Factors contributing to long covid outcomes using Artificial Neural Network (ANN)

Background: The COVID-19 pandemic has resulted in a significant number of individuals experiencing long­term symptoms, known as long COVID. Understanding the factors contributing to long COVID outcomes is crucial for effective management and intervention. Objective: This study aimed to identify and rank the factors contributing to long COVID outcomes, develop an Artificial Neural Network (ANN) model, and evaluate its performance in predicting long COVID outcomes. Method: Data from a cross-sectional study by Moy et al. (2022) were used, including variables such as gender, age, BMI, smoking status, comorbidities, and severity of acute COVID-19. The Multilayer Perceptron (MLP) model in IBM SPSS Statistics 27 was employed for data analysis and ANN model development. Findings: Age, smoking status, severity of acute COVID-19, and heart disease emerged as significant factors associated with long COVID outcomes. The developed MLP model achieved an accuracy of 81.3% in predicting long COVID outcomes, with an area under the curve (AUC) of 0.753. Conclusion: This study provides valuable insights into the factors contributing to long COVID outcomes. Age, smoking status, and severity of acute COVID-19 were identified as key predictors of long-term effects. The developed ANN model offers a useful tool for healthcare professionals and policymakers in understanding and addressing the challenges oflong COVID. Further research is needed to explore additional variables and validate the findings in diverse populations.