychotic drugs (APDs) in patients with schizophrenia with persistent psychosis. Subjects participated in a prospective longitudinal study of the effect of APDs on psychopathology were temporally split into discovery (n = 88) and replication (n = 42) cohorts. The primary endpoint was a change in Brief Psychiatric Rating Scale at 6-week or 6-month after treatment. rs204991 (β = 3.917, p = 3.72 × 10⁻⁶), the strongest signal associated with response at 6-week was located near C4A/C4B after a linear regression adjusted for covariates. xMHC SNP imputation disclosed much stronger signals (rs9268469, β = 5.140, p = 1.57 × 10⁻⁷) and other weaker signals (p < 1 × 10⁻⁵) spanning the entire xMHC region. All the variants were previously identified schizophrenia risk loci. Conditional fine-mapping revealed three subgroups of SNPs which were the eQTLs (p < 1 × 10⁻⁷) for C4A, HLA-C, and BTN3A2 in disease-relevant tissue. Epistasis between HLA-C and C4A was observed (p = 0.019). Minor allele (G) carriers of rs204991, eQTL for C4A, having decreased risk for schizophrenia and lower imputed expression of C4A, had a better response to APDs. Some imputed HLA alleles associated with a decreased risk for schizophrenia had a positive association with improvement in psychotic symptoms. An independent cohort validated the association of change in psychosis with C4A. We provide evidence that genetic risk factors for schizophrenia from the xMHC region are associated with response to APDs and those variants significantly alter the imputed expression of C4A, HLA-C, and BTN3A2. The minor alleles predicting higher C4A level are associated with diminished improvement in psychotic symptoms after APD treatment.