Projects at uah

Machine learning for human health monitoring


Research Mentor

Research problem and objectives

Wireless sensor technologies have revolutionized healthcare, enhancing human life quality. In smart healthcare, wearable sensors collect medical data for ML-enabled intelligent analytics, enabling applications like remote health monitoring. However, the reliance on a central server for health data learning in current ML systems poses privacy risks, especially in e-healthcare governed by regulations like United States Health Insurance Portability and Accountability Act (HIPPA). Federated learning (FL) emerges as a promising solution, enabling cost-effective smart healthcare applications with enhanced privacy. FL trains high-quality ML models by averaging local updates from multiple sensor devices, ensuring privacy preservation without direct access to local data.

 

Dr. Nguyen's team at UAH is actively exploring the applications of FL in health monitoring, focusing on three primary objectives: (1) creating a lightweight ML algorithm suitable for execution on wearable sensor devices, (2) developing communication-efficient FL platforms to facilitate federated health data training across a network of sensors and a computing server, and (3) designing privacy preservation techniques to enhance the security of data training against data attacks. The overarching aim of this research is to establish a robust health monitoring system characterized by efficient health data learning and advanced privacy protection, making it well-suited for real-world healthcare applications.

Student Research projects and Activities

Project 1: Federated on-device learning for intelligent mobile healthcare 

This project aims to propose an interesting federated on-device learning solution for intelligent health data analytics. Federated learning (FL) allows for training a shared powerful, intelligent machine learning (ML) model at a single computing server by using the training ability of distributed health devices. This innovative working concept eliminates the need for health data sharing for privacy preservation. 

The proposed project includes two main stages: 

The overview of federated on-device learning framework for intelligent health data analytics. Distributed devices collaborate with an edge server to build a strong ML model for privacy-preserved intelligent health data applications. 

Project 2: Blockchain for secure health monitoring  

This project aims to propose an interesting security framework using blockchain for secure health monitoring. Blockchain is a novel security platform operated by a decentralized data ledger with interesting security features, such as traceability and immutability. This project includes 2 main stages: (1) establishing the understanding of blockchain and its integration into health monitoring, and (2) developing a new blockchain simulator and its software using smart contracts to implement several cybersecurity tasks for health monitoring, including trustworthy health data collection and reliable health data sharing over mobile networks.

The students will be involved in the following activities:

Expected outcomes

Through the project, students will not only learn the theory and fundamental concepts of federated on-device learning in healthcare but also develop hands-on skills via FL algorithm deployment and software system designs. The skills learned from this project will be very useful for industrial job positions, such as software engineers and data scientists. Moreover, the outcome of this project is expected to be published at an international conference, which would be helpful for students’ future careers.

An overview of a blockchain framework for health monitoring, where blockchain is used to form a secure communication platform among mobile devices and cloud server. 

Required Qualification and Skills