Hyper-DREAM, a Multimodal Digital Transformation Hypertension Management Platform Integrating Large Language Model...

Area of Science:

• Digital Health
• Artificial Intelligence in Medicine
• Cardiovascular Disease Management

Background:

• Limited systematic research exists on mHealth frameworks for hypertension, particularly in establishing digital health archives and leveraging large language models (LLMs) for patient data analysis and clinical decision support.
• Effective management of hypertension requires comprehensive patient data and advanced tools to assist clinicians in monitoring and controlling blood pressure (BP).

Purpose of the Study:

• To describe the design, development, and usability evaluation of the Hyper-DREAM platform, an mHealth tool for hypertension management.
• To assess the performance of the platform's BP Coach component against other AI models and clinicians.
• To evaluate the impact of the Hyper-DREAM platform on clinician workload and patient management efficiency.

Main Methods:

• Iterative design and development of the Hyper-DREAM platform over one year, incorporating multimodal data collection, interventions, and interactions.
• Usability evaluation using the mHealth App Usability Questionnaire (MAUQ) with 51 hypertensive patients and the Doctor's Software Satisfaction Questionnaire (DSSQ) with 6 clinicians.
• A comparative experiment evaluating the BP Coach, ChatGPT-4o Mini, ChatGPT-4o, and clinicians on utility, completeness, clarity, and conciseness.

Main Results:

• The BP Coach component scored higher in utility and completeness than ChatGPT-4o Mini, ChatGPT-4o, and clinicians.
• Clinicians reported improved work efficiency and patient interactions with Hyper-DREAM, with decreased work intensity and routine disruptions.
• Hyper-DREAM demonstrated significantly higher overall patient satisfaction compared to standard WeChat-based management, with reduced patient management time.

Conclusions:

• Hyper-DREAM is a novel, AI-driven platform for hypertension management that effectively alleviates clinician workload and enhances patient care.
• The platform exhibits user-friendliness, high satisfaction rates, utility, and efficient information organization, showing significant promise for clinical integration.
• The BP Coach component highlights the potential of LLMs in advancing mHealth strategies for hypertension management.