A patient in a remote town gets a chest X-ray to check for tuberculosis. The image is ready, but a specialist who can read it may be hours away. Waiting for a result can mean another trip to the clinic, another day off work and another chance to drop out of care.
Ateneo de Manila University researchers have examined whether AI-assisted X-ray reading could help rural health units shorten that gap. Their study also asks a practical question for public clinics: would it be affordable?
The researchers compared AI-assisted screening with radiologist or teleradiology readings using a model of 1,000 people with suspected TB each year. Under the study’s main assumptions, the AI approach cost about ₱877,000 annually, against ₱1.14 million for manual interpretation.
Much of the projected saving came from avoiding unnecessary follow-up tests. In the model, AI produced about 286 false-positive X-rays, compared with about 454 under manual reading. It also correctly flagged slightly more TB cases. Patients with a concerning X-ray would still need a confirmatory test; the software would screen for possible disease, not make the final diagnosis.
The findings offer a possible way to stretch scarce clinic resources in a country that accounted for an estimated 6.8 percent of global TB cases in 2024. But the figures come from a theoretical model, not a trial in rural health units. When the researchers changed key assumptions about reading fees or AI accuracy, the technology no longer consistently saved money.
Researchers Harold Henrison Chiu, Bryan Christopher Lao and Gloanne Adolor propose starting with targeted trials in underserved communities. Local results would show whether AI can deliver its most useful promise: helping a patient leave the clinic with a clear next step, rather than another wait.
