Stage, 30 rue de Campo Formio 75013 PARIS.| Entreprise/Organisme : | Brain&Mind | | Niveau d'études : | Master | | Sujet : | In clinical trials, a panel of biomarkers is often investigated in combination with few clinical covariates & baseline characteristics to discover (see BEST guidance for more information):
• predictive signatures: to identify responders/non-responders of a drug
• prognostic signatures: to identify fast/slow progressors of a disease
• diagnostic signatures: to distinct a condition from another
The main objective of this internship is to provide a tool to the biostatistician community to accelerate biomarker signature discovery. The idea is to implement a framework within which as many steps as possible are automated with optional settings, while providing tailored outputs.
In particular, it will include:
- a data science pipeline: sequence of elements including resampling, reformating, dimension reduction, implementation of various ML/DL algorithms and prediction for a new patient.
- a HTLM report with key outputs (e.g., performance, algo choice, interpretability of results) | | Date de début : | mars-avril 2026 (négociable) | | Durée du contrat : | 6 mois | | Secteur d'activité : | Médecine/Pharma/Santé | | Description : | Brain&Mind is a French non-profit association, created by l'Institut du Cerveau de Paris, la Fondation Voir&Entendre and la Fondation FondaMental, whose objective is to accelerate innovation in Neuroscience (Neurology, Psychiatry, Sensory Disorders).
The production launch of a new proteomics platform dedicated to Neuroscience is planned by Q4 2026. The technology and the panel of neurological biomarkers (proteomics here) were deeply discussed with various experts, ensuring this platform to be highly sensitive to a large set of biomarkers (proteins here) identified as major in Neuroscience.
In practice, as already performed in clinical trials, a panel of biomarkers is investigated in combination with few clinical covariates & baseline characteristics to discover (see BEST guidance for more information):
• predictive signatures: to identify responders/non-responders of a drug
• prognostic signatures: to identify fast/slow progressors of a disease
• diagnostic signatures: to distinct a condition from another
The main objective of this internship is to provide a tool to the biostatistician community to accelerate biomarker signature discovery. All the definitions above have some commonalities and specificities. The idea is to implement a framework within which as many steps as possible are automated with optional settings, while providing tailored outputs.
Main deliverables:
• Flexible, user-friendly & automated pipeline with the following steps:
o Merging of clinical & biomarker datasets
o Data science pipeline: sequence of elements including resampling, reformating, dimension reduction, implementation of various ML/DL algorithms and prediction for a new patient.
• Associated HTML reporting with:
o Summary of potential QC steps
o Summary of performance metrics for several algorithms
o Best combinations for biomarker signature candidates
o Interpretability (at local and/or global level)
o Descriptive statistics & involved pathways
Qualifications :
• M2 or engineering schools (e.g. ENSAI, ENSAE, ISUP, Centrale-Supelec, …) specialized in Applied Mathematics, Statistics or Data Science
• A previous experience with Python: scikit-learn, lime, PyTorch or TensorFlow, …
• A previous experience with Git and Github or GitLab
• Fluent English
• Team spirit, curious, rigorous, autonomous
Optional knowledge considered a plus:
• R packages: mlr3, caret, tidymodels, DALEX, …
• Cloud Data Processing
• Bioinformatics
• Object-oriented programming | | En savoir plus : | https://brainandmind.sharepoint.com Internship_Brain&Mind_Biostatistics_DataScience.pdf | | Contact : | hr@brainandmind.fr |
|