TinyML-Powered Electronic Nose for Non-Invasive Diabetes Screening Using Exhaled Breath
Supervisor Name
Liana Qabajeh
Supervisor Email
liana_tamimi@ppu.edu
University
Palestine Polytechnic University
Research field
Computer Science
Bio
Liana Khamis Qabajeh was born in Hebron, Palestine, in 1977. She received her B.Sc. degree in Computer Systems Engineering from Palestine Polytechnic University (PPU), Hebron, Palestine in 2000 and joined the Engineering and Technology Faculty, PPU, as a research and teaching assistant. In 2003 she was awarded a DAAD-scholarship to continue her education in Jordan. She received her M.Sc. in Computer Engineering from Jordan University of Science and Technology, Irbid, Jordan in 2005. Between 2005 and 2008 before pursuing her study, she was primarily involved in academic teaching and research in PPU. In 2008 she was awarded an IDB-scholarship to attain her Ph.D. degree in Malaysia. She obtained her Ph.D. in Computer Science from University of Malaya, Kuala Lumpur, Malaysia in 2012. In 2012 she has been appointed as assistant professor in Information Technology and Computer Engineering Faculty, PPU. She has been appointed as Master of Informatics program coordinator at PPU during 2016-2019. Recently, she has been granted associate professor title and appointed as dean of Information Technology and Computer Engineering Faculty at PPU. Her research interests include distributed systems and wireless networks. She is particularly interested in security issues and Quality-of-Service (QoS) in Ad-Hoc networks routing protocols.
Description
Diabetes is one of the fastest-growing chronic diseases, yet many people are diagnosed only after complications have set in. Although accurate diagnostic tests are available, conventional screening relies on blood testing, which can be invasive and requires access to healthcare facilities and laboratory services. Exhaled breath contains volatile organic compounds (VOCs), including acetone, whose patterns may be associated with metabolic changes related to diabetes. Electronic noses (e-noses) can capture these patterns using low-cost gas-sensor arrays. This project aims to develop a portable e-nose for preliminary diabetes screening by analyzing exhaled breath using machine learning and TinyML. In addition to classifying participants with and without diabetes, the study will investigate the relationship between gas-sensor readings and clinically measured glycated hemoglobin (HbA1c) levels, which reflect average blood glucose over approximately two to three months. Breath samples will be paired with relevant clinical reference data, subject to clinical collaboration and ethical approval. The project will also include a web-based dashboard to display sensor readings and model outputs.