Master’s Thesis Internship Offer
Explainable Machine Learning and Microbiome for Alzheimer’s Risk Stratification

Scientific Context
Recent scientific advances suggest a probable link between the gut microbiome and neurodegenerative diseases, including Alzheimer’s disease. Certain microbial signatures may influence mechanisms such as inflammation, intestinal permeability, or the microbiome–brain axis. This internship aims to analyze public datasets combining microbiome and cognitive variables to assess the feasibility of predicting Alzheimer’s status and identifying relevant microbial biomarkers. The project is conducted jointly by Sorbonne University (SCAI) and Université Laval (Quebec, Canada), with co-supervision by experts in microbiome research, data science, and neurobiology.

Data
The intern will work on cohorts containing taxonomic microbiome data (16S or shotgun sequencing) as well as clinical measurements related to cognition. Several data sources are being considered, including the PROTECT cohort (University of Exeter), the Knight ADRC (Alzheimer Disease Research Center, Washington University School of Medicine), and public resources such as ADNI-microbiome or the Qiita platform from the University of California San Diego (UCSD). Part of the internship will involve selecting, documenting, and harmonizing the final dataset.

Objectives
The main objective is to develop an artificial intelligence (AI) model capable of predicting Alzheimer’s status or the level of cognitive decline from microbiome data. The intern will identify the most informative taxa (biomarkers), compare different statistical and machine learning models, and use explainable AI approaches to interpret the results. The internship will also include the development of a reproducible pipeline and the preparation of a scientific synthesis (report or preprint of a scientific article).

Methodology
The work will begin with data preprocessing and normalization, followed by exploratory analyses of microbial diversity and association tests. The modeling phase will involve variable selection, cross-validation, and the comparison of several AI approaches. Explainable AI tools will be used to highlight potential biomarkers and interpret the models. The internship will conclude with a synthesis of the results aimed at scientific dissemination.

Required Skills
Master’s level (M2) in data science, bioinformatics, statistics, AI, or a related discipline.

Required Skills
Good command of Python or R (scikit-learn, tidyverse, phyloseq, QIIME2).
Background in microbiome research, biology, or multivariate statistics is appreciated.
Rigor, autonomy, and strong motivation for an interdisciplinary AI–health project.

Supervision
Elsa Rousseau & Frédéric Calon – Université Laval
Anna Bonnet & Sylvain Le Corff – Sorbonne Université

Location, Duration, and Organization
Duration: 6 months
Location:

First part: Sorbonne University – SCAI (Paris).
Second part: Possible at Université Laval (Quebec), depending on logistics and the intern’s preferences.