Machine Learning for risk prediction in thoracic oncologic surgery and comparison with traditional models

• Grant number: 22/12874-4
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2023
• End date: January 31, 2025
• Field of knowledge: Health Sciences - Medicine - Surgery
• Principal Investigator: Ricardo Mingarini Terra
• Grantee: Renata Matheus Faccioli
• Host Institution: Instituto do Coração Professor Euryclides de Jesus Zerbini (INCOR). Hospital das Clínicas da Faculdade de Medicina da USP (HCFMUSP). Secretaria da Saúde (São Paulo - Estado). São Paulo , SP, Brazil

Abstract

Introduction: Classic risk prediction models are currently used as auxiliary tools for decision-making thoracic surgery. The application of technologies such as machine learning combined with progressively more robust databases presents the opportunity to develop models with increasing predictive capability. Objectives: This study aims to develop a machine learning model to predict risk of mortality in 90 days after pulmonary resection and compare its performance to that of well-established models in thoracic surgery, used as benchmark models. In addition, its secondary objectives are to develop machine learning models to predict the occurrence of cardiopulmonary complications and overall survival. Methods: The study will use data from 1,850 patients with lung cancer submitted to curative resection, extracted from the Brazilian Registry for Surgical Treatment of Lung Cancer (RBCP). The performance of the classification models will be evaluated according to the AUC and the O:E ratio. Additionally, the global survival prediction model will be evaluated according to the index of agreement, IBS and temporal evolution of AUC. The results of the 90-day mortality prediction model will be compared to the results of the benchmark models, which will be evaluated with the same methodology.