A new mental health diary app developed in South Korea leverages generative AI to early screen depression while securing sensitive user data. 

HOW IT WORKS

Doctorpresso came up with a mobile app called REDI that analyses a user's speech patterns and provides near real-time insights into their mental state. 

Through AI, the app analyses four voice characteristics: average pitch, pitch variability, spectral centroid, and spectral roll-off. These factors, which reportedly change as depression worsens, are used to classify users, whether they are not depressed, have mild depression, or have a major depressive episode. 

In a study in 2021, REDI's AI demonstrated over 90% accuracy in detecting depression. 

Another innovative feature of REDI's AI is it is based on a small language model. This enables the app to keep a user's voice data directly on their device, maintaining privacy. 

REDI also offers a calendar for tracking daily habits, such as sleep, alcohol consumption, and physical activity. 

MARKET SNAPSHOT

Doctorpresso is now preparing for a worldwide rollout of its mental health app. It is sharing a space that is currently dominated by the likes of Sonde and Kintsugi.

The latest research in South Korea has also employed genAI to detect neurodegenerative disorders. Researchers at the Daejeon-based Electronics and Telecommunications Research Institute integrated a large language model with existing voice and text analysis models to predict a senior individual's risk of developing dementia.