Matching science to reality: how to deploy a participant-driven digital brain health platform

Ileana De Anda‐Duran, Phillip H Hwang, Zachary Popp, Spencer Low, Huitong Ding, Salman Rahman, Akwaugo Igwe, Vijaya B. Kolachalama, Honghuang Lin, Rhoda AuFrontiers in Dementia 2023 · DOI

Abstract

The importance of digital data in its native format, near real-time data access, passive participant engagement, and availability of technologies across OS has been supported by the pattern of participant technology use and adherence across cohorts. The precision brain health monitoring platform will be iteratively adjusted and improved over time. The pragmatic study design enables multimodal digital phenotyping of distinct clinically characterized cohorts in both rural and urban U.S. settings.

Introduction: The Search for Early Detection

In the field of Alzheimer’s Disease (AD) research, early detection remains the "Holy Grail." Identifying cognitive decline before significant damage occurs is crucial for intervention. However, current gold-standard diagnostics—such as PET scans and cerebrospinal fluid (CSF) analysis—are invasive, expensive, and often inaccessible to the general population. Traditional neuropsychological testing, while less invasive, typically requires in-person clinic visits and can be biased toward highly educated, Western populations.

Digital phenotyping—the use of sensors in smartphones and wearables to track behavior—offers a scalable alternative. Yet, many digital studies struggle with high attrition because they are too rigid, requiring participants to use specific, often unfamiliar, hardware. This paper presents a flexible, participant-driven platform designed to capture cognitive signals in the "wild" while respecting the realities of participants' daily lives.

A Participant-Driven Methodology

The researchers developed a precision brain health monitoring platform centered on flexibility and inclusivity. Unlike traditional studies that enforce a single protocol, this platform allowed participants at the Boston University Alzheimer’s Disease Research Center (BU ADRC) to choose from a suite of nine remote technologies.

Overview of the BU ADRC digital precision brain health platform data description, including cognitive measurements, typing behavior, physiological data, and more.

The platform categorizes engagement into three main streams:

  1. Active Engagement: Cognitive games (e.g., Lumosity) and specific tasks within the Linus Participant app.
  2. Passive Engagement: Background tracking of typing cadence (NeuraMetrix) and general phone usage patterns (Longevity).
  3. Wearables: Devices like Fitbit for activity tracking, or specialized tools like the Dreem3 headband and SleepImage ring for high-fidelity sleep monitoring.

A key technical priority was the collection of native data. Rather than relying on proprietary "scores" generated by apps, the platform captures raw sensor data (accelerometer, gyroscope, voice recordings). This approach future-proofs the research, allowing for retrospective analysis as algorithms improve and hardware evolves.

Testing Versatility Across Cohorts

To test the platform's real-world viability, it was deployed across two very different groups:

  • BU ADRC (Urban): 55 participants, mean age 72, primarily highly educated and living in the Boston area.
  • Bogalusa Heart Study (BHS, Rural): 94 participants, mean age 56, a more diverse cohort in Louisiana with lower average educational attainment.

Key Insights: Flexibility vs. Friction

The study revealed that flexibility is a powerful driver of engagement. In the BU ADRC cohort, participants chose to use an average of 5.9 different technologies, suggesting that older adults are highly willing to engage with complex tech when given agency.

However, the "wearable burden" remains a significant hurdle. While app engagement was high, specialized hardware like the Dreem3 sleep headband saw a 60% opt-out rate due to physical discomfort. This underscores a critical trade-off in digital health: higher-fidelity data (like EEG) often comes at the cost of lower compliance.

Furthermore, the "digital divide" was evident in the rural BHS cohort. Over 30 participants were unable to register initially due to poor internet connectivity, highlighting that "remote" research still depends heavily on physical infrastructure.

Results and Feasibility

The results demonstrate that a multimodal digital platform is feasible even for older populations:

  • High Interest: 98.2% of BU ADRC participants were interested in using multiple technologies.
  • Adherence: 97.9% of BU ADRC participants who opted into the Linus app completed at least one task.
  • Longitudinal Success: Data was successfully collected in 96.1% of scheduled assessment periods at the BU ADRC.

Limitations and Future Directions

Despite the success, challenges remain. "Device-based exclusion" is a concern, as 58% of participants had devices too old to run certain research apps. Additionally, the transition from clinic to "reality" requires significant human support for setup and troubleshooting.

Future work will focus on "harmonization"—creating data science pipelines to merge voice, gait, sleep, and typing data into a unified "digital signature" for brain health. The team also plans to integrate more passive, in-home sensors to further reduce the burden on participants.

Summary updated

Abstract

The importance of digital data in its native format, near real-time data access, passive participant engagement, and availability of technologies across OS has been supported by the pattern of participant technology use and adherence across cohorts. The precision brain health monitoring platform will be iteratively adjusted and improved over time. The pragmatic study design enables multimodal digital phenotyping of distinct clinically characterized cohorts in both rural and urban U.S. settings.
BibTeX
@inproceedings{andaduran2023matching,
  author = {Ileana De Anda‐Duran and Phillip H Hwang and Zachary Popp and Spencer Low and Huitong Ding and Salman Rahman and Akwaugo Igwe and Vijaya B. Kolachalama and Honghuang Lin and Rhoda Au},
  title = {{Matching science to reality: how to deploy a participant-driven digital brain health platform}},
  booktitle = {Frontiers in Dementia},
  year = {2023},
  doi = {10.3389/frdem.2023.1135451},
}
Figure