DISPEL: A Python Framework for Developing Measures From Digital Health Technologies

Area of Science:

• Digital Health
• Neurodegenerative Diseases
• Biomedical Engineering

Background:

• Digital health technologies generate vast amounts of data relevant to neurodegenerative disease progression.
• Developing reliable sensor-derived measures (SDMs) from this data is crucial for therapeutic development.
• Existing frameworks often lack modularity, standardization, and robust handling of real-world data issues.

Purpose of the Study:

• Introduce DISPEL, a novel Python framework designed to streamline the development of sensor-derived measures (SDMs).
• Facilitate the use of digital health technology data in therapeutic development for neurodegenerative diseases.
• Provide a standardized, flexible, and integrable solution for SDM extraction and processing.

Main Methods:

• Employed an object-oriented architecture for modular data modeling and SDM extraction.
• Implemented systematic flagging for missing data and unexpected user behaviors common in unsupervised monitoring.
• Ensured standardization of SDM generation, naming, storage, and documentation.

Main Results:

• DISPEL is an open-source (MIT license) Python framework supporting various data formats.
• It enables traceable end-to-end processing from raw wearable and smartphone data to structured SDM datasets.
• Currently extracts SDMs from 16 structured tests (including questionnaires) assessing disability, quality of life, cognition, dexterity, and mobility.

Conclusions:

• DISPEL provides a production-grade framework to support SDM development for clinical trials.
• The framework includes a comprehensive set of pre-implemented SDMs, refined using clinical trial data.
• Users are advised to perform context-specific validation of the provided algorithms for their intended use.