GOOD @ ICPR’2020

Designing AI in support of Good Mental Health 

January 10, 2021

Call for Contributions

Reality 

The world is experiencing a mental health crisis that is in need of immediate attention. In 2017, within the United States alone, an estimated 43 million people (18% of the population) had a mental illness and 9.6 million people experienced thoughts of suicide1. About fifty-six percent of such adults did not receive the treatment they needed due to a variety of socio-cultural and economic factors including lack of access to healthcare, a shortage of providers, and a fear of the stigma that may be associated with treatment1.  Beyond diagnosed mental health conditions, it is estimated that at least 3 out of 5 people have had experiences of mental illness.  These statistics hold for global populations as well.

The genesis of mental health conditions is instructive as well.  The American Psychiatric Association estimates that 50% of mental illness begins by age 14, and 75% of mental illness begins by age 24.  The National Association on Mental Illness in the US tracks suicide as the 3rd biggest cause of death in ages 10-24; 90% of these individuals had an underlying mental illness.  These ages are some of the most technologically connected segments of our life span, providing unique opportunities for technological solutions to play a key role in addressing this crisis.

Responses

Globally, governments, clinical institutions, and technology companies are starting to respond to this escalating crisis.  Some examples include: Governments (especially in Europe) starting to establish cabinet level positions for Mental Health, mental health education becoming required in school grades K-12 at a rapidly increasing rate, and partnerships by clinical institutions with communities around general care and prevention.  Technology has enabled easier remote access to both active listeners and counselors and psychiatrists through text and video channels.  

Research: Technical topics & unique opportunities 

We are at the dawn of technology supporting good self-care and other-care, especially for good mental health.  The mental health crisis is a huge opportunity for the research, design, and deployment of AI/ML-assisted systems to improve good mental health at scale.

This domain comes with unique challenges.  In several mental health scenarios, there isn’t a high degree of agreement between experts on the ground truth of a situation, and culture and its dynamism adds complexity.  Datasets can be small or noisy as they capture human expression during periods of desired change.  There are often several effective paths forward in any given situation, and not just one “best move”.  There is a large cohort of trained providers who would like to use technology well to expand their scale, and to augment their efficacy.  There are several other community-centered ecosystems - school, work, religion, others - that play a key role in good mental health and can benefit from technology that leverages their presence.  And technology can augment all aspects of fostering good mental health: awareness, assessment, mediation, and maintenance.

This workshop will bring together industrial and academic experts and researchers and field practitioners to elucidate problems, solutions, and paths forward.  We invite short papers (anywhere from 2-4 pages + citations) that speak to the topic of designing artificially intelligent systems for good mental health.  Some topics that the papers might cover include the following (note that this is not a restrictive list!)

Important Dates

Workshop paper submissions: October 10, 2020

Workshop paper notifications: November 10, 2020

Camera-ready submissions: November 15, 2020

Workshop date: January 10, 2021

Schedule (Tentative)

9 - 9:05                Greeting and Connection exercise

9:05 - 9:20                 Introduction & Framing for the workshop

9:20 - 10:20                Fast & Focused paper talks (6-10), 5-7 minutes each.  

The talks will be structured as:  What’s the problem you are trying to solve, what you learnt (and can teach us), what do you think the important problems to solve are.

10:20 - 10:30        Break & Post-Talk Discussions  

10:30 - 11:15        Industry and Academia panel

The panel will discuss 3 key questions in the area, which will be clustered from the topics of the papers, and general open issues in the area.  There will be 10 minutes of discussion per question roughly, for 30 minutes in total, followed by 15 minutes of discussion with the room.

11:15 - 11:20                Stretch Break

11:20 - 11:50                Invited talk, 20 minutes + 10 mins Q&A

11:50 - 12:30        Problem Generation and Elucidation

This session will conclude the workshop in a generative manner.  With 20 minutes, different groups will work on enumerating open problems for different aspects identified during the day, and connecting them with ML techniques or open research problems.  We will synthesize these group results into a larger research map for the space that participants can take away with them for inspiration.

Program Committee

TBD

Organizing Committee

Ajay Chander, Fujitsu Labs of America

Martha Russell, Stanford University

Anssi Smedlund, Finnish Institute of Occupational Health

References

1. T. Nguyen. State of mental health in america. Mental Health America, 2017.