International Workshop on AI and Mental Health
- Organizers: Fani Deligianni and Alessandro Vinciarelli (University of Glasgow)
- Date: June 8th, 2021
- Time: 10.00-17.00 (UK time)
The World Health Organisation states that mental health problems affects more than 5% of the population worldwide, whereas in UK, mental health services report that 19.7% of people over 16 years old show symptoms of anxiety and depression. The latest interactive technologies of social robots and virtual reality, powered by Artificial Intelligence (AI) promise new approaches for clinical treatment of psychiatric, developmental and cognitive problems. This workshop brings together experts in AI, computing and mental health that work towards the development of new, computing based methodologies for detection, treatment and analysis of mental health issues. Besides providing an extensive overview of the state-of-the-art in the domain, the workshop aims at helping young researchers (including PhD students and postdocs) and raise awareness of multidisciplinary opportunities of computing and medical sciences in mental health.
Program
- 10.00 – 11.00. Esther Papies – The role of habits in health behaviour: Challenges and opportunities for digital interventions;
- 11.00 -12.00. Mohamed Chetouani – Putting the social interaction at the center of mental health analysis and modeling with AI-based approaches;
- 12.00 – 13.00: Pietro Cipresso – Computational Psychometrics integrating artificial intelligence, virtual reality and mental health.
- 13.00 – 14.00: Lunch Break.
- 14.00 – 15.00. Nicholas Cummins – Speech analysis for mental health: opportunities and challenges;
- 15.00 – 16.00. Helen Minnis – Social Signal Processing in Attachment;
- 16.00 – 17.00: Panel and Final Remarks.
Details about talks and speakers are available below.
The role of habits in health behaviour: Challenges and opportunities for digital interventions
Abstract Habits dominate much of human health behaviour. In this talk, I will briefly introduce habits from a psychological perspective, and address implications of this perspective for the development of digital interventions to improve health behaviours. In particular, habits develop if a behaviour is repeatedly performed to pursue a goal in a particular context. As a result, contextual cues can then trigger the activation and performance of a well-established habit, without relying on conscious awareness or intentions. This raises specific challenges and opportunities for situating just-in-time interventions to effect behaviour change. In addition, the development and maintenance of habits benefits from positive affect, which raises specific challenges for the content of interventions.
Bio Dr Papies directs the Healthy Cognition Lab at the University of Glasgow. She and her team study the cognitive processes underlying the regulation of health and consumer behaviour and behaviour change, especially in the domain of healthy and sustainable eating and drinking. Her research uses mainly social cognition methods and focuses on the cognitive processes of how behaviour is regulated as a function of cognitive representations shaped by personal goals, previous experiences, and environmental cues that trigger and shape these cognitive representations. The Healthy Cognition Lab addresses questions such as: how are appetitive stimuli represented cognitively, and how does desire for them develop? How can we leverage these processes to promote healthy and sustainable consumer behaviour? How can we design digital health behaviour interventions that optimally integrate personal motivation and environmental factors to promote healthy habit formation? Esther collaborates closely with key stakeholders for healthy and sustainable consumer behaviour, such as Danone and The World Resources Institute. Esther received her PhD in 2008 at Utrecht University, and after working at Utrecht University as Assistant and Associate Professor, she joined the University of Glasgow in 2015. Currently, her work is funded through various ESRC Research Grants and PhD studentships.
Putting the social interaction at the center of mental health analysis and modeling with AI-based approaches
Prof Mohamed Chetouani
Abstract Mental illnesses usually present a great variability of expression regarding both patients and stages of the pathology. Autism spectrum disorder (ASD) is among the most disabling neurodevelopmental disorders (NDDs) in children. It is characterized by impairments in social interaction, limitation in communication, and restricted and repetitive behaviors. The expression of these impairments presents a great variability, which leads to the concept of spectrum disorder. In such context, AI-based techniques have been successfully employed to characterize and quantify behavioral and physiological phenotypes, leading to the concept of digital phenotypes. However, there is clearly a need to develop approaches able to move from social perception to social interaction. Such approaches should be able not only to quantify individual behaviors of patients but also take into account the interaction with other partners.
In this talk, we will discuss such approaches and illustrate examples using social signal processing and social robotics methodologies. These approaches are built upon verbal, non-verbal, biological and physiological evidences of interpersonal emotional dynamics that have been reported in the literature. Within this context, the challenge is to develop machines that can decode social interaction by assessing individual and interpersonal dynamics of behaviors, with the goal of analyzing, quantifying and predicting human’s social behaviors. We will also show that the evaluation of such AI-based approaches needs to move beyond the traditional AI metrics. Children and adults with ASD, their families and the society are asking for a different kind of evaluation. We will show our recent efforts in this direction with preliminary results of the French Living & Learning in Neuro-Development.
Bio Prof. Mohamed Chetouani is Full Professor in signal processing and machine learning for human-machine interaction at the Institute for Intelligent Systems and Robotics (CNRS UMR 7222), Sorbonne University. He is part of the PIRoS research team (Perception, Interaction et Robotique Sociales). His activities cover social signal processing, social robotics and interactive machine learning with applications in psychiatry, psychology, social neuroscience and education. He is the Deputy Director of the Laboratory of Excellence SMART Human/Machine/Human Interactions In The Digital Society. Since 2018, he is the coordinator of the ANIMATAS H2020 Marie Sklodowska Curie European Training Network. Since 20219, he is the coordinator of the National Living and Learning Lab in Neurodevelopment (LiLLaB). He was the local co-chair of IEEE ICRA 2020 (Paris) and Program co-chair of ICMI 2020 (Utrecht). Since 2020, he is the Chair of the Ethical Advisory Board of Sorbonne University.
Abstract Although the neuroscience literature contains many examples of the use of computational models to describe processes and biological patterns, there are fewer reports of their use in modeling behavior, primarily due to difficulties in simplifying such complex phenomena. In fact human behavior is much too complex, nonlinear, and unpredictable due to the irrationality of even healthy humans. Even with very efficient models it is impossible to predict the influence of irrationality on human behavior.
If human behavior cannot be predicted, how can social simulation be useful in predicting society as a whole? Complexity science provides a paradigm to help answer this difficult question. In fact one focus of social simulation has been studying emergent properties and the simple rules that can be useful to policy-makers in understanding a social change due to a simple change at a lower level. Computational sociology and computational economics received much attention in the last decade, and the improved computational capability of commercial computers has extended its use in many fields, from industry to the arts and many other applications.
In the meantime, behavior, which is not predictable, has been considered only within limited application and simplified to what is simply recorded, such us gesture, voice, gait, and so on. This is not wrong but surely is incomplete. In fact human behaviors are a complex of human actions that are not limited to single steps like a gesture, but comprehend intentions, consciousness, willing, and many other wonderful concepts that unfortunately make human behavior unpredictable and difficult to model.
Therefore, researchers have had to resign themselves to the fact that there is no way to integrate real behavior in social simulation. If this is true it is unthinkable to use computational simulation at an individual level, and simulation is limited in its use either for social emergence or in low-level processes of cognition (such as specific neural mechanisms). At higher cognitive levels there are too many variables to make the processes predictable for specific purposes (such as decision-making), so simulation will be useful to neuropsychology only at low cognitive-levels in the understanding of specific mechanisms. To predict a patient’s behaviors will never be done with simulations.
However, there is a radical change in this idea that would allow researchers to integrate computational intelligence (and in particular computational simulations) into neuropsychology.
Since there is no way to predict human behavior, the only possible solution to have a realistic simulation for neuropsychology is to integrate real humans into artificial simulation. In particular our aim was to integrate real human behavior into computational models, which means bringing real experiments with human-beings into simulated worlds.
However, it is not enough to connect human beings and artificial simulation, we need to include human behavior and this means to immerse humans in prototypical situations within which human behaviors make clinical or experimental sense.
Since it would be very expensive, and sometimes impossible, to create a real scenario that would involve actors and artificial scenes, an alternative would be to use Virtual Reality (VR) to create a virtual scenario.
Virtual reality has already been used in clinical and experimental settings and is a validated method to elicit human behaviors within specific situations.
Bio Pietro Cipresso is Associate Professor of Psychometrics at Department of Psychology, University of Turin, and Senior Researcher at the Applied Technology for Neuro-Psychology Laboratory at Istituto Auxologico Italiano, Milan, Italy. Prof. Cipresso, has coordinated and coinvestigated several National, European and International Projects and has been Visiting Researcher at Massachusetts Institute of Technology (MIT), Cambridge, MA, USA, and at Monash University, Melbourne, Australia. He has been the unit team leader of the EU Project (BodyPass: API-ecosystem for cross-sectorial exchange of 3D personal data). Prof. Cipresso is also the Chief Editor for Frontiers in Quantitative Psychology and Measurement, IF: 2.463, Q1. In this role (jointly with Prof. Immekus), he is managing a Board of more than 600 Editors.
Computational Psychometrics integrating artificial intelligence, virtual reality and mental health
Prof Pietro Cipresso
Speech analysis for mental health: opportunities and challenges
Dr Nicholas (Nick) Cummins
Abstract: The production of speech is remarkably complex, combining conscious and subconscious cognitive thought with the physical actions of the respiratory system. This complexity means that changes in our health state can affect speech, often at a subconscious level. Such changes can then alter both the linguistic, acoustic and prosodic content in ways that are potentially measurable through intelligent signal processing techniques. The multifaceted nature of speech means it is uniquely placed as a signal of interest in remote monitoring (RMT) mobile health (mHealth) studies. No other RMT signal contains such a combination of cognitive, neuromuscular and physiological information. However, considerable research efforts are required to realise the potential of speech as a mHealth biomarker. This talk will consist of three parts, a general introduction to speech production, an overview of current works in this area and an outlook on the challenges and opportunities associated with this fascinating and unique health signal.
Bio: Nicholas (Nick) Cummins is a lecturer in AI for speech analysis for health at the Department of Biostatistics and Health Informatics at King’s College London. Nick’s current research interests include speech processing, affective computing and multisensory signal analysis. He is fascinated by the application of machine learning techniques to improve our understanding of different health conditions and mental health disorders in particular. Nick is actively involved in RADAR-CNS project in which he assists in the management of Work Package 8: Data Analysis & Biosignatures. Nick was awarded his PhD in electrical engineering from UNSW Australia in February 2016 for his thesis ‘Automatic assessment of depression from speech: paralinguistic analysis, modelling and machine learning’. After completing his PhD, he was a postdoctoral researcher at the Chair of Complex and Intelligent Systems at the University of Passau, Germany. Most recently, he was a habilitation candidate at the Chair of Embedded Intelligence for Health Care and Wellbeing at the University of Augsburg, also in Germany. During his time in Germany, he was involved in the DE-ENIGMA, RADAR-CNS, TAPAS and sustAGE Horizon 2020 projects. He also wrote and delivered courses in speech pathology, deep learning and intelligent signal analysis in medicine
Abstract This presentation will discuss the social signals associated with attachment and attachment disorders and some work, in collaboration with Alessandro Vinciarelli and other computing science colleagues to use this knowledge in development of new understanding and clinical measures.
Bio Helen Minnis has a longstanding clinical and research focus on the mental health problems associated with abuse and neglect. This has included investigating the clinical features, population prevalence and behavioural genetics of Attachment Disorders and investigating the way trauma and neurodevelopmental problems contribute towards the development of severe mental illness. She is currently leading three randomised controlled trials of complex interventions for children who have experienced, or are at risk of, maltreatment. These involve a wide range of multi-agency partners including colleagues from social care, education, and the judiciary. Helen has collaborated with Computing Science colleagues to develop the first computerised measure of attachment and is working with Alessandro Vinciarelli to supervise a social signal processing PhD on Reactive Attachment Disorder.