Artificial Intelligence in Process Modelling in Oncology

Carlos Fernandez-Llatas¹ Roberto Gatta^2* Fernando Seoane^{3, 4, 5, 6} Vincenzo Valentini⁷

• ¹ITACA-SABIEN Technologies for health and well-being, Polytechnic University of Valencia, Spain
• ²Department of Clinical and Experimental Sciences, University of Brescia, Italy
• ³Department of Clinical Science, Intervention and Technology, Karolinska Institutet (KI), Sweden
• ⁴Department of Textile Technology, Faculty of Textiles, Engineering and Business, University of Borås, Sweden
• ⁵Department of Clinical Physiology, Karolinska University Hospital, Sweden
• ⁶Department of Medical Technology, Karolinska University Hospital, Sweden
• ⁷Agostino Gemelli University Polyclinic (IRCCS), Italy

From the early days of process mining in healthcare, oncology has consistently been one of the most compelling application areas (Rojas et al. (2016)). The various oncological diseases, in fact, from the perspective of screening, diagnosis, treatment, and follow-up pathways, although sharing some common elements, also differ considerably depending on the anatomical district involved, while maintaining a generally well-structured practice, codified in consensus, protocols, and guidelines. In addition, the significant impact on patients and the high social costs contribute in making this sector one in which Data Analysis seeks to provide a meaningful contribution through the application of innovative techniques.Since in this context, it is particularly important to capture the temporal evolution of relevant factors, the application of Artificial Intelligence (AI) techniques for modeling clinical/healthcare processes can be a key tool in understanding what may play a significant role in disease control or the induction of iatrogenic events in the care pathway.In this collection, many applications covering different areas of Process-Modelling in the oncology theme have been explored.In Tozzi et al. (2022), by a systematic review of systematic reviews, an exploration of the current contributes of AI in Pediatric Oncology is given, in Europe in particular. A set of 34 reviews and 304 articles were considered, retrieved by querying the Web of Science platform. The number of original papers, relatively stable from 2004 to 2016, quadruples in 2018 and subsequently doubles compared to 2018 in 2020. This is interpreted as a sign of the growing interest in paradigms, methods and tools provided by AI and how, nowadays, it is seen as promising in the field of oncology. Notably, despite the considerable amount of retrieved papers, there was no evidence found for AI utilization in process mining, clinical pathway modeling, computer-interpreted guidelines to enhance healthcare processes. This indicates 22 that AI in oncology is probably just beginning, and there is room for various tools, like those focused on 23 process analysis, for proposing efficient solutions for many unmet needs in pediatric oncology.