Objective: To present the protocol of the Artificial Intelligence-based Parkinson’s Risk Assessment (AI-PRA) study which aims to validate a novel personalised PD risk prediction model in individuals at higher risk of PD.

Background: An early, accurate PD diagnosis is a research priority. MDS diagnostic criteria for prodromal PD provide a research diagnostic tool although with suboptimal performance. The AI-PROGNOSIS is a research project (funded by the European Union under Grant Agreement No. 101080581) that aims to advance PD diagnosis and care through novel predictive artificial intelligence-based models integrating health, genetic and digital wearable data.

Method: This proof-of-concept, prospective multicentre study will recruit 60 participants (aged ≥50 years, across three European countries) at higher risk of PD defined as polysomnography-confirmed isolated REM sleep behaviour disorder, neurogenic orthostatic hypotension, or objective idiopathic hyposmia documented on smell test, followed-up for 12 months.

The AI-PROGNOSIS model developed in first phase of the project will use digital biomarker data on motor function, physical activity and sleep (tracked continuously via a smartwatch), clinical information logged by health care professionals and participant-reported information on PD risk factors (collected through dedicated apps – mAI-Insights and mAI-Health, respectively) to estimate PD risk.   

Participants will undergo in-person clinical assessments at baseline and end of the study for completion of questionnaires on a range of PD symptoms, neurological examination and blood sampling.

Results: Validity measures (sensitivity, specificity, positive / negative predictive value, balanced accuracy and area under the curve) will be compared between the AI-PROGNOSIS PD risk model and the MDS diagnostic criteria for prodromal PD, using binary (normal / abnormal) and quantitative dopamine transporter SPECT imaging as the ground truth for PD risk.   

Feasibility and usability data will be collected: descriptive smartwatch use and app user engagement data, and qualitative usability and satisfaction questionnaires.

Conclusion: Based on unintrusive everyday digital biomarkers and routine clinical information, the AI-PRA study will provide prospective proof-of-concept validation of a novel individualised PD risk model aiming to enable scalable tools to improve early accurate PD diagnosis.

References: Schrag A, Horsfall L, Walters K, Noyce A, Petersen I. Prediagnostic presentations of Parkinson’s disease in primary care: a case-control study. Lancet neurology 2015;14:57-64.
2. Simonet C, Bestwick J, Jitlal M, et al. Assessment of Risk Factors and Early Presentations of Parkinson Disease in Primary Care in a Diverse UK Population. JAMA neurology 2022;79:359-369.
3. Berg D, Postuma RB, Adler CH, et al. MDS research criteria for prodromal Parkinson’s disease. Movement disorders : official journal of the Movement Disorder Society 2015;30:1600-1611.
4. Heinzel S, Berg D, Gasser T, Chen H, Yao C, Postuma RB. Update of the MDS research criteria for prodromal Parkinson’s disease. Movement disorders : official journal of the Movement Disorder Society 2019;34:1464-1470.
5. Noyce AJ, R’Bibo L, Peress L, et al. PREDICT-PD: An online approach to prospectively identify risk indicators of Parkinson’s disease. Movement disorders : official journal of the Movement Disorder Society 2017;32:219-226.
6. Bestwick JP, Auger SD, Simonet C, et al. Improving estimation of Parkinson’s disease risk-the enhanced PREDICT-PD algorithm. NPJ Parkinsons Dis 2021;7:33.
7. Kaasinen V, Vahlberg T. Striatal dopamine in Parkinson disease: A meta-analysis of imaging studies. Annals of neurology 2017;82:873-882.
8. Cummings JL, Henchcliffe C, Schaier S, Simuni T, Waxman A, Kemp P. The role of dopaminergic imaging in patients with symptoms of dopaminergic system neurodegeneration. Brain : a journal of neurology 2011;134:3146-3166.
9. Hastings A, Cullinane P, Wrigley S, et al. Neuropathologic Validation and Diagnostic Accuracy of Presynaptic Dopaminergic Imaging in the Diagnosis of Parkinsonism. Neurology 2024;102:e209453.
10. Mahlknecht P, Leiter S, Horlings C, et al. Preferences regarding Disclosure of Risk for Parkinson’s Disease in a Population-based Study. Mov Disord Clin Pract 2025;12:203-209.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/an-artificial-intelligence-based-model-to-predict-parkinsons-disease-risk-the-ai-pra-study-protocol/