Santé publique France is France’s national public health agency. A public institution under the authority of the Minister of Health, formed through the merger of several public institutions and established by Ordinance 2016-246 of April 15, 2016, the agency works to promote public health. As a scientific, expert, and public health safety agency, its missions include:

1. Epidemiological observation and surveillance of the population’s health status; 
2. Monitoring health risks threatening the population;
3. Promoting health and reducing health risks;
4. Developing prevention and health education;
5. Preparation for and response to health threats, alerts, and crises;
6. Issuing public health alerts.

The agency is organized into 12 departments—some scientific, some cross-functional, and some providing operational support.

The agency’s strategic priorities and work program, established by its Board
of Directors, are organized into three areas: Strengthening the capacity for anticipation and rapid response to health threats; Measuring and assessing the extent of diseases and risk factors to guide their prevention and control; Strengthening the health impact of all public policies and the prevention and promotion of health.

Data Support, Processing, and Analysis Division — AMETIS Unit (Support and Methods for Studies and Investigations in the Field of Surveillance)

Internship Topic

The objective of this internship is to contribute to the implementation of a large-scale, automated pipeline for processing and structuring data from the SurSaUD system, tailored to the analysis of multivariate time series in the context of public health surveillance (detection of weak signals, modeling, forecasting). The intern will participate in ensuring the reliability, enriching, and organizing a massive dataset consisting of several hundred thousand time series, describing the evolution of health indicators across various geographic, temporal, syndromic, and demographic dimensions.

Internship Responsibilities

To structure and automate the data, the intern will be expected to participate in the following tasks:

• Data collection and structuring: leveraging the various sources within the SurSaUD system and their associated repositories to build a coherent and structured dataset.
• Creation of a reference dataset: participate in the development of a structured and documented corpus intended for the comparative evaluation of signal detection and forecasting methods.
• Building a time-series repository: organizing the data into a corpus structured across multiple dimensions (geographic, temporal, syndromic, and demographic), suitable for statistical and exploratory analyses.
• Contextual enrichment: integrate external data (sociodemographic characteristics, school calendar, holidays, weather conditions, etc.) to refine the analysis of the time series and guide future methodological choices (pre-selection of explanatory variables).
• Quality control: Implement procedures for evaluating data quality, including the detection and, if necessary, imputation of missing values, the identification of anomalies, and the definition of summary reliability indicators.
• Processing documentation: Produce comprehensive technical documentation (guides, scripts, dataset structuring) and implement practices that ensure the reproducibility and generalizability of the work to other data sources (versioning, script modularity).
• Performance optimization: Adapt processing workflows to Santé publique France’s computing infrastructure, taking into account constraints related to the large volumes of data being handled.
• Process automation: Design and implement an automated processing pipeline that ensures traceability, timestamping, and the reproducibility of processes.

For data analysis and utilization, the intern will be expected to participate in the following tasks:

• Extraction of temporal descriptors: compile indicators describing the behavior of time series (overall statistics, seasonality, trends, variability, etc.) to classify the observed dynamics.
• Analysis of correlations and temporal dependencies: studying the relationships between time series to identify significant interactions or dependencies among health indicators or geographic dimensions.
• Exploratory analysis and unsupervised classification: grouping time series according to their temporal profiles to identify recurring or atypical trends and better characterize the diversity of behaviors.