RefuCare: Revolutionizing Refugee Mental Health with AI-Driven Wearable Support
by Jonanne Talebloo
Coauthors: Coauthor: Ishita Srivastava (UC Berkeley Student, MI4 Intern)
Refugees and migrants are one of the most vulnerable members of society. Migrants are frequently excluded from national health programs, disease prevention, and financial protection in health. They also face inadequate interpreting services, high user fees, and poor cultural competency among health providers. Particularly, unaccompanied minors and refugee children are at a higher risk of experiencing traumatic events – abuse, inadequate housing, interrupted schooling – where parents and caregivers are unable to provide adequate help due to their own trauma, leading to difficulties accessing mental health care.
RefuCare uses artificial intelligence (AI) in order to address the mental health care needs of refugee and immigrant children. RefuCare is an AI-driven system that delivers personalized and accessible mental health care interventions tailored to the distinct experiences and needs of minors who are refugees and immigrants.
Our solution includes a wearable device featuring an AI-powered chatbot that constructs a comprehensive digital profile, utilizing personality traits, and that allows users to communicate with other refugees, mitigating feelings of isolation. This facilitates effective stress management and mental health support for individuals. Furthermore, the wearable device contains the capability to detect and interpret voice and speech across various languages, including Pashto, Dari, Farsi, Kurdish, and Arabic. Leveraging AI algorithms, the device analyzes voice tone and physiological indicators to identify distress signals and allow for timely interventions to be implemented.
Moreover, the wearable device incorporates skin-sensing technology capable of discerning alterations in electrodermal activity. When emotional arousal occurs, the sympathetic nervous system transmits pulse-like signals to the sweat gland, resulting in variations in sweat activity.
To ensure user privacy and reduce reliance on external cloud infrastructure, the wearable device performs on-device signal processing and analysis. This necessitates implementing low-latency algorithms with minimal computational requirements, enabling real-time processing and accurate detection of distress signals within the wearable itself. Techniques like adaptive filtering, feature extraction, and machine learning-based classifiers are used, dramatically reducing costs and increasing accessibility. Additionally, encryption protocols to protect data transmission and storage, access controls to ensure authorized usage and compliance with data protection regulations such as General Data Protection Regulation (GDPR) are implemented.
The technology allows users to detect early physiological reactions to stimuli, employing personalized interventions such as soothing music to prevent escalation. Our product can be adapted to cater to the general public. However, RefuCare is tailored for refugee communities, drawing insights from experienced social workers embedded in those communities. Moreover, our approach involves a rigorous research process that includes surveys and soliciting input from refugees, ensuring that recommended practices and wearable responses encompass culturally relevant elements.
Lastly, the device can be linked to a purpose-built mobile application aimed at addressing depression and anxiety by leveraging cognitive behavioral therapy and evidence-based coping techniques. The app delivers personalized recommendations and, through the utilization of natural language processing, enhances its suggestions by extracting thematic and pattern-based insights from user data. This valuable information contributes to an improved understanding of participants' experiences and augments the efficacy of the AI system.