We know CBT-I works. But can it be better?

Our lab is developing a machine learning system (Lumeni) trained on continuous wearable data—including heart rate variability, activity patterns, light exposure, and sleep history—that predicts the optimal moments to deliver CBT-I interventions. When Lumeni detects elevated risk, such as reduced physical activity that may decrease sleep drive, irregular light exposure that may disrupt circadian rhythm, or increased stress levels that may elevate arousal, it delivers CBT-I-based guidance to the user in the moment.

This just-in-time approach transforms CBT-I from a static clinical protocol into a dynamic, adaptive system that supports users when they need it most.