ADELAI
ADELAI collects longitudinal data from patients with dementia during routine neurological assessments. The collected data include clinical history, neurological examinations, neuropsychological test results, and patients’ written and spoken responses to structured questions. Digital interaction data, including keystroke press and release timing, are also acquired. These heterogeneous data are used to develop and evaluate AI models capable of identifying patterns associated with cognitive impairment and disease progression, with the long-term goal of enabling continuous, non-invasive monitoring approaches.
Technology Transfer
ADELAI
ADELAI collects longitudinal data from patients with dementia during routine neurological assessments. The collected data include clinical history, neurological examinations, neuropsychological test results, and patients’ written and spoken responses to structured questions. Digital interaction data, including keystroke press and release timing, are also acquired. These heterogeneous data are used to develop and evaluate AI models capable of identifying patterns associated with cognitive impairment and disease progression, with the long-term goal of enabling continuous, non-invasive monitoring approaches.
Overview
ADELAI is a longitudinal study aimed at collecting real-world clinical and behavioral data from patients with dementia. The study investigates the potential of artificial intelligence to support the diagnosis, prognosis, and longitudinal monitoring of cognitive decline through non-invasive and multimodal data collected during routine neurological visits.
Research Directions
- Development and validation of artificial intelligence methods for dementia, with a focus on multimodal learning, missing modalities, longitudinal disease modeling, early diagnosis, prognosis, and non-invasive patient monitoring. Research activities include the integration of clinical, neuropsychological, speech, and keystroke dynamics data, as well as the exploration of machine learning, deep learning, and foundation models.