ML4-CD

Complex diseases suffer from the limitation that models based solely on traditional genomic data fail to capture full biological heterogeneity. Therefore, patient stratification would benefit from integrating multi-omics data using advanced machine learning approaches. In this project (focusing on complex conditions such as Alzheimer’s disease, its prodromal stage MCI, and Parkinson’s disease), we will apply unsupervised subtyping methods to multi-modal datasets to identify distinct disease subtypes. These subtypes will be integrated into predictive frameworks to improve case-control discrimination and risk stratification. In parallel, we will extend this approach to early-stage disease manifestations to evaluate progression trajectories, assess biological concordance across states, and use survival models to predict disease conversion and time-to-progression.