Promoting Engagement in Remote Patient Monitoring Using Asynchronous Messaging

Document Title

Promoting Engagement in Remote Patient Monitoring through Asynchronous Messaging

Document Information

• Subject Area: Human-Computer Interaction, Telemedicine, Digital Health
• Keywords: Remote Patient Monitoring, Asynchronous Messaging, Disease Management, Data Quality, Patient Engagement, COVID-19
• Conference: CHI ’24 (Human Factors in Computing Systems)

Research Background and Problem

• What issues or challenges did the authors identify?

  • Despite the vast potential of remote patient monitoring, it may weaken communication between patients and healthcare providers, leaving patients feeling uncertain or undervalued.
  • Patients may feel uncomfortable with the platform's interface or technology, leading to a decline in data quality.
  • Over time, patients may lose motivation to upload health data, affecting engagement and long-term data collection.

• Why is this problem important?

  • Remote patient monitoring reduces time and financial burdens for both patients and doctors, optimizing the use of acute medical resources.
  • During the COVID-19 pandemic, strained medical resources necessitated better tools to support patient recovery monitoring and reduce hospital burdens.

• Research Motivation and Related Work

  • Compared to other remote monitoring platforms, this study focuses on the positive impact of asynchronous messaging on communication between patients, doctors, and researchers.
  • While asynchronous messaging features exist in some electronic health record systems, few studies have thoroughly analyzed their specific mechanisms and benefits within remote monitoring platforms.

Solution

• What methods or solutions did the authors propose?

  • The COVIDFree@Home platform integrates asynchronous messaging functionality, allowing doctors to monitor patients' symptoms and vital signs.
  • The study uses user-initiated messages for thematic analysis to evaluate how the messaging feature enhances protocol adherence, supports patients in discussing health issues, and enables doctors to convey care.

• What is innovative about this solution?

  • Asynchronous messaging not only improves communication between patients and healthcare providers but also directly facilitates more efficient and accurate data collection, enhancing the predictive performance of machine learning models.
  • The specially designed messaging interface reduces the burden on doctors while providing patients with a personalized feedback channel.

• Implementation Steps—What key technologies were used?

  1. Platform Design: Separate mobile applications for patients and web-based dashboards for doctors to facilitate data input and message exchange.
  2. Messaging Feature Usage Statistics: Detailed analysis of the time distribution and interaction patterns of message exchanges.
  3. Thematic Analysis: Open coding method applied to categorize message content and identify communication patterns and themes.
  4. Machine Learning Analysis: Few-shot learning techniques used to demonstrate how message-driven data corrections improve predictive models.

Research Outcomes

• What specific outcomes were achieved?

  • The asynchronous messaging feature improved communication between patients and doctors while adding an emotional support layer to the platform.
  • The platform supported 359 patients, with 16 requiring escalated care.
  • The asynchronous feature recorded and corrected data issues, significantly enhancing data integrity.

• What advantages does it have compared to existing solutions?

  • The asynchronous messaging feature allows flexible scheduling for doctors' responses, increasing patient data submission rates.
  • Compared to traditional phone or face-to-face communication, asynchronous messaging reduces the need for time synchronization.
  • The messaging feature directly records and links patient issues, supporting timely detection of device malfunctions and data errors.

• What were the experimental or evaluation results?

  • Data corrections significantly improved the performance of machine learning-based hospitalization prediction models: AUROC increased from 0.53 to 0.59.
  • Message exchanges facilitated the identification of device malfunctions and data entry errors, reducing data quality issues in the study.

• Limitations and Future Directions

  • This study was conducted primarily in a metropolitan area in North America, and its international applicability has not yet been validated.
  • The data is biased toward mild cases, which may limit the model's applicability to broader healthcare scenarios.
  • Future research could explore the applicability of asynchronous messaging features in interactions between patients and doctors across diverse demographic and cultural contexts.

Appendix and Example Information

• Example Message Content:
  • From Doctors:
    • “Your oxygen saturation is very low. Please measure it a few more times and upload the results.”
    • “Hello, why haven’t you submitted information today? If there’s a problem, feel free to contact me.”
  • From Patients:
    • “Should I be worried about oxygen readings below 92?”
    • “I’m feeling much better. Can I stop uploading data?”