Postural abnormalities are highly disabling complications of Parkinson's disease (PD). These include camptocormia, anterocollis, and Pisa Syndrome (PS). PS is characterized by a lateral trunk flexion (LTF) typically exceeding 10 degrees, often accompanied by axial rotation, asymmetric shoulder positioning, and poor awareness of the postural alteration. This condition worsens during upright activities and improves in a supine position. Patients with PD and PS are characterized by more pronounced motor asymmetry, a disorganized trunk muscle activity, back pain, balance issues, and reduced quality of life compared to PD patients without postural disorders. Camptocormia, another disabling postural anomaly, involves an anterior trunk flexion that also improves when lying down. Both PS and camptocormia are challenging to treat, with limited and short-lasting benefits from current multidisciplinary approaches, including medication, physiotherapy, botulinum toxin injections, and transcranial direct current stimulation (tDCS). Given the limitations of traditional rehabilitation strategies, there is a growing need for innovative and personalized approaches. In this context, advanced technologies such as artificial intelligence (AI) offer new possibilities for home-based treatment. This study aims to evaluate the feasibility of using a real-time visual feedback system powered by AI as a complementary intervention following inpatient neurorehabilitation for PD patients with trunk postural disorders (PS or camptocormia). A secondary objective is to assess whether an AI-guided, personalized exercise program can help maintain improvements in posture, mobility, and quality of life in the medium term. By integrating quantitative and qualitative outcomes, this study seeks to fill a gap in the literature and explore the potential of AI-driven home rehabilitation to support long-term functional gains and foster greater independence and well-being in people with PD.