JMMLRC

The software tool Joint Multi-Modal Longitudinal Regression Classification (JMMLRC) is a novel computational approach designed to study the biological mechanisms underlying Alzheimer's disease (AD). By combining clinical data from various modalities, such as genetic information and brain scans, JMMLRC performs regression and classification tasks to identify AD-relevant biomarkers and predict patients' cognitive scores and clinical diagnoses.

The method utilizes a variety of regularizations to intelligently combine the multi-modal data and address the challenging problem of determining the biological mechanisms causing AD development. As the proposed objective is a non-smooth optimization problem, the authors have derived an efficient iterative algorithm and proved its convergence.

Comprehensive experiments on the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort have validated the effectiveness of JMMLRC in predicting patients' cognitive scores and clinical diagnoses. The promising results demonstrate the benefits and flexibility of the proposed method, which may be of interest to clinical communities beyond AD research.