Psychoeducational mediators and auditable machine learning

Leadership of the scientific and methodological design of a decision support platform, within an R&D&I project with funding from the European Union Just Transition Fund.

The system integrates whole genome sequencing, psychoeducational measurement and machine learning models to identify students performing below their estimated biological and cognitive capacity.

A genetic variant enters the model only when three conditions converge: a known biological mechanism, replicated evidence and an applicable psychoeducational intervention. Biology is conceptualised as a set of modifiable learning barriers, in which the psychoeducational constructs act as latent mediators.

The engagement covers the study design, the statistical power calculation, two-phase sampling with a calibration subsample, the genomic and bioinformatic pipeline, auditable machine learning and the regulatory architecture: EU AI Act, ALTAI, GDPR and research ethics. It brings together personalised health, data science and Clinical processes and governed AI.