DiabAgent: Toward Agentic Physiological Control
arXiv cs.AI · 3 Nov 2026, 04:41 AEST
DiabAgent introduces a framework for agentic physiological control in managing Type 1 diabetes. The research suite includes a simulator and benchmark for evaluating glucose control strategies.
Key facts
- As pretraining scales, medical language models have advanced rapidly in clinical question answering.
- We introduce DiabAgent, a research suite for agentic physiological control comprising the Agentic Physiological Control (APC) framework, the DiabetSim simulator, and the SweetBench benchmark.
- Using DiabetSim, we construct SweetBench by instantiating 32 scenario designs as 212 glucose-control tasks.
- With Claude Opus 5.5 and a fixed decision interval of one hour, APC achieves a COGI of 73.5, compared with 71.2 for the open-source closed-loop baseline OpenAPS.
Read aloud
DiabAgent introduces a framework for agentic physiological control in managing Type one diabetes. The research suite includes a simulator and benchmark for evaluating glucose control strategies. As pretraining scales, medical language models have advanced rapidly in clinical question answering. We introduce DiabAgent, a research suite for agentic physiological control comprising the Agentic Physiological Control APC framework, the DiabetSim simulator, and the SweetBench benchmark. Using DiabetSim, we construct SweetBench by instantiating thirty two scenario designs as two hundred twelve glucose-control tasks. With Claude Opus five point five and a fixed decision interval of one hour, APC achieves a COGI of seventy three point five, compared with seventy one point two for the open-source closed-loop baseline OpenAPS.
Sources in this cluster
- DiabAgent: Toward Agentic Physiological Control
arXiv cs.AI · 3 Nov 2026, 04:41 AEST