Indonesia’s healthcare system remains challenged by tuberculosis (TB), with estimated deaths reaching 125,000 in 2024. Recognizing the urgency, a team of scientists from the Department of Computer Science and Electronics (DIKE), Faculty of Mathematics and Natural Sciences at Universitas Gadjah Mada (FMIPA UGM) has developed TBScreen.AI, Indonesia’s first AI-based TB screening application, accessible via http://tbscreen.ai.
Team member Dr. Wahyono noted that this initiative aligns with government efforts to tackle TB by applying research and technology innovations, including the advancement of computer-aided diagnosis (CAD) using artificial intelligence (AI).
“This development is intended to support the screening process. The World Health Organization (WHO) has recommended the use of CAD technology to assist in interpreting chest X-rays,” he said on the UGM campus on Friday, Aug. 8, 2025.
According to Dr. Wahyono, the application is designed for both healthcare professionals and the general public.
Its user-friendly approach allows users to simply upload a chest X-ray image, which the system then automatically analyzes and presents a percentage likelihood of TB indication.
However, Dr. Wahyono emphasized that this output is not a definitive diagnosis; it is an initial screen requiring follow-up by a physician for a valid diagnosis.
This application is part of a research project funded by the KONEKSI program, an initiative of Australia’s Department of Foreign Affairs and Trade (DFAT).
The project is led by Dr. Antonia Morita I. Saktiawati, from the UGM Faculty of Medicine, Public Health, and Nursing (FK-KMK UGM), serving as Principal Investigator for the KONEKSI X-ray AI TB Project, with Dr. Wahyono as AI Team Coordinator.
It involves cross-institutional collaboration with the University of Melbourne, Monash University Indonesia, Universitas Sebelas Maret, YAKKUM Rehabilitation Center, Center for Advocacy for Women, Disabled and Children (SAPDA), and Papuan Health and Community Development Foundation (PHCDF).
Dr. Wahyono explained that the project was initiated using data from Sardjito Hospital in Yogyakarta.
The team ensured data validity through validation by the clinical team led by Dr. Antonia Morita I. Saktiawati, and a radiology team headed by Professor Lina Choridah.
Following validation, the team developed the AI model using digital image processing, computer vision, and machine learning.
The dataset, comprising approximately 936 chest X-ray images, was split into training data for model development and validation data to assess model accuracy.
“We currently achieve around 64% validity, and we are awaiting additional data from the Mimika Regional Public Hospital (RSUD Mimika),” Dr. Wahyono said.
The app’s core feature is its automated analysis of chest X-ray images, delivering a probability score ranging from 0 to 100% indicating TB likelihood.
While fully accessible to healthcare professionals, the public version offers limited features.
“The development team also includes functionality to collect additional dataset inputs to diversify the data and enhance AI accuracy,” he added.
At present, TBScreen.AI has been released in a limited capacity.
The team conducted outreach with healthcare workers at Balkesmas Klaten on Aug. 2 and RSUD Mimika on Aug. 7, 2025.
These sites will serve as pilot locations.
“This limited release aims to gather feedback from healthcare professionals in the two locations to refine the AI application. We hope for a wider release by year’s end,” Dr. Wahyono explained.
The introduction of TBScreen.AI is expected to accelerate active TB detection, especially in hard-to-reach areas and facilities lacking doctors.
This innovation is a tangible contribution toward achieving the Sustainable Development Goals (SDGs).
Author: Lintang Andwyna
Editor: Gusti Grehenson
Post-editor: Lintang Andwyna
Photographs: TBScreen.AI Research Team