Artificial Intelligence Predicts GBA1 Mutated Status in Parkinson’s Disease Patients

GBA1 mutations represent a major genetic risk factor for Parkinson’s Disease (PD), contributing to a significant proportion of cases depending on population characteristics and age at onset. This study investigated whether Artificial Intelligence (AI) could be harnessed to predict GBA1 mutation status in PD patients, aiming to develop a machine learning model capable of providing an accurate pre-test estimate based on key clinical and demographic indicators. To this end, a cohort of 58 GBA1-PD patients was matched with 58 non-mutated PD patients, and 124 features were recorded for each participant. Using a Leave-One-Out cross-validation approach and SHapley Additive exPlanations (SHAP) to quantify feature contributions, XGBoost emerged as the most effective algorithm for this supervised classification task. The resulting model relied on five primary clinical variables—including family history, cognitive scores (MDS-UPDRS 1.1), and motor function measures (MDS-UPDRS 3.8a, 3.8b, and rigidity subscores)—and achieved 73% overall accuracy, rising to 94% in patients with high SHAP confidence. These findings demonstrate the potential of AI to support targeted genetic screening in PD, particularly in settings with limited clinical resources. Limitations include the relatively small sample size and absence of external validation, highlighting the need for further research in larger, independent cohorts to refine and validate the predictive model.

Artificial Intelligence Predicts GBA1 Mutated Status in Parkinson’s Disease Patients

Autors: Giulia Di Rauso MD, Alessandro Ghibellini MSc, Sara Grisanti PhD, Valentina Fioravanti MD PhD, Edoardo Monfrini MD PhD, Giulia Toschi PhD, Giacomo Portaro MD, Giacomo Argenziano MD, Ruggero Bacchin MD, Jessica Rossi MD, Rossella Sabadini MD, Valeria Ferrari MD, Andrea Melpignano MD, Francesca Pacillo MD, Maria Scarano MD, Anna Groppi MD, Luca Gherardini MSc, Chiara Lucchi PhD, Giuseppe Biagini MD, PhD, Sara Montepietra MD, Maria Chiara Malaguti MD PhD, Isabella Campanini PhD, Andrea Merlo PhD, Andrea Castellucci MD, Angelo Ghidini MD, Alessandro Fraternali MD, Annibale Versari MD, Augusto Scaglioni MD, Jefri J. Paul PhD, Luciano Bononi Eng, Maurizio Gabbrielli Eng, Alessio Di Fonzo MD PhD, Peter Bauer MD, Francesco Cavallieri MD PhD, Franco Valzania MD

DOI: 10.1002/mdc3.70334