PRAMA

Proteomics, RAdiomics & Machine learning-integrated strategy for precision medicine for Alzheimer’s

Funded under: Par Fas Salute Toscana 2014-2020

Start date 22 October 2020     End date 24 April 2025

Keywords: Alzheimer's disease; Artificial Intelligence; Disease Phenotypes

Common clinical trials for Alzheimer's disease (AD) rely on outdated hypotheses on disease pathogenesis and on approximate criteria for patient selection, grouping together patients with diverse manifestations of the disease. Recent studies have suggested that AD may come with several clinical phenotypes and that the differentiation between disease subtypes can be due to the pathway followed by the AD precursor beta-amyloid (Aβ) peptide when it self-assembles into amyloid aggregates in the brain. An integrated survey taking advantage of multiple marker modalities is, thus, perceived as a desirable solution to support clinicians in identifying different disease subtypes, even in their early stages, and to accordingly decide on personalized treatments for individual patients.

In the PRAMA project, we intend to build up a strategy for personalized prediction of the disease based on the hypothesis that the main precursors of AD can form specific aggregates responsible for distinct clinical pictures of the disease, with consequent different sensitivity to drugs. In detail, a combined biochemical, biophysical and optical spectroscopy characterization of molecular biomarkers found in the cerebrospinal fluid of 100 individuals will be carried out, by including patients with progressive clinical signs of AD. This data will provide information on biomarker composition, structure, aggregation level and toxicity. This will constitute the proteomic profile of the biomarker content for each individual. The same patients will be subjected to magnetic resonance imaging (MRI) followed by a radiomics-based image analysis. The entire set of biochemical, optical, MRI data including clinical parameters and neuropsychological evaluation of patients will be elaborated through data analytics techniques to, firstly, discover correlations among novel and gold-standard biomarkers and, then, to mine and identify different AD phenotypes. The most recent Artificial Intelligence and Machine Learning techniques will be employed to model and process the complex high-dimensional data gathered in PRAMA. Data analyses will also aim at discovering specific diagnostic, prognostic or predictive responses at the different stages of disease stages, on a personalized basis.

The outcomes of PRAMA are expected to have a high socio-economic impact, with significant advantages that include reducing healthcare costs and improving the well-being of the ageing population.

The project is coordinated by IFAC-CNR and will last three years. SILab is involved in the analysis of the multimodal data to define the disease phenotypes.