A retinal photograph, taken to check a patient with diabetes for eye disease. A CT scan of the head, taken to look for a bleed. A chest X-ray, taken to catch tuberculosis. Clinics across Asia-Pacific were already capturing these. Software is now reading them as well, and the change is easy to see in a working day: a result in about three minutes instead of an hour, a radiologist alerted in about six minutes, a public health scheme willing to pay for the reading, an X-ray unit small enough to take into a community.

That is where this is heading. The routine case starts to need less of a person's time. The urgent case reaches a clinician sooner. In Thailand and South Korea, the government is paying for the software as part of care, which is how a tool leaves the pilot stage.

Singapore is getting to the urgent case faster

In Singapore, the diabetic-eye programme is the place this is already ordinary.

SELENA+, the Singapore Eye Lesion Analyser, reads those retinal photographs for diabetic retinopathy, glaucoma and age-related macular degeneration. Lianhe Zaobao, writing on 31 August 2026 about the national research report, says the technology was designed with the Singapore Eye Research Institute and the National University of Singapore, and that the local company EyRIS makes the system. Zaobao puts it in the national programme since 2020, across 23 polyclinics, screening 120,000 to 200,000 diabetic patients a year.

A clinic would notice the clock. The software returns a reading in about three minutes. Doing the same photograph by hand had taken about an hour. Zaobao says a person still reviews the result, and describes what comes next: cases that do not need a referral could be filed on their own, about half the workload, while the photographs that might show disease stay with a person. The grader's day shifts toward those.

The urgent version is a head scan. Healthcare Asia reported in September 2026 that Singapore General Hospital, Synapxe and Iota MedTech have iScreen, which reads a head CT for a bleed and alerts the radiologist. It is the first head CT tool on AimSG, the public imaging platform running across Singapore's three public hospital clusters since 2023. Healthcare Asia says the software's result is ready in about six minutes, where an urgent report had been taking half an hour to an hour, and that a radiologist still checks it before the report goes out. When the scan does show an acute bleed, the report is coming back within 30 minutes, from a previous 30 to 60. The hospital does about 50 of these scans a day, and about 10% show that bleed. Synapxe has also put a fracture-reading tool into the emergency department at Woodlands Health, and has said a tuberculosis tool on the same platform will go into the National Tuberculosis Screening Centre.

The medical professional still has to review the images. What is different is that they should see the urgent case sooner.

Japan is asking how many doctors still have to look

For years, Japan's local screening for lung, stomach and breast cancer has put two doctors on every film, so a miss by one still meets a second pair of eyes. On 23 March 2026 an expert panel at the Ministry of Health, Labour and Welfare approved a trial that lets the software look first. The Asahi Shimbun reported the shape of it the next day. If the software flags something, one doctor can decide whether the person needs a closer exam. If the software sees nothing of concern, both doctors still read the film. The images are chest X-rays, gastric X-rays and mammograms. The ministry has not yet said when the trial starts or how it will be run, so there is no count of patients.

Kyoto has started on a smaller piece of the same question. On 8 June 2026 the Kyoto Medical Association, with Preferred Networks and PSP, said a chest X-ray aid would go into the prefecture's lung-cancer screening in fiscal 2026, after a try-out in Uji the year before. When the software thinks an abnormality is unlikely, the screen shows "No Findings". Both doctors still read, and the diagnosis stays with them.

Kyoto's film is a chest X-ray. Several other governments are now putting software on that same kind of film, mostly to find tuberculosis, and they are backing it with public money and with machines that can leave the hospital.

Tuberculosis screening is where the spending has started

Since 2021, the World Health Organization has said that software may read a digital chest X-ray in place of a person, when the task is screening people aged 15 and older for pulmonary tuberculosis. A health service still has to be satisfied the software is up to the job, and a positive screen still goes on to a proper test. India, Thailand, the Philippines and South Korea are each using that opening differently.

India asked what the reading is worth. The Department of Health Research commissioned the Indian Institute of Public Health Gandhinagar to set two tools, Qure.ai's qXR and Deeptek's Genki, against a radiologist reading a digital chest X-ray. Frontiers in Digital Health published the study, using 2023 screening from Maharashtra and Tamil Nadu. qXR came out about INR 9,865 cheaper per case interpreted, so in the model's terms it saves money. Genki came out about INR 11,287 more per case, which the study still counted as cost-effective. Separately, the institute's brochure, drawing on other studies, put the software's sensitivity at 90% and its specificity at 68%, and said that pair meets the World Health Organization's standard for this kind of screen. Most suspicious films get caught. A large share of healthy ones still go on to a person. The health system would be paying less per film, and living with that second number.

Thailand has gone ahead and put the reading on the public tab. On 7 July 2025 the board of the National Health Security Office made an AI reading of the chest X-ray a benefit under the universal coverage scheme, the gold card. The tool comes from Siriraj Hospital at Mahidol University and the company Perceptra. Thai regulators and the Royal College of Radiologists of Thailand had already cleared it, and it has been on the national innovation list since December 2023. The first allocation is up to THB 55 million, for 167 public hospitals in fiscal 2025. Later years can add hospitals as the service can absorb them. In its English note, the office talks about a later annual figure of THB 225 million by 2027, and about using the tool to pick up tuberculosis and lung cancer earlier while taking some of the reading off clinicians.

In the Philippines, the machine itself moves. On 12 November 2025 the World Health Organization's office there quoted Health Secretary Teodoro Herbosa: the country is already using ultra-portable, AI-powered chest X-rays, together with rapid molecular tests, in the national tuberculosis plan. The page shows health workers carrying those units in the Cordillera. The plan aims to screen 12 million people in 2026. That figure is the ambition for the tuberculosis programme as a whole. On the ground, the X-ray unit and the software travel together, so a screen can start in a community.

South Korea is paying for the software in hospitals outside Seoul. Asia Business Daily reported in June 2026 that the Ministry of Health and Welfare had selected Lunit to supply chest X-ray and mammography tools to six regional hospitals: Dankook University Hospital, Sejong Chungnam National University Hospital, Ulsan University Hospital, Jeju National University Hospital, Jeonbuk National University Hospital, and Hwasun Chonnam National University Hospital. The ministry is paying a commercial company to put that software into care away from the capital. What those six hospitals will do with it, day to day, is not in the report yet.

Software is now sitting on images clinics were already going to take. The difference is in the working day around them: a result in minutes, a public insurer willing to pay for the reading, a chest X-ray that can start in a community. The clinician still reviews. They should meet the urgent case sooner, and spend less of the day on the case that can wait. Filing a routine eye result without a second pass, and testing whether one doctor is enough once a film is flagged, are the next moves. Public payment is what makes either of those ordinary care. The tuberculosis chest X-ray has gone furthest because the film, the screening programme and the World Health Organization's rule for a software reader were already in place.

What to watch

• Whether AI starts changing staffing requirements. A faster queue can still need the same number of people. The shift is a health system that can show fewer readers on the routine case, or a different job for the people who remain.

• Whether the workload saved is published. Minutes off one scan are a clock. A saved workload is a count of cases that no longer need a second look, and a record of what the clinic did with that time.

• Whether the economics hold outside high-volume screening. A cheaper assisted read is easiest to show when thousands of similar films go through one programme. The harder test is a hospital that sees fewer cases, and more difficult ones.

• Whether the chest X-ray becomes the entry point to a broader AI imaging layer. It is the scan that already has a screening programme, a payment decision and a rule for a software reader. The tell is the same buyer extending that arrangement to other images.