E1.M1

DATASET DESCRIPTION

Disease Area

Cardiovascular Disease

Specific Research Focus

Atrial Fibrillation Risk Prediction in Older Adults (55–90 yrs) 

Dataset Theme / Edition

AFib Risk Stratification — Community + Cardiology Screening Mix

Short Dataset Summary

10,000 synthetic patient records encoding cardiovascular demographics, comorbidities, vitals, labs, cardiac indicators, and wearable data for binary AFib diagnosis classification. Prevalence calibrated to 17.5%, consistent with real-world AFib rates in adults 55+. All features generated under CHA₂DS₂-VASc-aligned clinical dependency rules with physiological constraint enforcement. 

Intended Research Applications

AFib risk classification · CHA₂DS₂-VASc model benchmarking · SHAP/LIME explainability · Healthcare screening analytics · Missing data imputation research · Federated learning simulation |

Total Number of Patients - 10,000 

Number of Features / Parameters

34 (9 demographic/lifestyle · 6 medical history · 3 vitals · 5 labs · 7 cardiac/functional · 3 wearable · 1 target) 

Data Modalities Included

Structured EHR-style tabular · Clinical vitals · Laboratory values · Cardiac biomarkers (HRV, ECG) · Consumer wearable signals (steps, HRV trend) 

Recommended AI Tasks

Binary classification · Explainable AI (SHAP/LIME) · Feature selection · Risk score calibration · Class imbalance benchmarking 

Dataset Size

 ~1.2 MB (CSV) + 7 KB data dictionary 

Release Date

May 2026 

Why This Dataset Exists

No large, publicly available synthetic dataset exists specifically for AFib risk prediction in older adults with CHA₂DS₂-VASc-aligned feature architecture, multi-ethnic representation, and structured wearable data — while preserving clinical realism and physiological constraint compliance. 

Real-World Problem Addressed

 Atrial fibrillation affects ~43.6 million people globally and confers a 4–5× increased stroke risk. Early AI-assisted detection in community screening is hampered by lack of high-quality, privacy-safe training data that reflects real older adult clinical populations.