From Data to Daily Decisions: A Safety-Constrained, AI-Assisted N-of-1 Framework for Back Rehabilitation

A Structured, Rule-Based Approach to Consistent Daily Rehabilitation Decisions

Abstract

This paper presents a lightweight, safety-constrained, semi-automated agentic decision-support framework for daily rehabilitation. The system combines structured user inputs, rule-based safety triage, and conservative plan generation to support consistent day-to-day decisions. In a single-case (N-of-1) implementation based on self-reported daily logs, the framework was associated with reduced symptom volatility, clearer identification of triggers, and improved standing and walking tolerance over the observation period. Here, “agentic” refers to a human-in-the-loop system that translates inputs into constrained, rule-based actions. The framework is non-diagnostic and is intended to support structured, safe decision-making under uncertainty.