Automated mHealth Application Development: A Case Study on the P-STEP App for Atrial Fibrillation

Project Highlights

1. This project leverages NLP and LLMs for automated, precise code generation, drastically reducing mHealth app development time.
2. It enables dynamic adaptation of app functionalities using real-time data, ensuring personalized and context-sensitive health recommendations.
3. Rigorous clinical validation will establish new benchmarks for AI integration in healthcare, enhancing the reliability and scalability of mHealth solutions.

Project

The integration of advanced computational techniques such as Natural Language Processing (NLP), Large Language Models (LLMs), and model-driven engineering offers groundbreaking opportunities to revolutionize mHealth application development. The high prevalence of Atrial Fibrillation (AF) — impacting around 1.4 million individuals in England[1] — demands innovative technological solutions to manage this widespread cardiovascular condition effectively. The P-STEP app, designed to promote physical activity under specific environmental considerations, exemplifies a prime candidate for leveraging such sophisticated computational methodologies to address real-world healthcare challenges.

Aim

This project will redefine the code generation process for mHealth applications through the innovative use of NLP, LLMs, and model-driven engineering, with the enhancement of the P-STEP app serving as a central case study. The specific objectives include:

1. Automated Code Generation: Developing an NLP and LLM framework to automate the translation of natural language specifications into model-driven app features and functionalities.
2. Dynamic System Modeling: Implementing model-driven engineering to convert these specifications into executable code that dynamically adapts to user-specific data and environmental variables in real-time.
3. Clinical Validation: Assessing the functional accuracy and clinical efficacy of the P-STEP app against established medical standards to demonstrate the viability of these technologies in enhancing mHealth app development.

Methodology

1. Framework Development:

• NLP and LLM Integration: Establish an NLP framework capable of understanding and processing user and clinician inputs describing desired functionalities and app behaviour. LLMs will be utilized to expand these capabilities to include more complex, conversational interactions and richer context understanding.
• Model Specification: Use the processed inputs to automatically generate detailed system models that describe not only functionalities but also the adaptive behaviour of the app under various user conditions and environmental factors.

1. Code Generation and System Integration:

• Automated Software Engineering: Apply model-driven principles to transform system models into deployable code. This involves developing a scalable architecture that supports real-time data integration from multiple sources, including wearable technologies and environmental sensors.
• Dynamic Functionality Adaptation: Ensure that the app’s algorithms are capable of receiving and reacting to continuous streams of health data, adjusting recommendations based on activity levels, physiological feedback, and external environmental conditions.

1. Validation and Implementation:

• Simulated Environment Testing: Before clinical trials, simulate a range of environmental and user scenarios to test the robustness of the app’s response to varied conditions.
• Clinical Validation: Conduct comprehensive testing with AF patients, comparing app-generated data and recommendations against gold-standard measurements. This phase will evaluate not just accuracy but also the usability and acceptance of the technology by both clinicians and patients.

By employing cutting-edge computational technologies to automate and refine the development of mHealth applications, this project promises to significantly enhance the functionality and reliability of health interventions. The P-STEP app will serve as a pioneering example of how model-driven engineering, supported by NLP and LLMs, can be effectively applied to create sophisticated, adaptive mHealth solutions that cater to specific clinical needs, paving the way for broader applications in healthcare technology.

[1] https://cks.nice.org.uk/topics/atrial-fibrillation/background-information/prevalence/ 

Project enquiries to supervisor: Dr Artur Boronat artur.boronat@leicester.ac.uk

General enquiries CMSpgr@le.ac.uk

How to Apply

Please refer to the application advice and use the application link on our web page

https://le.ac.uk/study/research-degrees/funded-opportunities/cms-gta