BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:America/New_York
X-WR-TIMEZONE:America/New_York
BEGIN:VEVENT
UID:941@fds.yale.edu
DTSTART;TZID=America/New_York:20260220T150000
DTEND;TZID=America/New_York:20260220T160000
DTSTAMP:20260212T163025Z
URL:https://fds.yale.edu/events/fds-seminar-greg-holste-ut-austin-bridging
 -the-gap-building-medical-ai-that-meets-clinical-reality/
SUMMARY:FDS Seminar: Greg Holste (UT Austin)\, "Bridging the Gap: Building 
 Medical AI that Meets Clinical Reality"
DESCRIPTION:\nZoom Password: 123\n\n\n\nAbstract: Artificial intelligence (
 AI) has enabled breakthroughs in the analysis of medical images\, electron
 ic health records\, and diverse sources of health data. However\, most med
 ical AI systems can only answer narrow questions like “Does this patient
  have based on this single image?” This framing fails to capture how med
 icine is actually practiced. Clinicians integrate information across modal
 ities\, reason over noisy longitudinal data\, and assess the presence and 
 future risk of multiple conditions at once—our AI models should too. In 
 this talk\, I present efforts to bridge this gap between academic AI and c
 linical reality with contributions spanning cardiology\, ophthalmology\, a
 nd radiology.\n\n\n\nI will start by presenting my work building AI system
 s to interpret echocardiograms\, the most widely used cardiac imaging tool
 . I will trace the evolution from single-view aortic stenosis detection to
  PanEcho\, a multi-task model that fully automates echocardiogram interpre
 tation by integrating multiple views like a cardiologist. Next\, I address
  the temporal nature of progressive eye disease monitoring with an approac
 h for disease prognosis based on irregularly-spaced longitudinal imaging\,
  matching the clinical workflow. I will then discuss the long tail of abno
 rmal findings on chest X-ray and the promise of vision-language models for
  zero-shot rare disease detection. I conclude with a vision for multimodal
  health monitoring\, AI-driven clinical knowledge discovery\, and prospect
 ive trials of human-AI collaboration to understand how these systems can l
 ift ongoing burdens on the healthcare system and improve patient outcomes.
 \n\n\n\nBio: Greg Holste is a PhD Candidate in Electrical and Computer Eng
 ineering at The University of Texas at Austin\, NSF Graduate Research Fell
 ow\, and MIT Tech Review 35 under 35 semifinalist. His research bridges ar
 tificial intelligence (AI) and healthcare\, developing deep learning metho
 ds to solve clinical decision-making and medical image analysis problems. 
 Greg collaborates with clinicians to develop AI systems that help physicia
 ns interpret complex health data more quickly and accurately with innovati
 ons in multimodal\, self-supervised\, and long-tailed learning. His resear
 ch has been published in journals such as JAMA\, JAMA Cardiology\, Nature 
 Communications\, The Lancet Digital Health\, among others\, and presented 
 at top conferences including CVPR\, ICML\, ICCV\, and MICCAI.\n
CATEGORIES:FDS Events,Postdoctoral Applicants
END:VEVENT
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:STANDARD
DTSTART:20251102T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
END:STANDARD
END:VTIMEZONE
END:VCALENDAR