BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260421T090513Z
LOCATION:Bldg. 6 - Room 104
DTSTART;TZID=Europe/Stockholm:20260701T090000
DTEND;TZID=Europe/Stockholm:20260701T110000
UID:submissions.pasc-conference.org_PASC26_sess154_msa169@linklings.com
SUMMARY:Computational Prediction of Anti-Cancer Drug Response in Preclinic
 al Cancer Models Using Deep Learning and High-Performance Computing
DESCRIPTION:Yitan Zhu, Justin Wozniak, Alexander Partin, and Thomas Bretti
 n (Argonne National Laboratory) and Rick Stevens (Argonne National Laborat
 ory, The University of Chicago)\n\nCancer is a complex and heterogeneous d
 isease. Tumors of the same histological type can respond differently to th
 e same anti-cancer therapy. Therefore, accurate prediction of anti-cancer 
 drug response is of paramount importance for both patient treatment design
  and therapeutic development. We have developed computational Drug Respons
 e Prediction (DRP) approaches for preclinical cancer models using Deep Lea
 rning (DL) and High-Performance Computing (HPC). In this talk, we will fir
 st introduce several of our technical advances in building DRP models usin
 g diverse data sources, DL methods, and feature-engineering approaches. We
  will then present our project, Innovative Methodologies and New Data for 
 Predictive Oncology Model Evaluation (IMPROVE), which establishes a compre
 hensive benchmarking framework for evaluating and comparing DRP models. To
  build trustworthy and robust AI systems for DRP, we examine model general
 izability across drug-screening studies and across different subpopulation
 s defined by drugs and cancer cases. Model construction and evaluation tas
 ks, such as cross-validation and hyperparameter optimization, require larg
 e-scale computing. To address this need, we have implemented pipelines tha
 t parallelize computations on HPC systems by leveraging HPC software platf
 orms such as the CANcer Distributed Learning Environment (CANDLE), which w
 e developed to support large-scale artificial intelligence workloads for c
 ancer research.\n\nDomain: Applied Social Sciences and Humanities, Life Sc
 iences, Computational Methods and Applied Mathematics\n\nSession Chairs: J
 ustin M. Wozniak (Argonne National Laboratory, University of Chicago); Tho
 mas Brettin (Argonne National Laboratory, University of Chicago); and Eric
  Stahlberg (MD Anderson)\n\n
END:VEVENT
END:VCALENDAR