Cariplo

METHODS

1. Data Harmonization and Quality Control

•Inter-lab data normalization (e.g., Reference Range Normalization)

•Automated derivation of reference ranges when missing

•Missing data handling using Multiple Imputation and Maximum Likelihood methods

2. AI for Recruitment and Eligibility

•Deep learning models to classify inclusion/exclusion criteria

•Automated eligibility assessment

•CNN-based prediction models to identify patients likely to enroll in trials

3. Generative Models and In-Silico Trials

•GAN-based synthetic data generation

•Creation of digital “twin” trial populations

•Data-driven in-silico clinical trial simulation

4. Representativeness Metrics

•Statistical indicators (e.g., homogeneity metrics, Gini index)

•Quantitative evaluation of minority inclusion improvements

Impact

•Scientific Impact

•Advances data science methodologies in clinical trial design

•Establishes robust pipelines for normalization and missing data handling

•Introduces scalable in-silico trial validation frameworks

•Societal Impact

•Reduces health disparities through improved trial inclusivity

•Enhances equity in access to evidence-based treatments

•Supports public health policy with data-driven generalizability metrics

•Economic Impact

•Improves efficiency and cost-effectiveness of clinical trials

•Enables scalable and commercially viable digital tools for healthcare innovation

PEOPLE