Master thesis: Application of AI and Raman Spectroscopy for non-invasive diabetes diagnosis

Diabetes remains a significant global health challenge, characterized by its increasing prevalence and the discomfort associated with conventional invasive diagnostic methods. The necessity for patient-friendly, non-invasive diagnostic solutions is paramount to improve early detection and patient compliance. This study addresses this critical need by exploring the integration of Artificial Intelligence (AI) and Raman Spectroscopy. This innovative approach aims to develop a reliable and efficient non-invasive diagnostic tool, leveraging the transformative potential of AI-driven medical technologies to revolutionize diabetes detection and management, ultimately enhancing patient care and reducing the burden of current diagnostic procedures.

Nghiên cứu sinh, nhà khoa học và chuyên gia trong lĩnh vực trí tuệ nhân tạo, quang phổ học, kỹ thuật y sinh, chẩn đoán y tế, cũng như các bác sĩ lâm sàng quan tâm đến phương pháp chẩn đoán đái tháo đường không xâm lấn.

This research comprehensively investigates the application of Artificial Intelligence (AI) and Raman Spectroscopy to establish a robust non-invasive diagnostic method for diabetes. Recognizing the limitations and patient discomfort associated with traditional invasive diagnostic techniques, the study proposes an integrated approach designed to simplify detection and enhance patient experience. The methodology involved a thorough literature review, selection of appropriate AI algorithms such as Support Vector Machines (SVM), ExtraTreesClassifier, and 1D-Convolutional Neural Networks (1D-CNN), and meticulous data acquisition using Raman Spectroscopy. Furthermore, the study detailed data preprocessing steps, including fluorescence background subtraction, polynomial fitting for baseline determination, and data denoising, to ensure data quality for analysis. Machine learning models were subsequently employed to analyze biochemical changes, facilitating the identification of specific diabetes biomarkers. The findings demonstrate promising potential, indicating that the synergistic integration of AI and Raman Spectroscopy can accurately identify diabetic conditions without invasive procedures. This innovative diagnostic tool offers a significant advancement towards patient-friendly diabetes detection, with profound implications for improving early intervention strategies, streamlining healthcare processes, and ultimately alleviating the burden of this chronic disease on both individuals and healthcare systems.