Predictive analysis
Models that predict symptoms, behaviors or outcomes from multimodal data.
Predictive models that hold up in the real world.
We build and validate machine learning models on longitudinal wearable, smartphone and EMA data, and explain what drives each prediction.
What we deliver
Machine learning and forecasting, end to end.
Risk stratification
Identifying participants at higher risk, for example of preterm birth from heart rate and HRV.
Treatment response prediction
Who is likely to benefit from an intervention.
Explainable AI
Feature importance and SHAP explanations that reviewers and clinicians can follow.
Time-series forecasting
Forecasts of affect, sleep or physiology, personalized to each participant.
Natural language processing
Analysis of free-text responses, diaries and conversations.
How we analyze it
- Personalized and adaptive models
- Gradient boosting and deep learning
- Validation designed to avoid leakage across participants and time
- Explainability with SHAP
What you receive
- Validated models and performance reports
- Explainability figures
- A methods section and code for reproducibility
- Plain-language summaries for your team
Evidence
Selected peer-reviewed work by our team.
- Feli et al. Preterm birth risk stratification through longitudinal heart rate and HRV monitoring in daily life. Scientific Reports, 2024. doi:10.1038/s41598-024-70773-0
- Jafarlou et al. Objective prediction of next-day’s affect using multimodal physiological and behavioral data: algorithm development and validation study. JMIR Formative Research, 2023. doi:10.2196/39425
- Yang et al. Integrating wearable sensor data and self-reported diaries for personalized affect forecasting. Smart Health, 2024. doi:10.1016/j.smhl.2024.100464
- Borelli et al. Detection of depressive symptoms in college students using multimodal passive sensing data and light gradient boosting machine: longitudinal pilot study. JMIR Formative Research, 2025. doi:10.2196/67964
- Tazarv et al. Active reinforcement learning for personalized stress monitoring in everyday settings. ACM/IEEE CHASE, 2023. doi:10.1145/3580252.3586979
- Jiang et al. Personalized and adaptive neural networks for pain detection from multi-modal physiological features. Expert Systems with Applications, 2024. doi:10.1016/j.eswa.2023.121082
- Wang et al. Differential private federated transfer learning for mental health monitoring in everyday settings: a case study on stress detection. IEEE EMBC, 2024. doi:10.1109/EMBC53108.2024.10782516