Thinking at the Edge: Neuromorphic Computing's Biomedical Moment

Abstract. Neuromorphic computing — event-driven, spiking neural network (SNN) inference on purpose-built silicon — offers a potentially transformative solution to the power and latency constraints of biomedical edge AI. Recent hardware demonstrations have achieved seizure detection at below 300 μW on wearable platforms, cardiac arrhythmia classification compatible with coin-cell operation, and neuroprosthetic control systems with sub-millisecond latency. Yet not one neuromorphic SNN-based medical device has received regulatory clearance anywhere in the world. This paper provides the first systematic analysis of that gap. We assign technology readiness level (TRL) estimates across three primary clinical domains — epilepsy monitoring (TRL 5–6), cardiac arrhythmia detection (TRL 4–5), and neuroprosthetics (TRL 3–4) — and assess hardware and algorithmic readiness across five dimensions: device variability, long-term stability, patient specificity, benchmark standardisation, and algorithm–hardware co-design. We then identify and characterise five structural regulatory gaps: absence of a cleared neuromorphic predicate; incompatibility of stochastic spike dynamics with deterministic verification standards; failure of the Predetermined Change Control Plan framework to accommodate continuous on-device learning; lack of standardised clinical benchmarks; and insufficient biocompatibility data for novel neuromorphic substrate materials. We propose a phased translational roadmap targeting first-in-human feasibility studies by 2028 and first regulatory clearance by 2030. The roadmap assigns specific actions to chip manufacturers, standards bodies, regulators, and clinical research teams. The analysis concludes with ethical, equity, and data governance considerations relevant to the deployment of adaptive neuromorphic inference in vulnerable patient populations. Keywords: neuromorphic computing, spiking neural networks, medical device regulation, epilepsy, brain-computer interfaces, edge AI, FDA, translational roadmap