Charles P. Rizzo, Daniel Mallett, Maria Hernandez Rivero, Jamin Hanna, Chris White, Steven Dao, Jonathan Tran, Bryson Gullett, Catherine D. Schuman and James S. Plank

July, 2025

ICONS: International Conference on Neuromorphic Systems

https://iconsneuromorphic.cc/Schedule-2025/

View Article

Abstract

The artificial pancreas (AP), a system consisting of a continuous glucose monitor, insulin pump, and control algorithm, represents the state of the art in automated blood glucose control for individuals with type 1 diabetes. Nevertheless, there is ongoing research aimed at improving the overall efficacy of these systems. The control algorithm has been a particular area of focus, with machine learning (ML) approaches such as deep learning and reinforcement learning (RL) having recently been applied to the problem of blood glucose control. A notable drawback of traditional ML solutions, however, is their large size which can make them inadequate for deployment to edge environments such as an insulin pump or smartphone. The control algorithm must be effective at regulating blood glucose as well as efficient to minimize the need for human intervention (e.g., to recharge the device on which the algorithm is running). In our work we apply neuromorphic computing to the problem of blood glucose regulation, training personalized spiking neural networks (SNNs) to act as the control algorithm for a simulated AP and using virtual patients provided by the Simglucose implementation of the UVA/Padova metabolic simulator. Our results demonstrate that an SNN acting as the control algorithm can outperform existing reinforcement learning models—achieving an average Time in Range (TIR) of 92.4%, 83.8%, and 79.9% for the adult, adolescent, and child cohort, respectively—while simultaneously consuming minimal power.

Citation Information

Text

author     C. P. Rizzo and D. Mallett and M. {Hernandez Rivero} and J. Hanna and 
           C. White and S. Dao and J. Tran and B. Gullett and C. D. Schuman and J. S. Plank
title      {NAP}: Neuromorphic Artificial Pancreas
booktitle  International Conference on Neuromorphic Computing Systems (ICONS)
publisher  ACM
month      July
year       2025

Bibtex

@INPROCEEDINGS{rmh:25:nap,
    author = "C. P. Rizzo and D. Mallett and M. {Hernandez Rivero} and J. Hanna and 
               C. White and S. Dao and J. Tran and B. Gullett and C. D. Schuman and J. S. Plank",
    title = "{NAP}: Neuromorphic Artificial Pancreas",
    booktitle = "International Conference on Neuromorphic Computing Systems (ICONS)",
    publisher = "ACM",
    month = "July",
    year = "2025"
}