DeepNoid, a first-generation domestic medical AI company, will participate in a national research and development project worth 18 billion won.
On the 30th, DeepNoid announced that it has been selected as a joint research institutional investor for the Korean ARPA-H project. In this initiative, DeepNoid is responsible for developing an AI foundation model for managing elderly mental health.
The Korean ARPA-H project is a national R&D program supported by the Ministry of Health and Welfare and the K-Health Future Promotion Team of the Korea Health Industry Development Institute. It aims to resolve critical national healthcare challenges through mission-oriented medical R&D that, despite high costs and complexity, yields significant impact.
South Korea entered the super-aged society at the end of 2024, when the proportion of the population aged 65 and over surpassed 20%. The suicide rate among the elderly also stands at 37.9 per 100,000 people as of 2024, the highest level among OECD countries. This is why addressing elderly mental health management is considered an urgent issue.
The newly selected 2026 Korean ARPA-H project is 'BCI-INSIDE-OUT,' aimed at transforming elderly mental health management. The research team was selected after proposing the 'AI Caregivers' task. A total of 18 billion won in government R&D funding will be invested in this project from 2026 to 2030.
The lead R&D institutional investor is the Jeon Hong-jin (Prof.) research team from the Department of Psychiatry at Samsung Seoul Hospital. DeepNoid, Seoul National University, Sungkyunkwan University, and Kyung Hee University Hospital are participating as joint research institutional investors. Seoul National University Hospital and Kyung Hee University are participating institutions, while NewLive, Wonderful Platform, and DeepMedi are collaborating as technology partners.
The goal of the task is to use AI to integrate and analyze non-invasive brain-computer interfaces, brain imaging, electroencephalograms (EEG), and biosignals to measure changes in elderly mental health status in real time. The team is also working to develop a closed-loop management system where measurement, analysis, and intervention are automatically linked within one second.
In this consortium, DeepNoid plays a central role in AI development. It will safely integrate and standardize various forms of data collected from multiple medical institutional investors and outpatient settings, and use this data to train a multimodal AI foundation model called the 'Elderly Mental Health-Multimodal Foundation Model' for objectively assessing and predicting elderly mental health status.
The core of DeepNoid's role is ensuring that the developed model does not remain confined to research. The company will stably deploy and operate the model in clinical settings and on public clouds, implementing a closed-loop structure where measurement, analysis, and intervention cycle in real time. It plans to connect this with sustainable private-public linked services.
This role aligns with DeepNoid's vision of becoming an 'agent AI-based medical service company.' Based on its technology for supporting image reading workflows, DeepNoid is developing agent AI technology. Through this national project, it plans to expand into areas connecting multimodal data such as EEG and biosignals with measurement, analysis, and intervention.
Choi Woo-sik, CEO of DeepNoid, stated, "DeepNoid is moving forward as an agent AI-based medical service company that supports the entire reading workflow." He added, "In this national project, we will develop a multimodal agent AI that connects measurement, analysis, and intervention, further enhancing our foundation model and agent AI capabilities accumulated in imaging."
in TECH
