As demand for telemedicine treatment continues to grow, global health companies are turning to AI to reach patients in underserved areas.
MORE Health is a Silicon Valley-based company that provides access to top international physicians for patients faced with critical illnesses such as cancer or heart disease. The company was founded in 2013, and recently took a leap forward by partnering with Houston-based Melax Technologies, which applies natural language processing techniques to clinical documents in a Mandarin-English medical information translation system.
China is the biggest overseas market for MORE Health. Patients in megacities such as Beijing, Shanghai, and Guangzhou can face long waiting times to see physicians, especially specialists. In 2017, 500,000 Chinese nationals sought medical treatment abroad, with an average per patient cost of US$150,000.
MORE Health CTO Bo Hu told Synced that AI techniques such as Natural Language Processing have significantly improved efficiency in translating and processing patients’ medical records. “We used to have three case managers spending a whole day dealing with one patient record. Today, a case manager can take care of five to seven cases each day with the help of NLP.”
MORE Health Co-Founder Dr. Robert Warren is a UCSF professor and highly esteemed physician and oncologist. Dr. Warren’s vision is a platform that connects a patient’s attending doctor(s) with remote specialists to develop a collaborative diagnosis. A treatment plan typically includes drug prescriptions and delivery, concierge overseas treatment service if necessary, and up to six months of follow-ups.
With over 700 top-tier US physicians on board, MORE Health has the medical expertise required for comprehensive diagnosis and treatment, and plans to further expand into the international marketplace, creating “a hospital without borders.” However, as the company expands it faces a new challenge: scaling telemedicine to fill the void in underserved areas.
Before MORE Health physicians can conduct a diagnosis on patients outside the US, they need a translated and reformatted patient medical record that complies with the US electronic health records (EHR) format. This is time-consuming and limits efficiency.
Machine learning researchers have made progress translating texts using deep neural networks — Microsoft software now matches human level for example when translating news stories. However, dealing with medical jargon in different languages is a total different situation, with more both specific vocabulary and more at stake.
For example, Google Translator converts the Mandarin medical phrase “面热潮红，肝火旺盛” into “Face hot flashes, Strong anger”, when in reality it indicates a more complicated irritablity syndrome involving anxiety, hot flashes and abnormal metabolism.
Previously, global health companies would turn to human translators with knowledge of the appropriate medical or pharmaceutical terminology. MORE Health researchers are now able to develop their NLP systems based on collected labeled datasets to perform tasks like automated entities recognition, medical jargon translation, and content classification.
MORE Health expects even more from AI. CTO Hu told Synced that MORE Health also intends to use deep neural networks for medical image screening, matching patients with appropriate physicians, and medical data structuring. The company has developed the Capsule format, which can systematically store medical records and preserve patient privacy.
Dr. Warren’s main interest meanwhile is finding ways that AI can help doctors perform better. “It really comes down in medicine to say: Can we make a diagnosis better? Can we treat a patient in a better way or more effective way? Can we treat more patients effectively than we are now with AI? Those are really the questions we have to ask when it comes to AI,” says Dr. Warren.
Journalist: Tony Peng | Editor: Michael Sarazen