PHA for Mental Health


The SIS Lab will be overseeing the design of personal healthcare agents (PHA) in the context of therapeutic support for mental health interventions. PHA will be processing massive amounts of sensing, language, and interaction data to interpret his/her behavioral patterns while eliciting personal narratives related to emotion-triggering events or stressors. PHA will elaborate raw signals collected from multiple sensors and extract the knowledge to make orchestrated inferences. PHA will engage into coherent and sustainable conversations with users and facilitate the interaction with their therapist.

For managing change successfully it is important to manage individual stress in the workplace. This goal may be reached by means of individual counselling. Several studies supported the view that individual-directed interventions may be effective in promoting the acquisition of stress management skills and a healthy lifestyle. The key factor is providing support to the individual workers for increasing their adaptability to a changing environment.

Currently available conversational agents in the domain of mental well-being have limited interaction, understanding, and domain capabilities at best. Users’ language interpretation is bounded by predefined schemas and conversational flows. Our aim is to design a conversational agent capable of engaging users in more flexible, in-depth conversations to better support them through their daily life thus improving their mental well-being. This goal will be solely achieved by an appropriate integration of domain knowledge and data-driven algorithms for dialogue systems.

The mental state of a person correlates with her/his physiology. The usage of language and the way s/he speaks constitute other important behavioral signals that may evoke emotions. We aim to capture these signals through wearable devices and a smartphone application for providing real-time feedback on mental wellbeing. We believe that users may build a personal relationship with a personalized application and that this may help them reduce stress and anxiety.




Danieli M., Ciulli T, Mousavi M. and Riccardi G., “A Participatory Design of Conversational Artificial Intelligence Agents for Mental Healthcare”, JMIR Formative Research Journal, 2021 .

Mousavi M., Cervone A., Danieli M. and Riccardi G., “ Would you like to tell me more? Generating a corpus of psychotherapy dialogues ” NAACL, Workshop on NLP for Medical Conversations, 2021.

Tortoreto G., Stepanov E. A., Cervone A., Dubiel M., Riccardi G., “ Affective Behaviour Analysis of On-line User Interactions: Are On-line Support Groups more Therapeutic than Twitter? ” Association for Computational Linguistics Conference, Workshop on Social Media Mining for Health Applications, Florence, 2019.

Stepanov E. A., Lathuilierev S., Chowdhury S. A., Ghosh A., Vieriu R.D., Sebe N. and Riccardi G., “ Depression Severity Estimation from Multiple Modalities ” IEEE 20th International Conference on e-Health Networking, Applications and Services (Healthcom), Ostrava, 2018. EXCELLENT Paper AWARD

Ghosh A., Stepanov E. A., Danieli M., and Riccardi G., “Are You Stressed? Detecting High Stress from User Diaries”, Proc. IEEE International Conference on Cognitive Infocommunications, Debrecen, 2017.