Sepsis Early-Warning CUSUM Detector (2D)
Interactive 2D ICU big-data simulator: a streaming heart-rate trend feeds a CUSUM (cumulative sum) statistical detector, catching a slow physiological drift toward sepsis-driven tachycardia hours before a fixed-threshold alarm would fire. Includes a Monte Carlo comparison of detection lead time and false-alarm rate.
Modern hospitals generate a continuous flood of vital-sign data per patient — the "medical big data" problem is turning that stream into an early warning instead of noise. This 2D simulator plots a live heart-rate trace alongside a CUSUM (cumulative sum) statistical detector tracking the same stream. Trigger a slow physiological deterioration and compare how quickly the CUSUM statistic accumulates enough evidence to alarm against a conventional fixed-threshold monitor that only reacts once a single reading crosses its line. A built-in Monte Carlo mode replays hundreds of independent synthetic patient streams offline to measure the average detection lead time and each detector's real false-alarm rate at your chosen k/h settings — the same tradeoff hospital early-warning systems are tuned against in practice.
A streaming ICU heart-rate trend feeds a CUSUM (cumulative sum) statistical detector that catches a slow drift toward sepsis-driven tachycardia hours before a conventional fixed-threshold alarm would fire.
2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install