The Faculty of Medicine Siriraj Hospital at Mahidol University has developed an artificial intelligence (AI) programme to interpret lung and chest X-rays with accuracy above 95%, comparable to specialist radiologists. The system has already been used in more than 500,000 real patient cases. It can deliver results in under 10 seconds per case, significantly speeding up diagnosis.

Get today's headlines by email

The innovation was created by Siriraj’s Department of Radiology and was first developed during the Covid-19 pandemic to manage high patient volumes. Work on the system has continued beyond the pandemic, refining its performance and expanding its clinical use. The AI has been certified by the Royal College of Radiologists of Thailand as achieving accuracy above 95%, on par with expert radiologists.

Siriraj says the technology’s key strength is its speed, enabling rapid interpretation while maintaining high diagnostic standards. The system has already been transferred to more than 145 hospitals under the Ministry of Public Health nationwide. The hospital aims to expand deployment to 450 hospitals in 2026, with the goal of reducing inequality in access to quality healthcare across Thailand.

The move is intended to support hospitals with limited specialist resources, particularly in regional and rural areas. By enabling faster and reliable readings of chest X-rays, the AI system is expected to ease workloads for radiologists and improve turnaround times for patients. Officials say the expansion forms part of a broader effort to strengthen public health infrastructure through technology.

Siriraj is also extending its AI development into other diagnostic fields. Current plans include building AI systems to interpret brain CT scans and to support breast cancer screening using digital mammography. The hospital is collaborating with specialised private-sector partners to develop AI that better understands breast anatomy among Asian women.

The Nation reported that this collaboration aims to enhance screening accuracy and strengthen early detection of major diseases. Siriraj says the broader objective is to improve health outcomes for Thai people on a national scale. Further rollout and development are expected to continue throughout 2026.

Key Takeaways

• Siriraj’s AI interprets chest X-rays with over 95% accuracy and reports results in under 10 seconds.

• The system has been used in more than 500,000 cases and transferred to 145 public hospitals.

• Expansion to 450 hospitals is planned for 2026, alongside new AI for brain CT and breast cancer screening.

Join the discussion?

Already a member?

Adapted by ASEAN Now Nation 22 Feb 2026

View full record

This is quite impressive, a locally developed system deployed in over 145 public hospitals and reportedly used in 500,000 real cases.

The Royal College of Radiologists of Thailand certifying accuracy above 95% must certainly give it professional credibility here.

However, internationally AI in radiology is generally regarded as assistive, not autonomous. A 10-second result may be great for throughput and for flagging obvious pathology, especially in provincial hospitals with limited specialists, but it must not replace the experienced radiologist in suspicious cases.

The planned rollout to 450 hospitals in 2026 is certainly ambitious. However, I think its real value will depend on how well it complements rather than replaces human expertise.

I strongly suspected that radiologists would be the first specialist doctors to lose their jobs to AI. It is just a matter of time before AI hits the 97/98% mark in detection. At first there will be resistance, but the economic case is absolutely overwhelming.

The annihilation of middle-class jobs has just begun.

On 2/22/2026 at 5:22 AM, Georgealbert said:

The Faculty of Medicine Siriraj Hospital at Mahidol University has developed an artificial intelligence (AI) programme to interpret lung and chest X-rays with accuracy above 95%, comparable to specialist radiologists.

So 5 people out of a hundred get an incorrect diagnosis be it from a specialist or AI. Not sure if either of these methods instills confidence.

I read somewhere that a lot of the medical AI advice is crap because they give much data they've scraped from YouTube videos as medical fact. One would thing the medical LLMs would only use medical journals but I guess they didn't.

If the 95% is accurate then this is a good thing as most human doctors don't achieve that.