Software Reference Architecture for Real-Time Mobile Digital Phenotyping: Evaluation of System Designs

Area of Science:

• Digital Health
• Mobile Computing
• Biomedical Informatics

Background:

• Digital phenotyping leverages smartphone data for health monitoring.
• Current methods face computational limits on mobile devices, relying on cloud processing.
• This necessitates efficient on-device data analysis for real-time health insights.

Purpose of the Study:

• Introduce the Stanford Screenomics platform, a novel software architecture for on-device digital phenotyping.
• Enable scalable, real-time health monitoring through parallel processing and edge computing.
• Compare the performance of the Stanford Screenomics platform against traditional cloud-based approaches.

Main Methods:

• Developed two prototype apps: Stanford Screenomics (on-device, parallel) and a traditional cloud-based app.
• Conducted 48-hour experiments under varying data loads (10-60 MB/min).
• Assessed offline resource performance (CPU, RAM, battery, data loss) and end-to-end phenotyping latency.

Main Results:

• Stanford Screenomics showed lower CPU (3.9%-14.6%) and RAM usage (97-132 MB) than traditional methods.
• It exhibited reduced battery drain (0.9%-2.1%/h) and significantly less data loss (0.4%-1.5%/h).
• Phenotype updates were 34-43x faster (0.9-9.3s vs 30.1-398.1s) with lower variability.

Conclusions:

• The Stanford Screenomics platform enables high-fidelity, low-latency digital phenotyping directly on smartphones.
• This architecture provides a resilient foundation for scalable mobile health interventions.
• It paves the way for next-generation context-aware health monitoring via mobile devices.