<![CDATA[Tuberculosis (TB) remains a leading cause of death from infectious diseases worldwide, with Indonesia ranking second. Conventional TB diagnosis, which relies on the manual interpretation of chest X-rays, faces challenges due to a shortage of radiologists, particularly in resource-limited healthcare settings. This study aimed to develop a TB image classification model using Google Teachable Machine a no-code machine learning platform to create an easily deployable screening tool that requires no programming expertise. The model was developed using a transfer learning approach based on the Mobile Net architecture, trained on a dataset of chest X-rays labeled as either TB-positive or normal, and evaluated for accuracy across multiple training epochs. The results showed that both training and testing accuracies converged toward 1.0 after the 10th epoch; the curves moved in parallel without significant performance discrepancies, indicating good model generalization capabilities. Optimal results were achieved with an epoch count of 60, a batch size of 64, and a learning rate of 0.003. These findings demonstrate that the ease of implementation offered by no-code platforms does not necessarily compromise classification performance, making this a potentially practical solution to support early TB detection. Further research is recommended to conduct external validation and assess sensitivity and specificity to confirm the model's suitability as a clinical pre-screening instrument.]]>
Copyrights © 2026
