• Artificial intelligence (AI)
• Cognitive rehabilitation
• Cognitive training
• Digital health
• Human–computer interaction (HCI)
• Neurodegenerative diseases
• Neuropsychology

The present thesis describes the design and formative evaluation of Rehabitus, a multilingual, AI-assisted cognitive assessment and stimulation tool for neurodegenerative diseases. The approach used is significantly motivated by the limitations of static, one-size-fits-all interventions, and the project itself represents a human-centered, clinician-in-the-loop design process, where clinicians are still embedded decision-makers and algorithms support interventions to personalization. Rehabitus has modules for memory, attention, executive function, language, and visuospatial skills; with an adaptive controller of task difficulty changing according to moment-to-moment performance and clinician dashboards that identify trends to personalize each Human-Computer Interaction (HCI). A working prototype was specified in Figma and implemented in NextJS; a pilot at AOUP (Pisa) is designed as the future project which will examine patients clinicians and caretakers. Subsequently, a mixed-methods evaluation (task analytics, usability questionnaire, and brief interviews) will explore elements of usability, acceptability, perceived clinical utility, and feasibility of embedding AI into neuropsychological workflows. The work documents design constraints related to older adults (multilingual UI, voice-over options, low-visual-load layouts) and outlines ethics, informed consent, and GDPR-enabled compliance. The contributions are twofold: (i) a working prototype that operationalizes personalized cognitive support with oversight from clinicians, and (ii) initial evidence and design guidelines around where AI can responsibly augment neuropsychological practice for neurodegenerative conditions.