Data-Driven Oncology: Applying Machine Learning and Process Mining for the Optimization of the Oncology Patient Journey
Prof. Ricardo S. Santos
Instituto Tecnológico de Aeronáutica, Brazi
Abstract: The oncology patient journey is a complex and dynamic system involving multiple steps, actors, and variables that interact in multifaceted ways. This journey is influenced by both internal and external factors, and delays in diagnosis or in any other stage of treatment can significantly compromise patient outcomes.
Several studies demonstrate that delays in accessing care are associated with lower survival rates, increased treatment-related complications, and higher healthcare costs. Timely access to diagnosis and treatment is, therefore, a critical element of effective cancer control strategies.
In countries like Brazil, legislation mandates the initiation of cancer treatment within 60 days of diagnosis. However, despite this legal requirement, several challenges still hinder its effective implementation. Common obstacles include delays in diagnosis, extended waiting times for certain procedures, and insufficient coordination across different levels of care.
Evidence shows that a significant proportion of patients begin treatment beyond the legally mandated timeframe—and even among those who start on time, delays often occur at other stages along the care pathway. These inefficiencies reveal systemic bottlenecks that frequently go undetected by public health administrators.
In contrast to sectors such as manufacturing, which commonly leverage software tools to identify inefficiencies, oncology still lacks robust technological solutions tailored to this purpose.
Patient Journey Maps are an emerging concept that visualizes the patient’s path through the care continuum. When enhanced with machine learning and process mining techniques, these tools can help identify critical bottlenecks and inefficiencies early in the care process, offering valuable insights for both patient navigation and healthcare system optimization.
This lecture explores how machine learning and process mining can be applied to the development of software solutions designed to detect and predict bottlenecks in the oncology treatment workflow, ultimately aiming to improve the overall patient journey.
Brief Biography of the Speaker: Ricardo S. Santos is a highly skilled software solutions architect, research scientist, and systems developer with over 30 years of experience spanning both academic and industrial sectors — more than 20 of which have been dedicated to healthcare informatics. Throughout his career, he has designed and implemented over 25 software systems.
With more than 28 years of experience in teaching and research, he has lectured in both undergraduate and graduate programs, while also actively contributing to project development and student mentoring. His teaching portfolio includes courses such as Databases, Data Mining (Machine Learning), Business Intelligence, Systems Analysis and Development, Health Informatics, and Data Science Applied to Healthcare.
His areas of expertise encompass healthcare informatics, software development, data science, knowledge representation, clinical and oncology pathways, database systems, machine learning, and data and process mining.
He has published numerous scientific papers in journals and conferences and has contributed to 27 innovation assets, including patents, trade secrets, invention disclosures, and registered software.
Ricardo holds a postdoctoral fellowship from the Technological Institute of Aeronautics (ITA), a Ph.D. in Health Informatics from the Federal University of São Paulo (UNIFESP), as well as two master’s degrees — one in Biomedical Engineering and another in Information Technology — and a bachelor’s degree in Information Technology.
Currently, he serves as Technical Director at Compumedica Research & Innovation, a collaborating professor at ITA, and a professor at Municipal University of Sao Caetano do Sul (USCS), where he leads the Situational Awareness in Healthcare Research Group.