SMART DIAGNOSIS OF TUBERCULOSIS USING MACHINE LEARNING

Millions of new cases of tuberculosis (TB) are reported every year, and the disease is particularly impressive in low-resources countries where quick detection is sometimes difficult.TB is a major global health concern.In many remote areas, traditional clinical methods such as laboratory testing and radiography are unavailable, expensive and taking time.A web-based, machine learning-based approach is proposed for tuberculosis in this work to bridge this difference.Using 2,167 patient records with 14 clinical variables -such as weight loss, fever, night sweating, and a frequent cough -a decision tree classifier was trained as part of the process.A flask-based backend easily integrates the model with an responsible bootstrap-based user interface that provides real-time predictions that are available in all devices.The platform provides safe data storage using MySQL and dashboard for symptomatic trends and case allocation to increase access to patients and healthcare providers.According to the results, this scalable and user-friendly method can support healthcare providers, awarenessmaking, manufacturing, and helping healthcare providers to help in starting time intervention.Further studies may use more complex machine learning algorithms, more datasets and multimodal health data to increase accuracy and expand the use of systems for other respiratory disorders.