Joint CMU-Pitt Ph.D. Program in Computational Biology Seminar

  • Remote Access - Zoom
  • Virtual Presentation - ET
  • Assistant Staff / Professor, Machine Learning and AI Group
  • Lerner Research Institute
  • Cleveland Clinic

Spatial transcriptome, single cell and digital pathology approaches to understand mechanisms of response and resistance to Immunotherapy and celltherapy: Opportunities and challenges for machine learning and AI scientists.

Advanced/relapsed/refractory cancer patients treated with Immune checkpointinhibitors (ICI) and Chimeric Antigen Receptor (CAR) T-cell therapy showed thatthere are subsets of patients with who have significant response. However, themolecular mechanisms of resistance and response to ICIs and cell therapy arelargely unknown. In our lab, we are routinely a) generate and analyze varioustypes of data including digital pathology, spatial transcriptome and singlecell RNA/Protein sequencing data from patients treated with ICIs and CAR-T therapyat real world settings, and actively b) developing and applying novel machinelearning (e.g., Deep Gaussian Process) and deep learning algorithms utilizingsuch data sets to develop clinically useful biomarker and prediction models.

In this talk, I will present our ongoing research studying digitalhistopathological whole slide images with matched spatial transcriptome data to1) understand spatial organization of Tumor Infiltrated Lymphocytes (TILs) andmicrosatellite instability (MSI)/tumor mutation burden(TMB) High tumors, 2)perform spatial transcriptome analysis of TILs and MSI-H tumor regions tobetter understand of ICI response in bladder and gastric cancers, 3) performingsingle cell analysis to identify biomarkers and find therapeutic strategies toimprove efficacy of CAR-T therapy. Our preliminary analyses demonstrate thatspatial organization and cellular heterogeneity of TILs and TMB/MSI-H tumorcells could provide a novel biomarker to predict ICI response, andcomprehensive single cell analyses could be used to find potential biomarkerand therapeutic strategies to improve CAR-T therapy. Finally, I will discussnew opportunities and challenges about how machine learning and deep learningcould provide a new foundation of spatial profiling, image and single cellanalysis for predicting immune/cell therapeutic responsiveness and newtherapeutic targets.

Tae Hyun Hwang is an assistant staff/professor at Cleveland Clinic where he leads machine learning and AI research for immuno-oncology/cell therapy and heart at Cleveland Clinic. He is a co-founding member and steering committee ofnewly established Cardiovascular Radiobiomics AI Center (CRAIC) at Heart, Vascular and Thoracic Institute at Cleveland Clinic where he is responsible to oversee, plan, perform AI and machine learning research based on molecular, image, and large EMR data. He received his PhD in Computer Science at the University  of Minnesota Twin Cities and was a research associate in the department of Computational Biology and Bioinformatics at Genentech, a tenure track faculty member at the University of Texas Southwestern Medical Center where he served as a Data Analysis Core and Bioinformatics Core co-director for Uthe TSW Kidney Cancer SPORE and NASA Specialized Center of Research,  and led a bioinformatics team for UTSW-MDACC Lung Cancer SPORE.  He is a co-founder of KURE.AI developing manufacturing advanced T and NK cell products by machine learningand AI technologies utilizing genomics, single cell and spatial transcriptomedata.

Faculty Host:  Russell Schwartz (CMU)

For More Information, Please Contact: