SKIMS-ECG: A Clinically Validated ECG Feature Dataset for Multi-Class Cardiovascular Disease Detection

SKIMS-ECG is a clinically validated cardiovascular disease dataset derived from real-time 12-lead electrocardiogram (ECG) recordings collected at the Sher-i-Kashmir Institute of Medical Sciences (SKIMS), Srinagar, India. The dataset comprises 24,894 ECG examinations, each independently reviewed and annotated by experienced cardiologists, ensuring high diagnostic reliability and clinical relevance.The dataset includes 20 attributes per record, consisting of 10 ECG morphological features, 9 demographic and clinical risk-factor variables, and 1 target class label. ECG features capture key physiological characteristics such as heart rate, PR interval, QRS duration, QT/QTc interval, electrical axes (P, QRS, R, and T), RV5/SV1 voltage, and R-wave amplitude in lead aVR. Demographic and clinical attributes include age, gender, systolic and diastolic blood pressure, diabetes status, cholesterol status, smoking history, medication usage, and family history of cardiovascular disease.Each ECG record is classified into one of six major cardiovascular disease categories: normal, cardiac arrhythmia, coronary heart disease (CHD), cardiomyopathy, stroke-related ECG patterns, and heart failure. The dataset is balanced, with an equal number of samples in each class, making it well-suited for unbiased multi-class classification and benchmarking studies.This release contains two CSV files:(i) a raw feature dataset with original extracted ECG and clinical attributes, and(ii) a min-max normalized dataset, where all numerical features are scaled to the range [0,1] to support stable and reproducible machine learning experiments. Both files share identical structure, column names, and row ordering.ECG recordings were acquired using the BioCare ECG-1210 system at a sampling rate of 250 Hz with a 10-second duration, following standard resting ECG protocols. Signal processing, feature extraction, and data management were performed using BioCare ECG-1000 software, followed by expert cardiologist verification.The dataset was created in full compliance with ethical research standards. Ethical approval was obtained from the Institutional Ethics Committee, Lovely Professional University, Punjab, India (Ref. No.: IEC-LPU/2025/1/19). All data are fully anonymized, and no personally identifiable information is included.SKIMS-ECG is intended for academic and non-commercial research and supports a wide range of applications, including machine learning and deep learning-based cardiovascular disease detection, soft-computing and optimization studies, explainable AI (XAI), clinical decision-support systems, and research on underrepresented Himalayan populations.