Explainable machine learning algorithm predicting working memory performance in Parkinson’s disease using task-fMRI

• Published in: Journal of neurology Vol. 272; no. 10; p. 692
• Main Authors: Yasuda, Eiji, Hattori, Takaaki, Shimano, Kaoru, Hase, Takeshi, Oyama, Jun, Yamagiwa, Ken, Kawauchi, Miho, Horovitz, Silvina G., Lungu, Codrin, Matsuzawa, Hitoshi, Hallett, Mark
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 14.10.2025; Springer Nature B.V
• Subjects: Activity patterns; Aged; Algorithms; Alzheimer's disease; Brain - diagnostic imaging; Brain - physiopathology; Brain research; Cognition & reasoning; Cognitive ability; Consent; Cortex (parietal); Data processing; Decision making; Deep learning; Female; Functional magnetic resonance imaging; Humans; Learning algorithms; Machine Learning; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; Medical imaging; Medicine; Medicine & Public Health; Memory; Memory, Short-Term - physiology; Middle Aged; Movement disorders; Neural networks; Neural Networks, Computer; Neurodegenerative diseases; Neuroimaging; Neurological disorders; Neurology; Neuroradiology; Neurosciences; Original Communication; Parkinson Disease - complications; Parkinson Disease - diagnostic imaging; Parkinson Disease - physiopathology; Parkinson Disease - psychology; Parkinson's disease; Prefrontal cortex; Scanners; Working memory; Task-based fMRI; Parkinson’s disease; 3D convolutional neural network; Explainable machine learning; 3D convolutional autoencoder
• ISSN: 0340-5354; 1432-1459; 1432-1459
• DOI: 10.1007/s00415-025-13438-w

Background Parkinson’s disease (PD) is a neurodegenerative disorder that affects both motor and cognitive functions, particularly working memory (WM). Machine learning offers an advantage for decoding complex brain activity patterns, but its application to task-based functional magnetic resonance imaging (task-based fMRI) has been limited. This study aimed to develop an explainable machine learning model to classify WM performance levels in PD based on task-based fMRI data. Methods We enrolled 45 patients with PD and 15 healthy controls (HCs), all of whom performed an n -back WM task in an MRI scanner. Patients were stratified into three subgroups based on their 3-back task performance: better, intermediate, and worse WM. A three-dimensional convolutional neural network (3D-CNN) model, pre-trained with a 3D convolutional autoencoder, was developed to perform binary classifications between group pairs. Results The model achieved an accuracy of 93.3% in discriminating task-based fMRI images of PD patients with worse WM from HCs, surpassing the mean accuracy of three expert radiologists (70.0%). Saliency maps identified brain regions influencing the model’s decisions, including the dorsolateral prefrontal cortex and superior/inferior parietal lobules. These regions were consistent with both the areas with intergroup differences in the task-based fMRI data and the anatomical areas that are crucial for better WM performance. Conclusions We developed an explainable deep learning model that is capable of classifying WM performance levels in PD using task-based fMRI. This approach may enhance the objective and interpretable assessment of brain function in clinical neuroimaging practice.