# MIT-BIH four-record prestudy

This public derivative packages a bounded technical R-peak detection pre-study for records 100, 105, 118, and 119 of the MIT-BIH Arrhythmia Database. The fixed protocol uses the MLII lead, 360 Hz sampling, and a ±150 ms (±54 sample) matching window.

The canonical aggregate is TP 9092, FP 28, FN 18, with 46 error events and 16 reciprocal counterpart pairs / 32 linked events. The error-stratification figures are materialized from Build 544c45cd1dbe33d85a5de8184c5594061a5fb605; this is a result version, not a new analysis run.

The original MIT-BIH data are not included. The source archive contains self-authored code only. The dataset must be obtained from PhysioNet under its own ODC Attribution 1.0 terms.

本项目仅供教学、实验和辅助分析，不可替代医生诊断。