Login screen for Grab Health using Good Doctor Technology
Good Doctor Technology (GDT) in Indonesia emerged from the confluence of three Asian giants—Ping An Good Doctor from China, Grab from Southeast Asia, and SoftBank from Japan. Ping An Good Doctor is the largest telemedicine application in China in terms of user scale and coverage. It currently has over 72.6 million monthly active users and over 373 million registered users, making it the most widely used telemedicine platform in the world. Their proprietary software platform uses "mobile medical + AI technology" to support an in-house medical team of more than 2,200 doctors and related health professionals. It provides 24/7 medical services through online consultations (an average of over 903,000 per day), referrals, registrations, and online drug purchases and deliveries. Good Doctor was incubated by Ping An, a large financial services firm, and became a separate company with an IPO in 2018.
Grab, Southeast Asia’s first startup with a valuation over U.S. $10 billion, is the region’s leading “super app” providing everyday services, including ride-hailing, delivery (food, packages and groceries), mobile payments, and financial services, in over 390 cities across eight Southeast Asia countries. Over 9 million people earn money on the Grab platform as drivers, delivery partners, merchants, or agents, and over 198 million people have downloaded the app.
SoftBank Group, a Japanese multinational conglomerate, invests in many technology, energy, and financial companies, and also runs Vision Fund, the world's largest technology-focused venture capital fund. Softbank has previously invested in both Ping An Good Doctor and Grab, and more recently in the Good Doctor Technology joint venture.
Currently, GDT only provides services to customers in Indonesia, while corporate headquarters are in Singapore. The service was first made available to Jakarta Grab users for beta testing in October 2019 as Grab Health, and officially launched in Jakarta in December 2019. Grab is now available in a broader set of Indonesia cities and towns.
Behind the Grab Health icon within the Grab app in Indonesia is GDT and its health services technology platform for mobile users. Behind GDT is the Ping An Good Doctor company and its well-developed on-line platform, formidable AI and technology capabilities, and deep user experience. While Ping An Good Doctor is China’s most widely used, comprehensive and sophisticated mobile medical platform, the localization process for the platform is complex and time intensive. The AI capabilities of the software must be trained in a new language and setting, and the local medical staff team built to partner with the AI-enabled platform for consultations. Localization also requires an ecosystem of relationships with external healthcare delivery providers and pharmacies in any country GDT is transferred to.
Drs. Chet Yong and Wawan Harimawan
Dr. Chet Yong, a Singaporean, is the Chief Operating Officer, GDT Southeast Asia. He worked in Singapore’s private sector healthcare system and two different healthcare companies as a practicing doctor and a senior administrator of medical services. He also spent several years with a consulting firm, heading their Southeast Asia regional healthcare consulting industry practice.
When Chet saw the news announcement of the new Ping An Good Doctor/Grab joint venture in 2018, he wanted to be part of it, and explained his motivation:
“Healthcare has an ‘Iron Triangle’ paradigm that access, cost and quality cannot all be improved at the same time, largely because of healthcare’s brick and mortar facilities. I saw that GDT would be focusing on those aspects of healthcare service delivery that come before and after treatment at physical facilities, and not on healthcare delivery within facilities. I realized this could break this healthcare Iron Triangle paradigm in my home region of Southeast Asia. In most parts of Southeast Asia there are never enough doctors, especially outside of the major urban areas. GDT had the potential to expand access given this shortage.”
In July 2019, Chet joined the GDT corporate office as the Chief Operating Officer.
Dr Wawan Harimawan is the Medical Intelligence Project Manager at GDT Indonesia. Wawan had managed health promotion projects for low-income families in Medan and taught medical school students in the area. As a research assistant for a prominent medical researcher in Jakarta, he was exposed to data analytics and statistical predictions related to geriatric health conditions. As a physician treating many types of patients across different provider settings, he experienced first-hand the vast disparities in health care quality and availability across Indonesia’s socio-economic spectrum.
Like Chet, when Wawan heard about GDT he thought the technology could provide substantial improvements in healthcare access in Indonesia. Initially he still worked at the hospital while working as Medical Consultant at GDT to set-up the platform for online consultations. In November 2019, Wawan started patient consultations via the GDT on-line platform, doing 4600 in his first six months. He also helped with localized medical and health promotion content and guided the effort on improving workflows for on-line consultations and follow-up interactions.
In March 2020, Wawan became the full-time Medical Intelligence Project Manager for GDT. He continued to lead localization as well as ongoing training of the AI algorithms. Together with the Medical Management Team, he also oversees the on-boarding and training of the 100 plus doctors and other full-time medical staff using the platform to perform on-line consultations. When he converted to full time, Covid-19 was starting to spread through Jakarta and other parts of Indonesia. His role expanded to adapting the local version of the platform and the supporting work processes to respond to the pandemic. While Covid-19 presented unexpected challenges to the GDT start-up team, it also created opportunities. More people were willing to seek consultations online for fear of traveling to their nearest doctor clinic.
How the GDT Platform Works
The GDT platform consists of two main systems. There is the doctor’s front-end system used for all aspects of the patient consultation, called the Doctor’s Workbench. Patients interact with this system using their mobile phones and a chat-based interface. There is also a back-end system linking to pharmacy partners that transmits the prescription information to the pharmacy, which integrates with the Grab platform to arrange for a driver to pick up and deliver any prescribed medication or supplements.
As part of supporting their internationalization effort, the Ping An Good Doctor platform team in China had created an English version of the chatbot. The Indonesian GDT team started with this version and then created a Bahasa Indonesian version, associating local Indonesian word usage, ways of describing symptoms and common expressions, with the huge database of medical condition symptoms that were already part of the Ping An platform.
The very first step of the GDT consultation via the chatbot interface is to determine patients’ purpose. Do they have an actual need for a medical consultation, or are they approaching the doctor for other reasons such as vitamin promotions or seeking general health information? It is important to quickly identify those in need of an actual medical consultation, as the support staff and the intelligent platform can handle non-medical requests. The AI-enabled chatbot engages with the patient through a series of questions and answers to make this initial first step determination. Chet explains, “The role and persona of the GDT chatbot is similar to that of the nurse, who would do a pre-screening interview of a patient in preparation for the doctor’s follow-on interview.” The GDT platform can handle this initial determination of purpose in parallel for hundreds or thousands of patients at a time.
The second step is to identify the symptoms and match them to possible clinical pathways—causes that could possibly lead to this patient’s particular presentation of symptoms. The AI-enabled natural language processing (NLP) of the GDT Doctor’s Workbench chatbot suggests an initial set of screening questions for the patient to respond to, and then generates follow-on questions in response to the patient’s text inputs. During the intensive and long-running localization and AI algorithm training period, GDT doctors reviewed every question suggested by the chatbot and revised them if required. Revisions to the questions provide training feedback to the system’s NLP algorithms, and the system gets more accurate in understanding local responses.
An important triage takes place right at the outset of the online interaction to assess symptoms as the doctor, with the help of the GDT platform, must quickly determine if a particular consultation is a standard and simple situation (80 to 90% of all inquiries), an emergency situation, or a more complicated but non-emergency situation. Emergency situations are fast-tracked and immediately referred to appropriate physical facilities for the necessary in-person treatment.
For the standard simple situation and the more complicated non-emergency situation, the chatbot-facilitated dialogue between the patient and the GDT Platform/human doctor would continue onward. With guidance and oversight from the human GDT doctor, the automated Doctor’s Workbench continues to generate questions to the patient via the chatbot and uses its AI-enabled capabilities and large database of prior experience (from both China and Indonesia) to generate its prediction of the most likely clinical pathways. However, meticulous localization is also required here, as some medical conditions are common in Indonesia (e.g., dengue) that are not common in China, and vice versa.
The third step in the consultation process is to make a specific diagnosis of the patient’s medical condition that is consistent with their presentation and description of symptoms. Indonesian law requires that that a medical diagnosis be made by a qualified human doctor. The targeted end state is that the GDT doctor reviews the list of diagnosis predictions automatically generated by the Doctor’s Workbench system—shown in order of highest likelihood based on probabilities calculated by the system from prior cases—and makes the final decision on the diagnosis.
As the AI-enabled consultation support system is always learning and sometimes encountering unfamiliar symptoms and diseases, it will inevitably make some predictions that turn out not to be correct. Even so, Wawan states that its ordered list of predicted possibilities for diagnosis ends up helping the human doctor tremendously. The data-driven list of suggestions generated by the platform helps the doctor to systematically consider a range of possibilities. Chet comments, “This is why the Ping An Good Doctor experience in China that combines the human doctor with the AI-enabled workbench support system has been so successful and well received.”
The fourth and final step in the consultation process is the specification of the treatment plan. This includes specifying prescription medicines and non-prescription supplements, and providing educational materials for the patient about their situation, medication, and treatment plan. It could also include a referral to a specialist clinic or hospital. Again, Indonesian law requires that the prescription has to be issued by a qualified human doctor. Once the doctor makes the final choice of the diagnosis, the system uses its Master Drug Bank Database to suggest medications that match the diagnosis and specific conditions. The doctor reviews the recommendations from the system and makes the final choice of medication. The system also provides support on determining appropriate dosages and for screening the patient for potential harmful drug interaction effects.
GDT has created a network of pharmacies across Indonesia—from large regional and national chains to very small neighborhood or village providers—that are tied in with the back-end prescription management system of the GDT platform. The medication is placed in special tamper-proof bags for secure delivery to the patient. In addition, special protocols identify the Grab driver who picks up and delivers each prescription order, and verification is needed of final delivery to the patient.
Chet’s experience as a physician in Singapore turned out to be especially helpful for working out all of these details. He is the current standing Chairman of the Biomedical and Health Standards Committee (BHSC), which oversees the Singapore Working Group that worked out the protocols standards for the delivery of medications from a pharmacy to a patient’s doorstep, and used them as the basis for setting up the GDT system in Indonesia. To deliver medication to the patient, Grab used the already well-developed processes established for food pick-up and time sensitive deliveries. Grab’s platform was also used to handle payments to GDT in Indonesia.
Impacts of Using the GDT Platform for Consultations
Wawan observes that “many middle- and lower-income Indonesians refrain from seeing the doctor unless it is a severe or emergency situation due to the barriers of travel times and cost.” The GDT approach, he says, “addresses the challenges of access and fears associated with consultations being too expensive.”
GDT also improves a doctor’s productivity, and doctors who are using the AI-enabled Doctor’s Workbench to do online consultations can dedicate 70 to 80% of their work time providing consultations versus doing clinic support work. The system automatically takes care of data entry, data collection, payment, and other administrative and workflow tasks. In addition, doctors can consult with multiple patients in parallel. Productivity increases in other parts of the healthcare service delivery ecosystem as well. For instance, ‘pre-qualification and vetting’ in GDT consultations helps to reduce the number of people who appear at medical facilities who don’t need to be there, helping overloaded physical facilities use their time and resources more productively.
Future Directions for GDT
Because localizing the platform is complex and time intensive, GDT plans to localize one country at a time—Thailand is next. There are also plans to improve the GDT system for Indonesia and other Southeast Asian countries. Chet uses his 5C conceptual framework—Consultation, Commerce, Content, Continuous (Care), and Community—to explain the overall directions of GDT.
For consultation, the GDT Indonesia platform is already very good at the first step of determining the patient’s purpose (medical vs non-medical). They will continue to improve the capabilities of the Doctors Workbench and its chatbot interface to support the doctor with the other three steps. They are also improving the process for following up with a patient after referral by the GDT doctor to a specialist or a hospital. GDT has a special version of the app that can be used by partnering specialist doctors and hospitals for patient follow-ups after treatment at their physical facility. In addition, GDT will consider adding video as a premium service. This will not only help in Indonesia—for those patients willing to pay somewhat more for each consultation—but in other Southeast Asian countries (e.g., Singapore) where the national standards for conducting remote consultations require video.
In terms of content, GDT has created a growing body of educational material to support the consultation process, but plans to do much more. The GDT chatbot has also been getting a steadily increasing number of inquiries for health information.
With regard to commerce, the GDT platform already allows the doctor to prescribe and fulfil prescriptions for medication and orders for other health supplements, but GDT plans to expand the range of offerings to support medical needs and overall well-being.
For continuous care, home monitoring devices for chronic conditions continue to improve and become less expensive. They will enhance GDT’s ability to provide consultation support for such conditions as a doctor would. Chet notes that GDT has brought on specialist doctors as Clinical Program Directors to support expansion into chronic care online consultation and referral.
The final C, community, involves increasing the social aspects of the platform to enable people with common healthcare situations and needs to reach out to one another, share experiences and information, and discuss the educational information they obtain through the platform. Other online communities related to health and well-being exist in Indonesia and other Southeast Asian countries, but Chet believes that by integrating all five C elements, the GDT platform communities can offer an inclusive one-stop destination that other health information providers would find hard to match.
Looking forward, Chet will continue expanding the GDT related healthcare delivery ecosystem across Indonesia with the 5C framework, establishing GDT in Thailand, and planning eventual expansion to other Southeast Asian countries. Wawan will continue to oversee localization and training of the GDT platform in Indonesia. He will also work with the Ping An Good Doctor team in China and his own staff in Indonesia to deploy localized versions of new features and capabilities to the Doctor’s Workbench system. He will also continue training and supervising new doctors who join GDT.
Together they will continue their efforts to demonstrate that with a responsible and innovative approach to AI-enabled digital transformation, there are indeed ways in which the access, cost and quality of general practitioner-type healthcare—and of some specialist doctor consultations as well—can be improved simultaneously. By doing so, they can have positive effects on resource usage and productivity of medical resources in Southeast Asia and perhaps other regions as well.
