Parts | Days | Selection | Search | Updates | Downloads | Help
AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz
AKPIK 3: AKPIK Postersession
AKPIK 3.6: Poster
Thursday, September 30, 2021, 13:30–15:30, P
Design and validation of a Digital Twin for prostate cancer from a physics point of view — •Carlos Andres Brandl1, Anna Nitschke1, and Matthias Weidemüller1,2 — 1Physikalisches Institut, Ruprecht-Karls Universität Heidelberg, Im Neuenheimer Feld 226, 69120 Heidelberg, Germany — 2National Laboratory for Physical Sciences at Microscale and Department of Modern Physics, and CAS Center for Excellence and Synergetic Innovation Center in Quantum Information and Quantum Physics, Shanghai Branch, University of Science and Technology of China, Shanghai 201315, China
Digital Twins (DT) are virtual representations of physical assets and e.g. promise improved decision making. DT can help to personalize healthcare for complex diseases like prostate cancer by combining large amount of clinical parameters and answering the questions of tumor risk, tumor stage and optimal treatment. In medical applications interpretability and uncertainty quantification are crucial.
Combining data-driven approaches with interpretable machine-learning models and evidence based clinical guidelines will lead to more reliable and confidential outcomes. Correlation analysis gives an insight to the data and enables to unravel the dependencies of important clinical parameters like prostate specific antigen (PSA) with others and determine their distributions in the Heidelberg patient cohort. The backpropagation of outcome errors to the found distributions of the input parameters with Bayesian methods allows to determine the impact of the input parameter uncertainties on the twin predictions and helps clinicians to interpret the results accordingly.