Vietnamese student uses AI to digitize doctors' cryptic handwriting, wins Amazon funding

A Vietnamese high school student and his friends have developed an AI system that can decipher doctors' handwriting and digitize medical records, with Amazon providing US$35,000 to support its further development. Nguyen Hoang Gia Minh, a 12th-grade student at Newton Interdisciplinary School in Hanoi, and his friends developed SmileScan, a system that automatically digitizes medical records for Operation Smile Vietnam, a nongovernmental organization that provides free surgery to children and teenagers with congenital facial deformities such as cleft lips and cleft palates.The system uses a phone camera to scan and recognize handwriting on medical records, automatically fill in data fields and store the information. The process takes about 20 seconds per record. It can also generate statistical charts showing the number of patients treated and the most common locations of damaged teeth. Nguyen Hoang Gia Minh in photo provided by his family. "Our idea is to help doctors and medical staff reduce their workload and easily track treatment progress without having to spend time searching for records," he said.Minh said he had participated in volunteer activities since ninth grade, initially helping organize mobile clinics for Operation Smile Vietnam. Each dental checkup lasts four days and serves 2,500-4,000 children. For surgical programs, 130-150 children are selected from around 270 medical records.After observing the process, Minh noticed that doctors and medical staff still stored records on paper or entered medical information into Excel spreadsheets."This makes it difficult for the organization, doctors and volunteers to compile statistics, assess medical histories and prepare related medical reports," he said.In September last year, Minh and his friends began developing an AI-powered application that uses a large language model to scan and recognize handwriting and automatically enter the information into a database."Each doctor has their own way of writing and uses a lot of medical terminology, so without medical knowledge, you cannot give accurate prompts," Minh said, giving the examples that "apply veni" means "apply Vecni fluor" (or Flour Varnish)," while the letter "F" means "apply fluoride."The biggest challenge, he said, was teaching the system to recognize "doctors' handwriting" and specialized medical terminology.To address the problem, Minh and his friends compiled a list of terms commonly used by doctors and photographed nearly 2,500 sets of patient records to create training data."Our goal is for anyone, including ordinary logistics volunteers without medical expertise, to be able to use the system for data entry," he said.After multiple improvements and support from experts, the system achieved 90% accuracy. Its performance was verified when data from dental checkups in Quang Ngai and Hue in May were entered into the system. The project later won an outstanding presentation award at a young researchers' scientific conference in Japan.Operation Smile Vietnam said on its website that the project had digitized around 5,000 handwritten records and praised the students' efforts to apply technology to practical challenges in community healthcare.The system will continue to be developed and incorporated into Operation Smile Vietnam's digitization projects over the next two years, with US$35,000 in funding from Amazon. Nguyen Hoang Gia Minh at the International Young Researchers' Conference 2026 held in Japan in June 2026. Photo courtesy of his family In the future, Minh and his friends plan to tackle the digitization of surgical records."These records contain a lot of information and go through multiple processes, from admission and screening to diagnosis, so consolidating the data into a single system is a major challenge," Minh said.He hopes to expand the system to support multiple languages so it can be used across dental screening programs in other countries.