Integrating AI in Postural Rehabilitation for Parkinson's Disease

Postural abnormalities represent particularly disabling complications of Parkinson's disease (PD), significantly affecting patients' functional independence and quality of life. Among the most commonly observed postural deformities there are camptocormia, anterocollis, and Pisa Syndrome (PS). Pisa Syndrome is clinically characterized by a lateral trunk flexion (LTF) greater than 10 degrees, typically presenting with a uniform inclination from the sacrum to the C7 vertebra. It is often accompanied by axial rotation, leading to elevation and forward displacement of the contralateral shoulder. Additional clinical features include difficulty rotating the trunk away from the leaning side, worsening of symptoms during upright posture or walking, improvement in the supine position, and poor patient awareness of the postural deviation.

Patients with PD and PS exhibit greater motor symptom asymmetry, disorganized paraspinal muscle activation, back pain, balance and postural control issues, and reduced quality of life compared to PD patients without postural abnormalities. The current standard of care for these postural disorders typically includes a multidisciplinary approach involving pharmacological treatment, physical therapy, botulinum toxin injections, and transcranial direct current stimulation (tDCS). However, these interventions frequently yield only modest and short-lived improvements, highlighting the need for more sustainable and accessible treatment strategies.

Camptocormia is another severe postural abnormality seen in PD, defined as an involuntary forward flexion of the trunk that is mostly apparent during standing and walking, and tends to reduce when the patient is supine. Like PS, camptocormia is frequently associated with chronic back pain and is often resistant to dopaminergic medication, with many patients reporting no significant improvement in either "on" or "off" states.

The long-term management of these postural disorders remains a significant clinical challenge. Innovative, individualized rehabilitation strategies that can be seamlessly integrated into patients' daily routines are urgently needed. In recent years, technological advancements-particularly in the field of artificial intelligence (AI)-have opened up new avenues for delivering personalized home-based interventions.

This study investigates the feasibility and potential benefits of using a real-time, AI-based visual feedback system as a complementary treatment following hospital-based neurorehabilitation for PD patients with trunk postural disorders, specifically PS and camptocormia. The intervention uses the Kemtai platform, which analyzes movement through video-based AI to deliver customized exercise routines and real-time corrective feedback. This allows patients to engage in highly interactive and adaptive rehabilitation sessions from home, potentially improving adherence and functional outcomes.

The primary objective of the study is to assess the feasibility, safety, and usability of this AI-based system as a continuation of care after inpatient rehabilitation. The secondary objective is to evaluate whether this personalized exercise program can prolong, over the medium term, the gains in posture, mobility, and quality of life achieved during the structured hospital-based rehabilitation period.

By collecting both qualitative feedback (e.g., user satisfaction, perceived usefulness) and quantitative metrics (e.g., trunk inclination angles, mobility scores by means of cinemtaic analysis of movement), this study aims to contribute new evidence on the role of intelligent digital tools in managing postural abnormalities in PD. The overarching hypothesis is that a home-based, AI-monitored exercise program could offer a sustainable, effective, accessible, and patient-centered approach to maintaining functional gains and supporting greater autonomy in everyday life.