Figure 2.

The development lifecycle of DPSTs: (1) Multimodal data are collected and aggregated from digital devices along with contextual health data. (2) Data from diverse training data are processed and machine learning algorithms are leveraged to identify patient diagnostic likelihood and digital phenotypes. (3) Ethical deployment of the DPSTs to supplement current screening methods is tested in clinical practice with consideration for privacy and interpretability of patient data and experiences.