South Korea Launches $13.5M AI Project to Predict and Prevent Elderly Mental Health Crises in Under One Second

South Korea has launched a national research initiative to address its elderly mental health crisis using artificial intelligence designed to detect and respond to psychiatric distress in less than one second. The 18 billion won (~$13.5 million USD) project, part of Korea's ARPA-H "BCI-INSIDE-OUT" program, will run from 2026 to 2030 and aims to create a multimodal AI foundation model capable of real-time monitoring and intervention for aging populations.

Medical AI company DeepNoid has been selected as a key participant in the consortium, led by Samsung Seoul Hospital's Department of Psychiatry under Prof. Jeon Hong-jin. The initiative responds to pressing demographic challenges: South Korea became a super-aged society in late 2024, with over 20% of its population aged 65 or older, and currently has the highest elderly suicide rate among OECD countries at 37.9 per 100,000 as of 2024.

Building a Closed-Loop Mental Health Monitoring System

The project's technical approach centers on integrating multiple data streams into a unified AI system. Non-invasive brain-computer interfaces, brain imaging, electroencephalograms (EEG), and biosignals will feed into what researchers are calling an "Elderly Mental Health-Multimodal Foundation Model" designed for objective assessment and prediction of mental health status.

The system features a closed-loop architecture with a cycle time of less than one second from measurement to intervention. According to the project description, the AI will continuously monitor an elderly person's mental state through various biosignals and brain activity patterns, analyze the data in real time, and trigger interventions before a crisis escalates.

The consortium includes academic partners from Seoul National University, Sungkyunkwan University, Kyung Hee University Hospital, and Seoul National University Hospital, alongside technology companies NewLive, Wonderful Platform, and DeepMedi. The system is planned for deployment through both clinical and public cloud infrastructure with private-public linked services.

From Image Reading to Agent AI in Mental Health

For DeepNoid, the project represents a strategic expansion beyond its existing medical imaging capabilities. The company is positioning itself as an "agent AI-based medical service company," moving from image reading workflows to comprehensive multimodal data integration designed to autonomously monitor and respond to patient conditions.

The potential clinical impact centers on proactive, continuous monitoring without requiring constant clinical visits. The system aims to reduce emergency psychiatric admissions and suicide attempts among elderly patients by providing healthcare providers with automated alerts when the AI detects concerning patterns, potentially enabling intervention before a mental health crisis fully develops.

Implications for Global Geriatric Psychiatry

While the project addresses South Korea's specific demographic challenges, it establishes a technical blueprint that could influence geriatric mental health care internationally. The integration of diverse data streams—EEG, biosignals, brain imaging, and brain-computer interfaces—into a single foundation model represents a novel approach to psychiatric monitoring that prioritizes speed and automation.

The emphasis on non-invasive monitoring technologies suggests a pathway toward mental health surveillance that doesn't burden elderly patients with complex medical procedures or frequent hospital visits. As the global population ages, South Korea's investment in real-time AI mental health monitoring may offer insights into how technology can address the psychiatric needs of super-aged societies facing similar challenges with elderly suicide rates and mental health resource constraints.

Sources

1. Elderly Mental Health Managed in Real Time with AI… DeepNoid Joins 18 Billion Won National Project — MoneyToday