PsycAssist: digital innovation in adaptive and personalized assessment of executive functions and fluid intelligence

Neuropsychological assessment has undergone several transformations, moving from traditional paper-and-pencil tests toward digital, adaptive, and personalized approaches. This dissertation is situated within this transition and contributes to the development of innovative tools for computerized and adaptive assessment of executive functions and fluid intelligence. Although classical instruments such as the Tower of London and Raven’s Progressive Matrices are methodologically robust and clinically valid, they present important limitations, including long and inflexible administration or a lack of adaptation to individual ability levels. These issues are particularly critical for clinical populations, in which fatigue, attentional difficulties, and cognitive vulnerabilities may compromise reliability and validity.This work is part of the PRIN2020 project “Computerized, Adaptive and Personalized Assessment of Executive Functions and Fluid Intelligence” where the main objectives were to develop a new generation of adaptive tests for planning and fluid intelligence, design an intelligent web-based system for administration, data analysis, and personalized reporting, and apply innovative methodological frameworks, specifically Knowledge Space Theory and Procedural Knowledge Space Theory.These efforts led to the development of PsycAssist, an AI-based web platform for adaptive neuropsychological assessment with automated scoring and clinical report generation. Two main instruments were implemented: Adap-ToL, an adaptive version of the Tower of London for planning abilities, and MatriKS, a computerized adaptive version of Raven’s Matrices for fluid intelligence.The dissertation is organized into two parts. Part1 presents the theoretical and methodological background, including the evolution of neuropsychological assessment, executive functions and planning, fluid intelligence, and their relevance in clinical populations, with a specific focus on Specific Learning Disorders. Part2 reports five empirical studies addressing the development, validation, usability, and process-based modeling of PsycAssist, Adap-ToL, and MatriKS. This work supports a reconceptualization of neuropsychological assessment as a dynamic, adaptive, and user-centered process that integrates psychometric innovation and digital technology to enhance both clinical practice and research.