The Autonomous Clinic: Navigating AI, Sensors and the Equity Gap in Modern Physiotherapy

Background: The coronavirus disease 2019 pandemic converted telerehabilitation in physiotherapy from a marginal service model into a mainstream one within weeks, and generated an unusually rapid accumulation of trial evidence, professional guidance and health-system experience.

Purpose and Scope: This critical narrative review examines three interlocking questions that the post-pandemic literature has largely addressed in isolation: how strong the effectiveness evidence for physiotherapy telerehabilitation actually is, whether the model narrows or widens inequities in access to rehabilitation, and why utilisation contracted in many settings once emergency conditions ended despite favourable evidence.

Approach: Peer-reviewed literature was identified through structured searching of biomedical and multidisciplinary scholarly indexes, supplemented by citation tracking and authoritative institutional sources, with priority given to randomised trials, quantitative syntheses, implementation studies and equity-focused analyses.

Principal Findings: Equivalence with in-person care is reasonably well supported for supervised exercise-based interventions in musculoskeletal, cardiac and chronic respiratory populations, and adherence and satisfaction are generally comparable or better. Confidence weakens considerably where interventions depend on manual assessment or handling, where comparators are inactive, and where effect estimates are pooled across heterogeneous delivery architectures. Equity claims rest on an unresolved inversion: the populations with the greatest unmet rehabilitation need frequently have the weakest connectivity, device access and digital confidence, and they are systematically under-represented in the trials that underpin equivalence claims. The contraction of utilisation after the emergency phase is better explained by reimbursement instability, organisational routines, physical infrastructure and professional identity than by patient or clinician rejection of the model.

Implications: Telerehabilitation should be treated as a service configuration requiring deliberate design and targeting rather than as an intervention with a fixed effect.

Conclusion: The central unresolved problem is no longer whether physiotherapy telerehabilitation can work, but for whom, in what combination with in-person care, and under what system conditions it can be sustained without displacing the people it was intended to reach.