NLP-based Chatbot Development for Suicide Prevention

The internship will take place at LaTIM (Inserm UMR 1101), a multidisciplinary research laboratory in Brest, France, at the interface between medical imaging, clinical AI, and digital health innovation . LaTIM is a Joint Research Unit (UMR 1101) of Inserm (the French National Institute of Health and Medical Research), the IMT Atlantique Engineering School and the University of Western Brittany located in Brest University Hospital.

Context

Suicidal behavior is a major public health concern and remains one of the leading causes of death among young adults worldwide [1]. In France, suicide prevention strategies rely heavily on Brief Contact Interventions (BCIs), which maintain structured, long-term contact with individuals after a suicide attempt. In the Brittany region, the VigilanS program has implemented this approach since 2016, providing systematic follow-up telephone calls by trained nurses to monitor patients’ well-being and encourage re-engagement with care services [2].

While BCIs are effective, their deployment faces structural limitations:

• follow-up calls are time-consuming,
• structured assessments cannot be repeated frequently,
• large-scale epidemiological screening is impractical with manual processes.

The ACCORDIA project, led by Prof. Sofian Berrouiguet, addresses this challenge by exploring the use of conversational agents to support outpatient monitoring, automate non-urgent clinical assessments, and improve population-level screening [3, 4]. In this context, NLP-based chatbots appear as a promising solution to collect standardized patient-reported information, administer validated questionnaires (such as the PHQ-9 for depressive symptoms), and assist health professionals by streamlining routine, non-decisional tasks.

Objective

Design and prototype an NLP-based chatbot capable of:

• administering structured self-assessment questionnaires (e.g., PHQ-9),
• collecting standardized patient-reported information,
• interacting safely and predictably within a controlled conversational framework,
• supporting healthcare teams by automating low-risk, repetitive tasks.

The chatbot will operate within clear clinical boundaries and will not perform any autonomous diagnostic or urgent-risk assessment.

Work description

The intern will:

• review existing conversational agents in mental health and clinical screening;
• define safe and structured conversation flows;
• develop the NLP components required for robust interaction (e.g., intent handling, controlled generation, dialogue management);
• implement a functional prototype;
• evaluate usability, acceptability, and robustness through simulated interactions.

The specific technical approach is intentionally left open, and candidates are invited to justify their choices in their cover letter.

References

[1] https://www.who.int/health-topics/suicide

[2] https://sante.gouv.fr/prevention-en-sante/sante-mentale/la-prevention-du-suicide/article/le-dispositif-de-recontact-vigilans

[3] Li H, Zhang R, Lee YC, Kraut RE, Mohr DC. Systematic review and meta-analysis of AI-based conversational agents for promoting mental health and well-being. npj Digit Med. 2023;6(1):236. doi:10.1038/s41746-023-00979-5

[4] Vaidyam AN, Wisniewski H, Halamka JD, Kashavan MS, Torous JB. Chatbots and Conversational Agents in Mental Health: A Review of the Psychiatric Landscape. Can J Psychiatry. 2019;64(7):456-464. doi:10.1177/0706743719828977

• Background in AI / NLP
• Strong Python programming skills
• Experience with deep learning frameworks is a plus
• Interest in digital health and ethical design of conversational systems