ProCoG

BACKGROUND

Global population aging is placing growing pressure on healthcare systems and economies. By 2050, the proportion of individuals aged over 60 is projected to reach 22%, accompanied by a significant rise in age-related pathologies, particularly neurodegenerative diseases such as Alzheimer's disease (AD). This condition represents a major public health challenge, with an estimated cost of over 1% of global GDP. In this context, it is essential to develop personalized prevention and monitoring strategies, as well as earlier diagnosis to enable management tailored to each patient's specific trajectory.

AD is characterized by progressive cognitive decline, the evolution of which varies greatly from one individual to another. This variability makes personalized care planning difficult and poses new challenges for administering treatments currently in development. The ProCoG project addresses this by drawing on the MEMENTO database, a longitudinal cohort of 2,300 initially healthy individuals presenting with cognitive complaints. These participants are followed every six months for two years, undergoing clinical, neuropsychological, biological, and brain imaging evaluations, as well as questionnaires on their lifestyle and social environment.

The project's objective is to predict the evolution of cognitive status using data from neuropsychological assessments. Ultimately, patients are classified into three profiles: those who remain cognitively healthy, those who develop mild cognitive impairment, and those who progress to major cognitive impairment characteristic of AD. The analyses will seek to better characterize these cognitive trajectories using machine learning methods, applied not only to isolated scores but also to multidimensional cognitive profiles. This integrated approach, which remains relatively unexplored, offers increased potential for individualized prediction. It builds on solid methodological precedents within the team, particularly in graph analysis applied to cognitive data, allowing researchers to understand the dynamic relationships between different functions such as memory, language, or executive functions.

---

STUDENT CONTRIBUTION

The student recruited for this Master 2 research internship will play a central role in conducting the project. They will begin with an in-depth literature review focusing on both Alzheimer's disease and predictive analysis approaches applied to longitudinal data. They will then become familiar with the MEMENTO database, particularly the cognitive scores extracted from successive neuropsychological assessments. They will actively participate in the analyses, which will employ advanced machine learning methods to model the evolution of cognitive status over a 24-month period.

Concurrently, the student may also become involved in complementary analyses based on graph theory to explore the interactions between different cognitive dimensions. These analyses will help identify typical cognitive profiles associated with different trajectories and better understand the tipping points between a healthy cognitive state and a pathological one. This contribution is part of a multidisciplinary approach combining cognitive neurosciences, neuropsychology, and data science, providing the student with a rich experience on a research project with high societal and clinical stakes.