Feasibility-Guided Fair Adaptive Offline Reinforcement Learning for Medicaid Care Management
This work addresses fairness and safety in healthcare decision support for Medicaid populations, representing an incremental improvement over existing methods.
The paper tackled the problem of reducing harm and equalizing fairness across subgroups in offline reinforcement learning for Medicaid care management, achieving comparable value to baselines while improving fairness metrics.
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.