An artificial intelligence system that detects infectious diseases in children

Accurate recognition and classification of images of pneumonia patients is critical to ensure diagnosis and treatment. Deep learning techniques such as Convolutional Neural Networks (CNN) showed an accuracy of 96.8% and CNN Long Term Memory (CNN LSTM) accuracy levels of 97% in this task. Models were evaluated using a dataset of 5856 classified images of pediatric pneumonia patients obtained from Kaggle. The data set was divided into training (70%), test (15%), and validation (15%) sets. Both CNN RF models achieved an accuracy of 92% and the CNN SVM an accuracy of 93% which contributes to useful resources for clinicians and radiologists. In addition, we explored the application of logical rules in our CNN FUZZY model, which showed results by making decisions based on probabilities and the accuracy was 96.5%. Feature extraction from images is a component of the modeling process that enables predictions based on patterns acquired with an optimizer. To improve performance, modifications were made to the Adams parameters. The evaluation of model performance included metrics such as recall, accuracy, and F1 score along with an evaluation of accuracy. These measures provide an assessment of the effectiveness of the models to help identify areas for improvement and serve as a reference for investigations. The high accuracy of the models demonstrates their ability to correctly classify pneumonia patients potentially enhancing patient outcomes. The evaluation scales used in this study provide an analysis of the strengths and weaknesses of the models emphasizing the importance of integrating these scales to ensure reliable and accurate performance, in real-world situations.