The Tribhuvan University Teaching Hospital (TUTH), Maharajgunj, has begun trialling the artificial intelligence tool “ChestGuru” in its emergency chest X-ray system with the aim to improve diagnosis and care.

Doctors at the hospital said that the home-grown AI tool can detect 14 different ailments, including tuberculosis and pneumonia, at once through a single chest X-ray.

“This AI tool is helping doctors diagnose problems more easily. Even junior doctors can make accurate diagnoses without help from their seniors, who may not be available all the time,” Dr Ghanashyam Gurung, professor of radiology at TUTH, told the Post.

The AI system was developed by Nepali health-tech company ‘Second Eye’, initiated by engineering students. Doctors say the software can instantly read the chest X-ray and detect multiple chest diseases, including tuberculosis, chronic respiratory problems, pneumonia, and lung cancer, among others, within seconds.

“Sometimes even doctors can miss the problems and send patients home without a proper diagnosis,” said Gurung. “The use of AI helps reduce errors and ensures diagnosis.”

The system allows doctors to view images directly on laptops, so patients no longer need to carry fragile X-ray films. In overcrowded emergency rooms, ChestGuru helps teams prioritise patients with life-threatening conditions. By flagging urgent cases, the tool allows faster triage, reduces conflict in stressful environments, and improves patient safety.

Developers say ChestGuru fundamentally changes the care experience. Traditional workflows often require patients to stand in long queues to collect printed X-ray films and handwritten reports. This leads to delays in treatment, repeated travel, lost wages, and in many cases, complete loss to follow-up.

“This tool acts as a public health intelligence engine, and it offers strong operational and financial advantages, as health facilities can view imaging records without reliance on expensive high-bandwidth internet systems,” said Dr Latika Giri, founder of Second Eye who also works at TUTH. “This AI tool could be very useful in the peripheral health facilities, which often lack trained human resources.”

Developers claimed that healthcare workers can review AI’s interpretations, edit findings, add clinical context, and automatically generate standardised reports. These digital reports can be printed, shared across departments, or stored for future visits, allowing smoother referrals and reducing the longstanding problem of patients being lost to follow-up.

“With the AI tool we developed, report generation is possible on the same day, which ultimately helps doctors decide when to start treatment,” said Samyam Giri, an engineering student and one of the developers.

Before being used at TUTH, the system underwent scientific validation using real-world imaging from outpatient, emergency, inpatient, and intensive care units.

According to Giri, the results demonstrated sufficient safety and performance for frontline clinical use.

Developers said the interface supports the Nepali language, and this enabled nurses, health assistants, and medical officers to use the tool without linguistic barriers. Along with engineering students, medical students and clinicians from the TU Institute of Medicine were involved in designing the software, Giri said.