CardioBot - ECG Segmentation Challenge

CardioBot - ECG Segment

● RUNNING

Demo Mode Active

Observing manual segment isolation and CardioBot automatic analysis loops.

Click inside the canvas or press any key to regain control.

Telemetry Challenge Score

Clinician

CardioBot AI

Telemetry Stream

Normal Sinus Rhythm

P Wave Q Point R Peak S Point T Wave

Active Isolation Tools

Select a wave point, then click near its position on the grid. Keyboard hotkeys [P, Q, R, S, T] are supported.

System Settings

Telemetry Sound

Visual Representation & Marker Legend

To help differentiate between clinician inputs and the automated diagnostic algorithms, this simulator employs contrasting custom marker patterns:

Clinician (User) Markers

Represented on the grid as solid color-filled nodes with a bright outer white outline. Each node is labeled with its waveform letter (P, Q, R, S, T) and displays the label Clinician directly below it.

CardioBot (AI) Markers

Represented on the grid as hollow double-ringed target circles with a small white core point. Each node is labeled with its waveform letter and displays the label CardioBot directly below it.

Overview

CardioBot ECG Segmentation Challenge is a dynamic clinical simulation tool designed to train medical personnel, bioinformaticians, and telemetry students in analyzing electrocardiogram waveforms. By segmenting key complexes—specifically the P-wave, Q-point, R-peak, S-point, and T-wave—users develop an eye for cardiac waveform topology.

As rhythms stream across the visualizer grid, the user competes with CardioBot, an integrated decision system that simulates expert diagnostic routines in real-time. CardioBot leverages physiological thresholds to compute boundaries, presenting an educational playground.

How to Use

Select an Active Isolation Tool: Tap or click any of the five wave target buttons (P, Q, R, S, or T) inside the side panel, or use the physical keys P, Q, R, S, T on your keyboard.

Isolate on the Grid: Click on the signal path displayed on the oscilloscope canvas matching the chosen wave feature. The simulator accommodates minor offset errors with a forgiving circular coordinate selection window of 35 pixels.

Race Against the AI: If you take too long to lock in your detections, CardioBot continuously analyzes the active waveform and claims points step-by-step, plotting its markers in real-time.

Sonification Telemetry: Turn on Telemetry Sound to generate synthetic frequency tracking waves. The physical pitch corresponds directly to the microvolt deviation of the signal.

Automated Demonstration: Let the simulation idle for 45 seconds or press the Observe AI Demo button to launch a read-only demonstration loop. Simply touch or interact anywhere to reclaim manual clinical operations instantly.

Global Reset: Hit Reset Simulation at any time to clear historical performance logs, reset telemetry scores, and reinitialize local caches.

Technical Details

Signal Processing and Waveform Architectures

The underlying engine processes synthetic ECG complexes mapped directly from clinically realistic spatial profiles representing:

Normal Sinus Rhythm: Perfect baseline pacing driven by healthy Sinoatrial (SA) nodes with smooth sinusoidal transitions.

Atrial Fibrillation (A-Fib): Chaotic atrial noise consisting of constant small baseline oscillations (f-waves), absence of P-waves, and irregular ventricular intervals.

Premature Ventricular Contractions (PVC): Broad, aberrant ventricular depolarizations presenting massive anomalous, wide QRS complexes followed by discordant inverted T-waves.

Ventricular Tachycardia (V-Tach): Rapid, repetitive monomorphic wave shapes representing high-amplitude ventricular distress.

The Decision Matrix

CardioBot implements a multi-variate decision matrix modeling clinical rules of uncertainty. It measures temporal parameters like the PR interval, QRS duration, and peak-to-peak variability. By projecting numerical parameters into linguistic ranges (e.g., "Normal Duration," "Prolonged Block," "Irregular Ventricular Rate"), the automated decision layer computes physiological classifications with native browser speed, achieving Interaction to Next Paint (INP) frames under the target of 200 milliseconds.