“The faculties in the college of medicine at our school (here referring to National Yang Ming Chiao Tung University, NYCU) are also medical practitioners, so we can seamlessly connect with medical industries in terms of university-industry cooperation, which stands out from other universities. After the merger of Yang Ming and Chiao Tung University, the information and communication resources from both campuses were further integrated and the construction of 5G network then help the hospital to build more efficient AI applications, giving NYCU an advantage to develop smart medical care.” said Albert C. Yang, Director of the Digital Medicine Center at NYCU.
Market Demands Boost User-friendly Medical Products, Attracting Worldwide Attention with Visionary R&D Talents
During the past three decades, the technology of medical images, physiological signals, and electronic medical records flourishes, gradually realizing the digitization of hospital information. As the Artificial Intelligence Algorithm (AI Algorithm), data analysis, and other related technologies start to grow, how to use the collected medical data with well-established algorithms to help medical teams make safer and more effective decisions has become more and more important in the field of “smart medical care”. This trend has also led to the development of the “Digital Medicine Center” at NYCU.
According to Professor Yang, both Yang Ming and Chiao Tung campuses before the merger already had their own talents in the field of clinical medicine, science and engineering. The two campuses have also cooperated with each other under the Foresight Project of National Science and Technology Council. After the merger in 2021, the university founded the Digital Medicine and Smart Healthcare Research Center to integrate the talents of clinical medicine, computer science, as well as electronics and electrical engineering from the two former universities. Digital Medicine and Smart Healthcare Research Center also helps the university build a deeper relationship with the industry, so they can commoditize the technology and promote the idea of Smart Wards. “We focus on the development of medical AI and medical Internet of Things (IoT) technology, while also work with the industry to develop interdisciplinary R&D proejcts.” Yang said.
Clinical knowledge along with science and engineering creates seamless university-industry cooperations
While many manufacturers have been working on developments of medical equipment, there is a high threshold for these products to be put into use. Therefore, the university established the Digital Medicine and Smart Healthcare Research Center, hoping to find a way of putting these equipment into clinical settings, and may further promote these equipment to domestic and overseas markets. Since 2020, the university has been working with Chunghwa Telecom to develop applications in the field of sleep medical care. Team developed an “artificial intelligence (AI)-assisted household sleep detection platform” that allows users to detect their own sleep quality at home by connecting to wearable devices from a cell phone app. The platform won the National Innovation Award in 2022 and received a patent in Taiwan, and is currently utilized in clinical practice at the Taipei Veterans General Hospital.
The “online silent hypoxia monitoring platform” is another result of the interdisciplinary collaboration between NYCU, Chunghwa Telecom, and manufacturers of medical wearable device. When the first wave of COVID-19 outbreak in Taiwan occurred in May, 2021, the team quickly adopted the sleep detection app into the first remote hypoxia monitoring app in Taiwan within just two weeks. The platform utilizes technologies such as Internet, 5G transmission and communication, artificial intelligence, and information security. These applications not only met the three major requirements for quarantined medical care during the pandemic, which is “convenient usage”, “real-time integration of physiological information” as well as “accurate assistance in interpreting medical conditions”; they also beckoned others in medical care facilities industry to develop domestic hypoxia-detection facilities.
Intelligent Brain Imaging Platform Bringing out AI Algorithm Spirit
When being asked about the current focus of medical AI technology of the Digital Medicine and Smart Healthcare Research Center, Yang highlighted the “Explainable AI” at the first place, since the team needs to constantly verify whether the features found by AI models have special significance in clinical medicine during the process of technology development. Meanwhile, the AI models should not only determine whether patients are infected, but also need to pinpoint the location of lesions and other related information, which is to “discover explainable features when developing AI models”.
The “Smart Brain Imaging Diagnostic Service Platform” developed by Yang’s is one of the best applications of “Explainable AI Algorithms”. As a psychiatrist himself, Yang said that in psychiatric clinical practice, most of the symptoms are obtained through diagnostic interview, and the diagnosis mostly relies on doctors’ experience and subjectivity. He hopes that this kind of objective imaging tools can help us understand the blind spots of the brain so the management of mental disorders can be more precise than before, including the use of the core application of AI in Deep Learning. The team uses algorithm technology to mark the location of disease-related brain lesions so we may precisely locate the lesions; at the same time, they use data analysis and information and communication technology to visualize the lesion information, so images can be analyzed quickly online for doctors and patients view them in real time. From the development to application, the intelligent brain image diagnostic service platform has been awarded the National Healthcare Quality Award by the Joint Commission of Taiwan and has been validated by several hospitals in multiple centers. They also received patents in Taiwan (No. I744798) and in U.S. (No. 11,379,982).
This platform has also become a powerful tool for NYCU to enter the field of medical AI research worldwide. The team has started to cooperate with the top units in the field, such as the Medical School in Harvard University, University of Toronto in Canada, and National Institutes of Health of U.S., to make use of their unique and rich database to extend the application of brain imaging in different clinical settings. Even though the diseases they target are different, they all can be quantified using magnetic resonance imaging. For example, the University of Toronto in Canada and the National Institutes of Health in the United States have the richest databases on autism, hence the Center initiates collaborative research on autism with these institutes, hoping to use intelligent imaging to assess the changes in the course of autism from childhood to adulthood as a reference for future clinical diagnosis.
Collaborating with International Institutes on the Pioneering Concept of Federated Learning
While using the medical AI technology, doctors often encounter a problem when developing smart medical care: the AI models developed in a single hospital do not perform well in other hospitals. Such problem happened because the facilities used in collecting data are different, creating difference in numbers and specifications. Another factor that causes the problems is that the data among hospitals cannot be shared due to the privacy principle, which indirectly results in poor performance of AI models. In order to improve this weakness, the internationally popular “Federated Learning” application turns out to be the best solution.
“The concept of ‘Federated Learning’ is based on close-loop connections between institutions; after obtaining data from multiple centers as well as verifications, the regulations on medical AI models can thus be established, making it available to everyone.” Yang explains. Digital Medicine and Smart Healthcare Research Center not only cooperates with China Medical University Hospital and Chang Gung Medical Hospital to promote a federated learning model between hospitals; since 2022, the Center has also started a transnational project with Dr. Yang C., director of Intramural IT& Bioinformatics Program of the National Institutes of Health in the U.S. They conduct federal learning training on the topic of “computed tomography in cerebral hemorrhage” hoping to make effective, rapid screenings of computed tomography to determine the type and cause of hemorrhage.
“Although federated learning is a relatively new issue worldwide, the NYCU team has been working on it for a long time, making us ahead of others.” Yang said confidently.
When it comes to the future prospects and achievements of the Digital Medicine and Smart Healthcare Research Center, Yang admitted that as a smart medical center in a university, the main task of the center is to implement medical applications to hospital facilities. However, the current challenge for the technology itself is to pass the accreditation of the government, which is a difficult task relying solely on the center, the school or the hospital.
“The role of the Center is to develop smart medical applications from our medical knowledge and current clinical needs, combined with the engineering talents of both Yang Ming and Chiao Tung campuses. Although the Center can provide clinical resources for application testing and implementation, we still need to work with the industry on certification and commoditization.” Yang shared with us his opinion.
