Cerebellum‑Inspired AI Chip Detects Arrhythmias with 98% Accuracy Using 10,000× Fewer Calculations

A team of researchers from Northwestern University has engineered a cerebellum‑inspired AI chip that mirrors the brain’s instinctive motor response system.

Cerebellum‑Inspired AI Chip Breaks Efficiency Barrier

In simulated ECG streams the device flagged arrhythmias in less than one‑fifth of a heartbeat, delivering ↑ 98% accuracy and operating roughly ↑ 10,000× faster than conventional transformer models. The chip’s core is a memtransistor built from atom‑thin molybdenum disulfide, merging memory and compute to sidestep the von Neumann bottleneck. Neuromorphic computing inspired by the cerebellum allows the circuit to ignore predictable signals and fire only on anomalies, conserving power.

“The cerebellum excels at reserving resources for the unexpected, which translates into lower energy consumption,”

said Mark Hersam, professor of materials science at Northwestern, in a statement to Reuters. Early tests suggest the architecture could enable edge‑deployed health monitors, autonomous vehicles and robotics that react in real time without reliance on massive data‑center farms. Industry analysts at Bloomberg note that reducing AI’s data‑center footprint is pivotal as global electricity demand from AI‑driven servers nears 945 TWh by 2030.