The Care Plan Compiler as an Agentic AI Framework for Translating Clinical Recommendations into Feasible Patient-Centered Care

Sakshit Teotia

Citation: Sakshit Teotia, "The Care Plan Compiler as an Agentic AI Framework for Translating Clinical Recommendations into Feasible Patient-Centered Care", Universal Library of Innovative Research and Studies, Volume 03, Issue 04.

Copyright: This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

A clinical recommendation becomes care after someone confirms coverage, finds a facility that provides the service, arranges language support and transportation, books an appointment, and verifies that the visit took place. When no organization owns this work, it falls to the patient, and a missed service is recorded as nonadherence, even when the failure occurred within the health system. The review examines five adjacent fields, each addressing part of this problem, and develops the Care Plan Compiler, an agentic AI framework that treats the clinician-approved recommendation as a protected source specification and compiles it into a graph of tasks with named owners. A deterministic readiness checker releases a task once all prerequisites hold and classifies any unknown coverage, capacity, or accessibility as unresolved. Feasible plans are compared across a multidimensional burden vector, and a change in context triggers recompilation of the dependent tasks. A synthetic case of diabetic retinal screening illustrates the compilation, and a four-stage evaluation program sets a screening completed within the approved window and verified within 90 days of referral approval as its primary endpoint. The framework moves responsibility for missed care to the system layer where the failure occurs.


Keywords: Agentic AI, Care Planning, Treatment Burden, Patient Navigation, Clinical Workflow, Human Oversight, Health Equity.

Download doi https://doi.org/10.70315/uloap.ulirs.2026.0304004