SUPPORTING PERSONALIZING PALLIATIVE TREATMENTS FOR ESOPHAGOGASTRIC CANCER PATIENTS: AI-BASED APPROACH

Dr. ROB VERHOEVEN

Biography
Rob Verhoeven is trained as an epidemiologist at the Radboud university and received his PhD from the Erasmus university in 2012. Currently he is employed as senior research at both the Netherlands Comprehensive Cancer Organisation (IKNL) and the Amsterdam University Medical Centers (Amsterdam UMC). His research focusses on analyzing the real-world treatment and outcome of patients with esophageal or gastric cancer. By using innovative methodologies to analyze real-world data Rob intends to contribute to the academic knowledge of these methodologies. But at the same time, trying to create potential societal impact by presenting the output in such way that it can be used by physicians and patients to improve decision making and outcome in the daily clinical oncology practice. An overview of his scientific publications can be found here.

Dr. LAURA GENGA

Biography
Laura Genga received her Ph.D. degree in Science of Engineering at the Università Politecnica delle Marche, Italy, in 2016. She is an assistant professor in the Information Systems group at the Eindhoven University of Technology, the Netherlands. Her research revolves around data-driven process analysis and enhancement. Her core topics involve automated discovery and analysis of flexible processes, process compliance analysis, and on-line and monitoring and prediction to support process managers in taking decisions regarding current process executions, e.g., to react to potentially undesired situations. An overview of her scientific publications can be found here.

Abstract
Esophageal and gastric cancer are in the top ten most common cancers worldwide, both with high mortality rates. As the disease course is heterogeneous, personalized multidisciplinary palliative care to address the specific needs of individual patients is necessary. However, there is a lack of consensus about personalized palliative care options. This often leads to difficulties in determining the right treatment pathway for individual patients. Prediction models can aid in the discussion on personalizing treatment. Traditional regression-based prediction models for treatment outcomes are however static and are usually only aimed at primary treatment decisions shortly after diagnosis. To overcome limitations of previous models, we are developing innovative dynamic prediction models that can be applied at any time during the treatment process and take all recent information of the patient into account. In particular, we exploit techniques from the predictive process monitoring discipline. Given a set of historical patients’ care pathways, these techniques can be used to develop predictive models to provide a prediction on a current patient. We are especially interested in determining the remaining life-time at each stage of the patient’s pathway. This information will then be used to deliver recommendations aimed at tailoring the treatment selection to the patients’ characteristics. In this talk we will discuss the recent development and the current status of the project.