Use cases

Use Case 1: Neurodegenerative Diseases Phenotyping & Prognosis Evaluation

  • Diseases: Focus on ALS with potential application to other neurodegenerative diseases.
  • Data Modalities and Centers: Genetic, clinical, imaging, and laboratory data from UNITO, UNIPD, and public/open data.
  • Scientific Approach: Develop methods using horizontal federated learning and knowledge extraction to perform unsupervised learning from multimodal data across centers. Identify correlations and dependencies among genetic, clinical, imaging, and environmental variables.
  • Example Clinical Outcome: Cluster ALS patients into subgroups with specific genetic variants and phenotypes. Identify environmental risk factors and improve patient stratification in clinical trials.

Use Case 2: Next-Generation Diagnosis and Treatment Response for Neurodegenerative Diseases

  • Group of Diseases: Application to MS, FTD, and other neurodegenerative diseases.
  • Data Modalities and Centers: Clinical, imaging, genetic, and environmental data from UNITO, UNIPD, CRG, and public/open data.
  • Scientific Approach: Develop ontologies to integrate clinical diagnosis with genetic variant functions. Validate the approach by investigating similarities between diseases like ALS and FTD. Apply unsupervised learning to MS data to identify patient clusters and optimize treatment strategies.
  • Example Clinical Outcome: Identify MS patient clusters and best-fitting treatments. Improve management and quality of life by linking environmental exposures to disease progression.

Use Case 3: Signs of Parkinson’s Disease in Multimodal Data

  • Group of Diseases: Parkinson’s and related diseases.
  • Data Modalities and Centers: Imaging, fundus photographs, OCT, clinical notes from UCD and UNIPD, clinical scales, neurophysiology (EEG), brain imaging (MRI/PET/SPECT), biofluids, genetic screening.
  • Scientific Approach: Identify PD patients using ophthalmic imaging. Build classifiers using deep learning, feature extraction, and foundational models. Explore associations between eye and brain biomarkers, and apply unsupervised learning to identify patient subgroups.
  • Example Clinical Outcome: Identify PD biomarkers in the eye, predict PD development, and explore associations between multimodal biomarkers.

Use Case 4: Phenotyping of the Gut-Brain Axis in Healthy Individuals

  • Group of Diseases: Diabetes, obesity, inflammatory bowel disease, irritable bowel syndrome.
  • Data Modalities and Centers: Genetic, text, microbiome, neuroimaging from RUMC, UNIPD, and public/open data.
  • Scientific Approach: Analyze gut microbiota from fecal samples and associate it with brain structure and function using a deeply phenotyped healthy population. Evaluate associations between gut microbiota, environmental factors, and health outcomes.
  • Example Clinical Outcome: Identify relationships between gut microbiome alterations and health-related data, potentially linking specific bacterial genera to brain functions and behaviors.

Use Case 5: Gut-Brain Linkage and Disease Relevance

  • Group of Diseases: Neurological, stress-related, neurodevelopmental disorders (PD, depression, ADHD, anxiety, autism).
  • Data Modalities and Centers: Genetic, text, microbiome, clinical, neuroimaging, and digital pathology samples from RUMC, UNIPD, UNITO, and public/open data.
  • Scientific Approach: Apply methods from Use Case 4 to clinical disorder samples. Characterize gut microbiota in relation to disease populations and clinical parameters using deep learning.
  • Example Clinical Outcome: Evaluate the relationship between gut microbiome alterations and clinical data in stress-related and neurodevelopmental disorders. Identify relevant probiotic treatments based on symptom clusters.