Atai Life Sciences: Monitoring Depression Through Sleep

Industry: Healthcare & Life Sciences
Solution: ML-based depression severity prediction via actigraphy data
Impact: Identified sleep phenotypes and achieved 80% prediction accuracy within 2 months

Overview
atai Life Sciences is a biopharmaceutical innovator focused on advancing mental health treatments. As part of their digital medicine initiative, they sought new ways to monitor depression symptoms using passive, real-world data from wearable devices.

The Challenge
The goal was to classify depression severity and understand patient sleep subgroups by analysing actigraphy data from wearables like Fitbits and Apple Watches. The team needed a partner to help apply machine learning and signal processing techniques to large public datasets.

The Ishango Approach
Ishango developed a supervised machine learning model using gradient-boosting algorithms to predict depression severity based on actigraphy data. The team also worked to segment users into sleep phenotypes—offering a new dimension to how mental health symptoms are understood.

The Result
Within 2 months, the model achieved 80% accuracy in classifying depression levels and provided atai with actionable insights into how sleep patterns relate to mental health.