Institute of Information Science, Academia Sinica



Press Ctrl+P to print from browser

TIGP (BIO)—AI Roadmap in Biomedical Research: From Images to Omics


TIGP (BIO)—AI Roadmap in Biomedical Research: From Images to Omics

  • LecturerDr. Nam Nhut Phan (Centers of Genomic and Precision Medicine, National Taiwan University)
    Host: TIGP (BIO)
  • Time2022-04-21 (Thu.) 14:00 – 16:00
  • LocationVirtual only Link: Please See Abstract Page
Live Stream

Meeting link:【webex

Meeting ID:2512 265 7590



Last 5 years have witnessed the transition of Artificial Intelligence (AI) from academic research to industrial applications with incredible speed. The availability of the pre-trained models together with cloud computing services such as the Amazon Web Service, Microsoft Azure, Taiwan Computing Cloud (TWCC) have allowed the application of AI models as never before. With the availability of plethora of public datasets and in-house data, various AI models have been already developed for purposes of classification, object detection, and outcome prediction, just to name a few. It is important to emphasize that application of deep-learning (DL) on commonly available next generation sequencing (NGS) data and whole slide pathological images, have allowed us to dig deeper into tasks and accomplish those that are unattainable by expert pathologists. I will primarily focus on research topics that deals with application of AI into biomedical research, spanning across biomedical images to multiple omics data, alone or integrated, for prediction of diagnosis and prognosis of cancer patients. Another aspect to be emphasized is the interpretability of the model prediction,not only on model itself but alsothat is justifiable by domain experts. For instance, DL can successfully predict cancer patients’recurrence status by using onlypathological images, however, without theinvolvement of pathologistsfor interpretation, the final results lackthereliability and therefore, it is difficult to convince theend userswith the predicted results, which are patients.

To summarize, AI and DL strategies have worked wonders, however, domain expertise is an important factor that needs to be included for justifying the model prediction towards its application to real clinical practice. 


Dr. Nam Nhut Phan obtained his doctoral degree (Ph.D) in Bioinformatics in 2022 from Taiwan International Graduate Program which is jointly operated by the Academia Sinica and National Taiwan University, Taipei, Taiwan. Dr. Phan is currently a postdoctoral research fellow at the Center of Genomic and Precision Medicine, National Taiwan University. Dr. Phan is also a research fellow at Nguyen Tat Thanh University, Vietnam since 2017.

With the vision to applying bioinformatics tools, particularly deep learning (DL) modelling, to biomedical research problems, Dr. Phan aims to develop high performing DL models with ultimate goal achieve fast and cost-effective prediction tools. Leveraging the big data in biomedical field such as pathological images and multiple omics data such as genomic data (DNA), transcriptomic data (RNA levels), proteomic data (protein level), and metabolomic data (metabolite level), his primary research theme focuses on employing DL for early cancer diagnosis and prognosis. Using these data types, either single omic or integrating multiple omics, along with cutting edge technology such as DL with state-of-the-art performance, not only reduce the workload but also improves the precision of the prediction by human experts. Furthermore, deploying these DL models into stand-alone graphic user interface is also another direction that Dr. Phan will be involved in his future research endeavors.