Work

D3.2 / First report on predictive modelling for PD

Year: 2025

Type: article

Abstract: The deliverable D3.2 provides an overview of all methodologies, findings and future directions of the AI-PROGNOSIS predictive models for Parkinson’s disease (PD). The deliverable focuses on three main pillars of prognostic modelling: a) estimation of the risk of developing PD, b) estimation of progression in PD, and c) estimation of response to medication. Hence, three distinct modelling approaches are presented, following the TRIPOD+AI reporting guidelines. The models use multimodal data, including sociodemographic factors, wearable device data and genetic profiles from wealthy, open large-scale datasets, such as the PPMI and AMP-PD. Pre-modelling achievements of the project include the successful harmonization of the core dataset using the observational medical outcomes partnership (OMOP) common data model, as well as efficient data pre-processing, such as bias mitigation, outlier detection, data imputation to improve model robustness, generalizability and fairness. Challenges were faced and addressed to ensure seamless data harmonisation, paving the way for harmonising datasets to be accessed in the future.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.15542464

Author: Sotirakis, Harris

Language: English

Topic: Parkinson's Disease Mechanisms and Treatments

Subfield: Neurology

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: green

Funder: European Commission

Award: Artificial intelligence-based Parkinson’s disease risk assessment and prognosis