30-07-2025
17-year-old student develops AI-powered medical transcription tool
Sign up now: Get ST's newsletters delivered to your inbox
The TL;DR: As a kid, John Tow couldn't spend as much time as he wanted with his doctor parents due to their long working hours. This inspired him to develop SgScribe, an AI-powered medical transcription app, to help doctors cut down on admin work.
John Tow, a second-year ACS (Independent) student, coded SGscribe, an AI-powered medical transcription device that converts recordings of doctor's consultations to SOAP notes.
SINGAPORE – John Tow recalls being unable to spend as much time with his parents as he wanted when he was growing up .
His orthopaedic surgeon father and anaesthesiologist mother worked long hours seeing patients and performing surgery, with administrative work on top of that keeping them away from home.
This situation inspired John to come up with a product that would free doctors from such tasks and help them save time.
Over the course of two months, the Year 6 student at Anglo-Chinese School (Independent) student developed SgScribe, a medical transcription tool that records medical consultations and uses Artificial Intelligence (AI) to transcribe them into SOAP (Subjective, Objective, Assessment, and Plan) notes, a method of documentation widely used by healthcare providers.
Th e 17-year-old said: 'Doctors I spoke to say that note-taking is time-consuming and inconvenient. One of them even employed a scribe just to take notes during consultations.
'This tool will help reduce the time they spend doing administrative work in-clinic, while also improving the quality of medical note-taking.'
John is among a growing crop of teens who are developing apps to solve issues.
He started developing SgScribe in May and tested it by having it transcribe pre-prepared, simulated medical consultations, sometimes even playing three different recordings at once to conduct as many tests as possible.
Currently, SgScribe can transcribe recordings in four different languages – English, Mandarin, Malay and Indonesian.
Doctors log into the platform, where they can record live consultations or upload recorded consultations to generate SOAP notes.
SgScribe ensures patient confidentiality. Recordings are saved with no patient identifiers, and SOAP notes generated merely describe what took place during the consultation without including the names or other personal details of the patients.
The model produces SOAP notes with 90 per cent accuracy, said John, who added that inaccuracies occur when patients mumble or speak unclearly.
SgScribe is currently in its beta phase where it is being tested by three specialists to whom John reached out via his parents' contacts and school's alumni network. It has processed over 200 consultations.
An example of the landing page of SgScribe, an AI-powered medical transcription tool that converts recordings of doctor's consultations to SOAP notes.
ST PHOTO: MARK CHEONG
One of the doctors participating is orthopaedic surgeon Dr Lo Ngai Nung, who said SgScribe allowed him to focus and analyse patients rather than trying to multitask during consultations.
He added: 'It's like having a medical scribe record your notes as you carry out the consult.'
Dr Lo also said he found SgScribe's ability to transcribe in two languages simultaneously particularly useful. 'In our cultural context where patients and doctors often mix English-Chinese usage during the consultation, this is very critical.'
Currently, John has spent $100 of his own money to run the transcription and note generation AI models for SgScribe. He credits free access to coding models offered by tech companies to students for keeping costs low.
John plans to keep the web-based transcription tool free indefinitely. Instead, SgScribe will be monetised via additional paid features and advertisements on the website.
To expand his project, John also plans to raise a round of seed funding for the tool and expand his team to include some of his ex-classmates after he completes his International Baccalaureate examinations at the end of the year.
After that, he said, and after doing his National Service, he will probably have a gap year to work on the project full-time.
He added that developing SgScribe has helped him decide what he wants to do in the future.
He said: 'I always thought I would be a doctor. But since working on SgScribe, I realised that I can make the most impact through software.'
