COVID-19 Diagnosis Algorithm Based on Convolutional Neural Networks

This study investigates an automated COVID-19 diagnosis system based on convolution networks, which aims to improve the effectiveness and accuracy of diagnosis. In a systematic analysis of 1709 chest X-rays, the research team used the ResNet model of deep learning for training and evaluation. Based on a detailed analysis of the confusion matrix generated by the model, the study found that the weighted average accuracy of the model and the recall rate reached 94,04% with an F1 value of up to 0,9403. This data shows better performance of convolutional neural networks in automated diagnosis. The results of the study show that neural network technology can effectively relieve the workload of doctors and provide diagnostic support for future public health events.