Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.
Self-evolving health agent with a LoRA memory-admission gate. Wearable-adjacent (orchestrates longitudinal personal health / wearable summaries), not a PPG waveform FM.
Parameters
9B
Signals
EHR, Text, wearable summaries
Form factor
Agent
Year
2026
Architecture
HealthJev agent stack (Streamlit/Feishu/CLI) + JevGate-9B = frozen Qwen3.5-9B-Base + LoRA (r=16, alpha=32) + 256-d pointer head for memory admission
License
Code MIT; JevGate adapter/head Apache-2.0; Qwen base separate
Medical large language models and agents show strong single-encounter question answering and evidence interpretation but struggle to accumulate and govern experience across encounters. We present HealthJev, an open-source agent separating shared safety rules and medical knowledge from a private longitudinal memory and distilling lessons from completed cases. To filter out erroneous lessons before they affect future behaviour, we self-train JevGate, a lightweight System-One model that decides which lessons enter memory; each decision takes a median of 86 milliseconds and about 1,000 input tokens, sustaining experience accumulation at negligible run-time overhead. On a longitudinal benchmark of 90-day simulated daily interactions for 20 users built from real MIMIC-IV trajectories, HealthJev outperformed a memory-free ReAct agent and three memory-augmented frameworks, with all paired comparisons significant. Across six tasks spanning clinical records, medical imaging, genomics and clinical text, HealthJev improved every task with a mean gain of 13.3 percentage points over the base condition, surpassing all baselines. The frozen library transferred to six further backbones with macro-average gains of 6.8 to 15.1 percentage points, and one task's library brought gains on other tasks. These results indicate that governed, self-evolving memory supports longitudinal personal health management beyond single encounters. HealthJev is publicly available at https://github.com/HC-Guo/Self-Evolving_HealthJev.
Verbatim fromthe publisher's DOI record· Publisher licence: https://creativecommons.org/licenses/by/4.0/
How it works
Architecture
HealthJev agent stack (Streamlit/Feishu/CLI) + JevGate-9B = frozen Qwen3.5-9B-Base + LoRA (r=16, alpha=32) + 256-d pointer head for memory admission
Inputs
EHR, Text, wearable summaries
Approach
LLM agent · health agent · JevGate · Qwen · MIT · Apache 2.0
Cite the publication recorded for this model. Check it against the publisher before submitting.
Guo, H., Li, H., Deng, J., Jin, T., Han, J., Wang, Y., Wang, Z., Sun, J., Si, Q., Gong, W., Zhuang, X., & You, J. (2026). HealthJev: A Self-Evolving Agent for Longitudinal Personal Health Management (Guo et al., Research Square rs-11605391). Research Square rs-11605391 v1. https://doi.org/10.21203/rs.3.rs-11605391/v1
Guo, Hongcheng, Haoran Li, Jiebi Deng, Tong Jin, Jinghong Han, Yuxin Wang, and Zexin Wang, et al. 2026. "HealthJev: A Self-Evolving Agent for Longitudinal Personal Health Management (Guo et al., Research Square rs-11605391)." Research Square rs-11605391 v1. https://doi.org/10.21203/rs.3.rs-11605391/v1.
Guo, H., Li, H., Deng, J., Jin, T., Han, J., Wang, Y. and Wang, Z. et al. (2026) 'HealthJev: A Self-Evolving Agent for Longitudinal Personal Health Management (Guo et al., Research Square rs-11605391)', Research Square rs-11605391 v1. doi: 10.21203/rs.3.rs-11605391/v1.
Guo, Hongcheng, et al. "HealthJev: A Self-Evolving Agent for Longitudinal Personal Health Management (Guo et al., Research Square rs-11605391)." Research Square rs-11605391 v1, 2026. https://doi.org/10.21203/rs.3.rs-11605391/v1.
H. Guo et al., "HealthJev: A Self-Evolving Agent for Longitudinal Personal Health Management (Guo et al., Research Square rs-11605391)," Research Square rs-11605391 v1, 2026, doi: 10.21203/rs.3.rs-11605391/v1.
@misc{guo2026healthjev,
title = {{HealthJev: A Self-Evolving Agent for Longitudinal Personal Health Management (Guo et al., Research Square rs-11605391)}},
author = {Guo, Hongcheng and Li, Haoran and Deng, Jiebi and Jin, Tong and Han, Jinghong and Wang, Yuxin and Wang, Zexin and Sun, Jinyu and Si, Qingyi and Gong, Weikang and Zhuang, Xiahai and You, Jia},
year = {2026},
howpublished = {Research Square rs-11605391 v1},
doi = {10.21203/rs.3.rs-11605391/v1}
}
History
Added to the ChatPPG research hub
Newsletter
Get new PPG datasets, papers and models
Get alerts when HealthJev / JevGate-9B is benchmarked on new datasets, plus the weekly digest.
New datasets, papers and models
Practical guides from the hub
Unsubscribe anytime
The waveform is an illustration of a PPG pulse, not a recording.