Prediction of Skin Tumor Invasiveness: A National Analysis Through Explainable Artificial Intelligence (XAI)

Abstract

In Brazil, skin tumors represents the type of neoplasm with the highest incidence rate among the population. Because of this, this study explores the invasiveness of this disease using computational techniques to understand how specific patient characteristics influence its progression. Through the analysis of data provided by the Cancer Hospital Registry (RHC) of the National Cancer Institute José Alencar Gomes da Silva (INCA), and with the aid of Artificial Intelligence (AI) algorithms explained by the SHapley Additive exPlanations (SHAP) approach, the study reveals that the invasiveness of skin cancer is affected in a significantly different way by the individual characteristics of patients compared to analyses based on more general attributes. These findings underline the importance of personalization in medicine, suggesting that a deeper understanding of individual characteristics can lead to more accurate diagnoses and more effective treatments. Furthermore, the research highlights the role of XAI in clarifying these relationships, pointing to the need for more refined approaches in prevention, treatment, and the formulation of public health policies aimed at combating skin tumors, despite limitations such as data imbalance encountered during the study.