Title : Leveraging Artificial Intelligence and Biosensors for User-Friendly Detection of Urological Cancers
일시 : 2024년 3월 26일(화), 17시
Speaker : : Kwan Hyi Lee (KIST)
Abstract : Addressing the imperative for a more patient-centric and precise screening modality for urological cancers, we have amalgamated the prowess of artificial intelligence (AI) and biosensors. Our methodology integrates an electrochemical biosensor with sophisticated AI algorithms, enabling the analysis of signals derived from multiple biomarkers present in patient urine samples. This comprehensive approach encompasses the utilization of both conventional machine learning techniques and advanced deep learning algorithms. Through an extensive comparative study encompassing various AI algorithms, we have discerned the pivotal factors influencing the accuracy and clinical significance of our diagnostic framework. Notably, as we expanded the number of biomarkers up to four, most algorithms consistently demonstrated enhanced screening performance. Employing the optimal biomarker combination, our machine learning algorithms achieved a remarkable accuracy exceeding 96% in identifying prostate cancer patients, leveraging the analysis of 216 urine samples. This innovative integration, which marries a multi-marker urinary biosensor with AI analysis, exhibits significant promise as a pivotal strategy for precise cancer screening. The methodology offers the added advantage of simplicity, as it necessitates only the non-invasive collection of bodily fluid samples, thereby enhancing user-friendliness and accessibility in clinical settings.