Theme #3 of 5:
Timely mental health intervention in emergencies/war zones
New video demos uploaded
Theme #3 of 5:
Timely mental health intervention in emergencies/war zones
This is a supplementary website for the paper titled "Creative use of AI for humanitarian causes: HADR AI toolkit"
Note this paper contributes on 5 HADR themes.
This webpage presents Theme #3. For other 4 contributions of the paper, please click here..
Recommended viewing: This comprehensive video covers demos of all 5 themes in 20 minutes
Video Demo: Recommended for review
Contribution Theme #3 : Cognitive Behavior Therapy in emergencies/ conflict zones
Timely mental health intervention in emergencies/conflict zones
HADR challenge: 1 in 5 refugees are impacted by metal health disorder, as per WHO [2, 3, 6]. UNICEF records that the trauma caused in war can impact many years of life. The prolonged war has disrupted the social fabric & children in Ukraine. Emergencies & conflict disrupt mental wellbeing of significant population. Around 1 in 5 adolescents have a mental disorder as per WHO. United Nations has recommended actions to ensure people and societies are better protected from the mental health impact of COVID-19 & emergencies in Ukraine. Early intervention is an implementation challenge. The dimensions of the challenge are
o Scalable: How to reach millions of families given shortage of professional staff
o Timely: Early & timely intervention
o Privacy: Sensitive information handling
Given significant rise on Mental health concerns among sizable population who need mental health services, and the lack of capacity of mental health professionals to handle emergences, how to provide timely mental health intervention to population in a privacy safe approach. How to enable self-help using therapy science such as Cognitive Behavior Therapy (CBT)? How to identify who needs help? How to provide timely intervention?.
This paper's Contribution: HADR-AI solution: It is important to keep a tab on the person’s progression of mental health. To support natural human resilience, and to avoid over-dependency on counselling, the AI shouldn’t hinder the natural process of resilience of humankind. If the person shows signs of natural recovery from a depressed mental state due to natural resilience, therapy is no more required. So the AI therapy needs to be activated/de-activated when appropriate. Based on the progression, the decision to intervene is taken intelligently. This idea of intelligent activation of therapy based on how the person is doing over time to create a balance between natural is a unique contribution to the literature. We demonstrated this intelligent activation by building a lab scale prototype. A novel idea of NLU (Natural Language Understanding) activated NLG therapy is demonstrated with reproducible results using a BERT based classifier to activate the GPT2 based therapy. We have fine-tuned a GPT-2 model to provide AI based self-help based on CBT science. We have been in discussions with local UNICEF officials to advance the solution. We share the AI source code, the video demo on this paper’s project website.
AI challenge: In conflict zones, where staffing of human councilors are challenging, how to enable self-help based mental health therapy approaches? How to fine-tune language model (GPT-x) to provide appropriate counselling such as CBT?. How to develop privacy safe approaches using on-device ML on smartphone deployment? .