AI-Assisted Predictive Model for Tuberculosis Disease
Keywords: Tuberculosis, Artificial intelligence, Chest X-ray, Convolutional neural network, Machine Learning
Abstract: Tuberculosis (TB) is an infectious disease caused by the bacteria named Mycobacterium tuberculosis. It is of a major public health concern, causing millions of fatalities globally each year primarily in developing nations and among individuals with poor socioeconomic status. Effective disease management depends on early diagnosis and prompt treatment. This study is aimed to build an Artificial intelligence (AI)- assisted model that can be used for the prediction of TB based on radiological data from patients suspected to have TB with the aid of machine learning algorithms. This research also examines the potential of AI-assisted models in detecting TB at a level that will overcome the radiologist limitations. In addition, the possibility of AI-assisted predictive analysis method in improving the accuracy of TB diagnosis will be discussed. This study dataset contains 4200 Chest X-Ray (CXR) images of both TB infested and non-infected patients. In building this model, we will employ various transfer learning methods, including VGG16, VGG19, ResNet50, InceptionV3, EfficientNetB7, DenseNet201, MobileNet, Xception, AlexNet, and NASNet. These methods will be utilised to extract essential features for optimising the classification of TB within Convolutional Neural Networks (CNNs). The model was assessed based on Accuracy, precision, sensitivity, and F1-Score metrics. The result obtained indicates that the VGG16 model achieved the best result based on accuracy and recall values of 99.6% and 100% respectively.Conclusively, this study shows that AI may be used to predict TB, and it also emphasises the importance of feature selection and data preprocessing for model performance.
Submission Category: Machine learning algorithms
Submission Number: 30
Loading