Oportunidades Públicas

Multimodal integration in Alzheimer's Disease through Patient Similarity Networks

Patient Similarity Networks (PSNs) integrate heterogeneous data by turning each modality into a patient-by-patient similarity graph and fusing those graphs into one shared structure. The approach extends beyond omics to neuroimaging, clinical, and demographic data, yet multi-omics PSN work in Alzheimer's disease is still uncommon. In this project the student builds a PSN integration pipeline on real Alzheimer's data from ADNI, combining molecular and precomputed imaging features, and tests how the choice of distance metric, network fusion algorithm, and graph embedding shapes the patient subtypes it recovers. The goal is to find strata that single modalities miss, and to characterize them by diagnosis, APOE genotype, and cognition.