One of the big stories in health care this year is the way digitalization and technology are rapidly changing practice. Clinics and health systems are increasingly adopting different types of AI-powered platforms and digital assessment tools that can help patients get faster access to care and monitor progress more precisely. These tools are seen as complements to, not replacements for, human clinicians, helping with movement analysis, patient follow-ups, and data-driven care.
Currently, there is no human-like Artificial General Intelligence (AGI). Instead, today’s systems consist of different types of narrow AI designed to perform specific tasks.
| AI Application | Clinical Use | AI Method | Evidence Level (2026) |
|---|---|---|---|
| Movement Analysis | Gait analysis, Posture Assessment | Computer Vision, Deep Learning | Low – few small studies, mostly proof-of-concept |
| Personalized Exercise Prescription | Conventional and Post-Operative Rehab | Machine Learning, Reinforcement Learning | Low – limited RCTs, short follow-up |
| Clinical Decision Support | Risk Stratification, Red Flags | Predictive Analytics, Natural Language Processing | Low – mostly feasibility studies |
| Robotics & Neurorehab | Neuro Rehabilitation | Adaptive Control Algorithms | Low – small sample sizes, early-stage trials |
| Adherence Tracking/ Telerehab | Long-Term Rehab Programs | Behavioral Prediction Models | Low – preliminary evidence, short-term outcomes |
It is often claimed that AI systems analyze movement and patient data more quickly and objectively than humans.
Objectivity does not mean rigid rule-following; it means applying consistent criteria based on valid reasoning. If a clinician evaluates patients using the same justified criteria and adjusts interpretations when clinically appropriate, the assessment remains objective. Human clinicians are capable of interpreting context, weighing competing hypotheses, and integrating biopsychosocial factors into their reasoning.
While a computer follows programmed rules strictly and does not engage in human reasoning, strict rule adherence alone does not equal objectivity. Algorithms are limited by several factors, including the quality and structure of input data, model assumptions, and training datasets. For example, accurate biomechanical movement analysis typically requires appropriate measurement technology (e.g., 3D motion capture systems). A standard 2D camera does not provide true three-dimensional kinematic data.
AI can process large amounts of measurement data quickly (e.g., multiple video frames or sensor readings), but this is not the same as increasing clinical reasoning speed.
There is limited research on the effectiveness of AI-assisted
rehabilitation treatment. The usability of technological tools designed to address problems in healthcare should be assessed through standardized methodologies, with
careful consideration of regional, linguistic, and interdisciplinary variation within specific contexts [See Kreienbrinck, A., et al. 2025 and Sumner, J. et al. 2023].