Artigo - Atena Editora

Artigo

Baixe agora

Livros

SURVIVAL ANALYSIS APPLYING COX MODEL AND MACHINE LEARNING TO COVID-19 DATA IN THE CITY OF BUCARAMANGA BETWEEN MARCH 2020 TO MARCH 2023

Este estudio realiza un análisis comparativo del rendimiento de las técnicas de Machine Learning, redes neuronales y tradicionales de análisis de supervivencia. Las técnicas comparadas son el tradicional modelo de riesgos proporcionales de Cox (CPH), técnica de Machine Learning Random Survival Forest (RSF) y redes neuronales como DeepSurv. Estas técnicas se aplican para el estudio de casos de Covid-19 de los pacientes diagnosticados en la ciudad de Bucaramanga entre marzo del 2020 y marzo del 2023. Este estudio demuestra un mejor rendimiento se obtuvo con la técnica de Random Survival Fores en la predicción de la función de supervivencia medidos a través del C – índex, Brier score y AUC.
Ler mais

SURVIVAL ANALYSIS APPLYING COX MODEL AND MACHINE LEARNING TO COVID-19 DATA IN THE CITY OF BUCARAMANGA BETWEEN MARCH 2020 TO MARCH 2023

  • DOI: https://doi.org/10.22533/at.ed.15941224230110

  • Palavras-chave: Análisis de supervivencia; Covid-19; Random Survival Forest; riesgos proporcionales; Deepsurv.

  • Keywords: Survival analysis; Covid-19; Random Survival Forest; proportional hazards; Deepsurv.

  • Abstract:

    This study performs a comparative analysis of the performance of Machine Learning, neural networks and traditional survival analysis techniques. The techniques compared are the traditional Cox proportional hazards (CPH) model, Machine Learning Random Survival Forest (RSF) technique and neural networks such as DeepSurv. These techniques are applied to the study of Covid-19 cases of patients diagnosed in the city of Bucaramanga between March 2020 and March 2023. This study demonstrates better performance was obtained with the Random Survival Fores technique in predicting the survival function measured through the C – index, Brier score and AUC. 

  • Giann Axel Leguízamo Jordán
  • Emiliano Rodríguez Arango
Fale conosco Whatsapp