I'm a PhD student in Computer Science at the University of Neuchâtel and HE-Arc, supervised by Prof. Dr. Paul Cotofrei (University of Neuchâtel) and Prof. Dr. Hatem Ghorbel (HE-Arc).
My research sits at the intersection of large language models (LLMs) and neurodegenerative disease research. Through my thesis, CogniLens, I investigate whether language models can be used to simulate the functional cognitive effects of Alzheimer's disease (AD) and to support its early detection and monitoring.
CogniLens is a framework for systematically altering language models to simulate Alzheimer's-related cognitive decline, evaluated through both behavioral and neural alignment metrics:
- Cognitive assessment — an adaptation of the Montreal Cognitive Assessment (MoCA) into structured text-based prompts with pairwise comparison metrics, enabling continuous quantification of cognitive performance across attention, memory, language, and abstraction.
- Model alteration — three strategies to simulate cognitive decline: baseline neuron ablation, transcript-based fine-tuning on AD speech data, and neurobiologically informed targeted layer alterations grounded in brain region-to-model-layer mappings.
- Neural alignment — brain score, a measure of alignment between model internal representations and human fMRI recordings, computed at each stage of alteration to evaluate the neural plausibility of the simulated decline.
- LLM evaluation & cognitive assessment
- Alzheimer's disease detection from linguistic data
- Model interpretability & neural alignment (brain score)
- Digital twins of cognitive decline
- Brain–language mapping (fMRI ↔ model representations)
- COGNILENS: Analyzing Cognitive Decline in Language Models for Alzheimer's Monitoring — EACL 2026 Student Research Workshop (ACL Anthology)
- RAG: Unveiling the Power of Retrieval-Augmented Generation — SwissText 2024 (ACL Anthology)
- Creating a virtual receptionist to lighten your employees' workload: a case study with the canton of Jura and subsidies for training — SwissText 2024 (ACL Anthology)