The art of intelligent conversation
When the National University Healthcare Systems (NUHS)’ directory system was due for a refresh in September last year, Dr Ngiam Kee Yuan found himself standing on the cusp of an opportunity to solve a longstanding problem. The Group Chief Technology Officer of NUHS was no stranger to the bugbear of healthcare professionals when it came to information: there wasn’t a single, convenient interface to access it.
Dr Ngiam, who promotes interdisciplinary research on artificial intelligence in healthcare through his joint appointments with NUS Biomedical Engineering and the NUS Saw Swee Hock School of Public Health (SSHSPH), worked with a local medical start-up, Bot MD, to develop a chatbot app that allows clinicians to access hospital information through a free-text messaging interface. The NUHS-Bot MD A.I. clinical assistant was developed in six months and provides easy access to a repository of information that healthcare professionals need on a daily basis, including call rosters, hospital protocols, drug dosage information, and drug formulary.
Having all the information they need at their fingertips means that healthcare professionals don’t need to sift through multiple sources, whether it’s connecting to the intranet to find protocols, making calls to the pharmacy for drug enquiries, or searching through webpages to find out who is on call. Users only need to key in their queries, and they can receive answers instantly, wherever they may be.
The chatbot also contains custom tools for more specific groups of users, such as a chemotherapy cost calculator that instantly calculates treatment costs based on the dosing regimens and drug prices at the National University Cancer Institute Singapore (NCIS). This serves as a useful financial counselling aid for NCIS clinicians and frontline medical social workers to help cancer patients determine appropriate treatment options.
But the impact of the chatbot can perhaps be felt most keenly in the current battle against the COVID-19 pandemic. NUHS has been able to use it to put out the latest COVID-19 directives, protocols and contact numbers quickly for its time-starved frontline workers, who can also use the chatbot to record their daily temperature readings.
Its use in providing COVID-19-related information to healthcare professionals is an example of how the chatbot can be adapted, and constantly updated, to meet evolving needs. This, Dr Ngiam shares, is baked into the DNA of the chatbot build. “Understanding and seeking out user needs is a vital process,” he said, citing the pilot phase as one of the key steps in the project. Unrolled last November, the pilot allowed the development team to garner feedback not just on the usability of the interface, but also on user patterns, the type of information users seek most, and even questions that the chatbot could not answer, which they then trained it to do. That process continues today, said Dr Ngiam: “We are finding out the kind of content and features users want on the chatbot and making sure they are kept updated.”
Over 500 doctors, nurses, pharmacists and medical social workers from NUHS are now using the chatbot under the first phase of its rollout. There are plans to expand access to over 3,000 more healthcare professionals across the NUHS cluster by the end of 2020, including to Ng Teng Fong General Hospital, Alexandra Hospital, and the National University Polyclinics.
The NUHS-Bot MD A.I. clinical assistant app joins other initiatives by NUHS to incorporate artificial intelligence (AI) in healthcare. An example is DISCOVERY AI, a platform developed by NUHS, the NUS School of Computing and SSHSPH. A platform that accumulates healthcare data and enables AI tools to be developed and deployed, DISCOVERY AI is capable of processing huge amounts of data to make predictions, which in turn helps clinicians make better decisions.
“I think we’re still in the early stages of healthcare digital transformation. To be fair, healthcare has traditionally been lagging in terms of using technology because of the high barriers to entry and regulation,” said Dr Ngiam.
Whatever form these initiatives take, they are underscored by agility. By leveraging on modern software technology, the resulting frameworks are nimble and elastic, and help to bring medical systems out of old legacy-type frameworks. This agility also brings about a host of other possibilities. As Dr Ngiam shared, “The chatbot is designed to be user-centric — it can tell a joke, or even show pictures of cats! It is important to the users that the software is personable and easy to use.”