Uloženo v:
Podrobná bibliografie
Hlavní autoři: Syed, Toqeer Ali, Akarma, Ali, Ali, Ahmad, Lee, It Ee, Jan, Salman, Khan, Sohail, Nauman, Muhammad
Médium: Recurso digital
Jazyk:
Vydáno: Zenodo 2026
Témata:
On-line přístup:https://doi.org/10.5281/zenodo.18891708
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
_version_ 1866901472546914304
author Syed, Toqeer Ali
Akarma, Ali
Ali, Ahmad
Lee, It Ee
Jan, Salman
Khan, Sohail
Nauman, Muhammad
author_facet Syed, Toqeer Ali
Akarma, Ali
Ali, Ahmad
Lee, It Ee
Jan, Salman
Khan, Sohail
Nauman, Muhammad
contents Official implementation of the PAAI framework — a four-layer constrained reinforcement learning and BDI multi-agent architecture for chronic disease management in IoT healthcare. Combines a custom Gymnasium environment, MaskablePPO with Lagrangian safety constraints, a hash-chained audit log, and a three-tier Human-in-the-Loop governance model. Validated on a 12-month synthetic cohort of 500 patients and on MIMIC-IV real ICU data.
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_18891708
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle AgHealth+: Privacy-Aware Agentic AI for IoT Healthcare (PAAI Framework)
Syed, Toqeer Ali
Akarma, Ali
Ali, Ahmad
Lee, It Ee
Jan, Salman
Khan, Sohail
Nauman, Muhammad
agentic AI
IoT healthcare
reinforcement learning
BDI agents
privacy-preserving
human-in-the-loop
chronic disease management
MIMIC-IV
constrained MDP
Official implementation of the PAAI framework — a four-layer constrained reinforcement learning and BDI multi-agent architecture for chronic disease management in IoT healthcare. Combines a custom Gymnasium environment, MaskablePPO with Lagrangian safety constraints, a hash-chained audit log, and a three-tier Human-in-the-Loop governance model. Validated on a 12-month synthetic cohort of 500 patients and on MIMIC-IV real ICU data.
title AgHealth+: Privacy-Aware Agentic AI for IoT Healthcare (PAAI Framework)
topic agentic AI
IoT healthcare
reinforcement learning
BDI agents
privacy-preserving
human-in-the-loop
chronic disease management
MIMIC-IV
constrained MDP
url https://doi.org/10.5281/zenodo.18891708