Saved in:
Bibliographic Details
Main Authors: Basu, Sanjay, Patel, Sadiq Y., Sheth, Parth, Muralidharan, Bhairavi, Elamaran, Namrata, Kinra, Aakriti, Batniji, Rajaie
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.09655
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911150434680832
author Basu, Sanjay
Patel, Sadiq Y.
Sheth, Parth
Muralidharan, Bhairavi
Elamaran, Namrata
Kinra, Aakriti
Batniji, Rajaie
author_facet Basu, Sanjay
Patel, Sadiq Y.
Sheth, Parth
Muralidharan, Bhairavi
Elamaran, Namrata
Kinra, Aakriti
Batniji, Rajaie
contents We introduce Feasibility-Guided Fair Adaptive Reinforcement Learning (FG-FARL), an offline RL procedure that calibrates per-group safety thresholds to reduce harm while equalizing a chosen fairness target (coverage or harm) across protected subgroups. Using de-identified longitudinal trajectories from a Medicaid population health management program, we evaluate FG-FARL against behavior cloning (BC) and HACO (Hybrid Adaptive Conformal Offline RL; a global conformal safety baseline). We report off-policy value estimates with bootstrap 95% confidence intervals and subgroup disparity analyses with p-values. FG-FARL achieves comparable value to baselines while improving fairness metrics, demonstrating a practical path to safer and more equitable decision support.
format Preprint
id arxiv_https___arxiv_org_abs_2509_09655
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Feasibility-Guided Fair Adaptive Offline Reinforcement Learning for Medicaid Care Management
Basu, Sanjay
Patel, Sadiq Y.
Sheth, Parth
Muralidharan, Bhairavi
Elamaran, Namrata
Kinra, Aakriti
Batniji, Rajaie
Machine Learning
Artificial Intelligence
Logic in Computer Science
Applications
We introduce Feasibility-Guided Fair Adaptive Reinforcement Learning (FG-FARL), an offline RL procedure that calibrates per-group safety thresholds to reduce harm while equalizing a chosen fairness target (coverage or harm) across protected subgroups. Using de-identified longitudinal trajectories from a Medicaid population health management program, we evaluate FG-FARL against behavior cloning (BC) and HACO (Hybrid Adaptive Conformal Offline RL; a global conformal safety baseline). We report off-policy value estimates with bootstrap 95% confidence intervals and subgroup disparity analyses with p-values. FG-FARL achieves comparable value to baselines while improving fairness metrics, demonstrating a practical path to safer and more equitable decision support.
title Feasibility-Guided Fair Adaptive Offline Reinforcement Learning for Medicaid Care Management
topic Machine Learning
Artificial Intelligence
Logic in Computer Science
Applications
url https://arxiv.org/abs/2509.09655