Objective: To describe the co-design and staged evaluation of a novel AI-enabled digital health platform for Parkinson’s where structured patient and clinician feedback informs iterative refinement during clinical evaluation.

Background: In the UK, people with Parkinson’s typically spend fewer than 30 minutes per year with their specialist. Between visits, approximately 50% (Chaudhuri 2006) of symptoms go undetected, medication adherence averages 65% (Grosset 2009), and meaningful deterioration may take months to identify. Existing tools often rely on population-level thresholds, offer limited actionable recommendations, or lack sustained patient input beyond initial requirements.

Method: We developed a platform with five modules: (1) a mobile application with multi-device wearable integration for continuous passive motor, sleep and non-motor data capture; (2) an adaptive personalisation engine replacing population thresholds with patient-specific models; (3) an AI recommendation layer for personalised medication timing and lifestyle guidance; (4) a deterministic safety layer, architecturally independent from the AI, validating recommendations against drug interactions and clinical thresholds; and (5) clinician and care partner dashboards. Patient and clinician feedback is incorporated into development to refine monitoring priorities and functionality. Evaluation is staged: Stage 1 (Class I medical device) assesses feasibility of data capture, engagement and workflow integration; Stage 2 (Class IIa) will activate AI recommendation modules.

Results: Stage 1 evaluation is underway with over 250 participants. Deployment shows stable passive data acquisition across wearable integrations and sustained symptom monitoring engagement. Mean engagement is 58 minutes per participant, indicating active use beyond passive monitoring. Continuous data has identified symptom fluctuations and medication response patterns. Outcomes confirm reliable data capture, meaningful longitudinal engagement, and potential for integration into outpatient workflows.

Conclusion: Combining patient-directed development, adaptive personalisation, deterministic safety validation and staged evaluation, this platform addresses key limitations of existing digital Parkinson’s tools, offering a scalable approach to continuous personalised care between clinic visits.

Platform architecture

Staged regulatory evaluation pathway

Agentic AI feedback workflow

References: Chaudhuri, K.R., Healy, D.G. and Schapira, A.H., 2006. Non-motor symptoms of Parkinson’s disease: diagnosis and management. The Lancet Neurology, 5(3), pp.235-245.

Grosset, D., Antonini, A., Canesi, M., Pezzoli, G., Lees, A., Shaw, K., Cubo, E., Martinez‐Martin, P., Rascol, O., Negre‐Pages, L. and Senard, A., 2009. Adherence to antiparkinson medication in a multicenter European study. Movement disorders: Official journal of the Movement Disorder Society, 24(6), pp.826-832.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/a-patient-directed-ai-enabled-platform-for-continuous-personalised-parkinsons-management/