AI in the early detection of TB among people living with HIV

Overview

This AI-based tool uses deep learning, specifically Convolutional Neural Networks (CNNs), to detect tuberculosis from chest X-ray images among people living with HIV. By supporting accurate and timely diagnosis, the model helps address a critical diagnostic challenge in high-burden settings. The algorithm achieved 92% accuracy, demonstrating strong and consistent performance in identifying TB cases and supporting improved clinical decision-making.

Responsible AI Practices

The model was trained on a diverse dataset of approximately 3,000 patients from public health clinics in Tanzania, promoting fairness and representativeness. To support explainability, a heat map was developed to visually show which parts of the input data influenced the model’s decisions