SmileScan works by pointing a smartphone camera at a paper record. The image is sent to an AI model that recognises the handwriting, extracts the relevant fields and stores the information automatically, completing the whole process in about 20 seconds per record.

To teach the AI, the team first compiled a list of medical terms that doctors frequently use. They then photographed nearly 2,500 patient records, creating a large set of examples for the model to learn from.

After several rounds of improvement and expert guidance, the system reached 90% accuracy. The team confirmed this performance by entering data from dental check‑ups carried out in Quang Ngai and Hue in May, and comparing the results with the original paper files.

Operation Smile Vietnam reported that SmileScan has already digitised roughly 5,000 handwritten records, making it easier for doctors and volunteers to track treatment progress and generate statistics. The project will continue to grow with US$35,000 in funding from Amazon.