A Multimodal Deep Learning Approach for Predicting Dementia’s Disease Progression Using MRI Clinical Data

Dementia’s disease (DD) is a progressive neurological disease that primarily affects the elderly. Its primary symptom is cognitive impairment. There is currently no known cure; the only available treatments are meant to improve the patient’s quality of life. Given the challenges in treating DD, predicting its trajectory can be crucial in guiding management and treatment strategies. In this work, we propose a multimodal approach that uses MRI scans and clinical data to predict the phases of DD. The VGG19 deep learning model is utilized to interpret MRI images for image classification, and LightGBM (LGBMClassifier) is used to evaluate patient clinical data in CSV format. A multimodal deep learning technique is then used to combine the predictions from both modalities in order to increase the overall prediction accuracy. The integrated model aims to provide a comprehensive and accurate assessment of DD’s progression. This multimodal method offers a promising step toward a better understanding and management of Dementia’s disease by providing doctors with improved prediction insights into the development of the disease.