MedTex Summit Asia 2023: Illustrating the Evolving Landscape of Healthcare
The 6th MedTex Summit Asia not only featured technology giants discussing new trends in the global smart healthcare industry, but also healthcare leaders from Taiwan and overseas, who shared their insights into the applications of AI and ICT in the clinical field based on their front-line experiences.
Creating a Rapidly Advancing AI Healthcare Technology Ecosystem
René Torres, Vice President, Sales, Marketing and Communications Group, GM Networking & Edge Solution Sales at Intel Corporation, delivered a keynote speech titled “Creating a Rapidly Advancing AI Healthcare Technology Ecosystem – How to Drive More Efficient and Secure Medical AI Innovation”, sharing how the leading chip manufacturer is driving AI innovation for enhanced efficiency and security. He emphasized that medical applications of AI can be found everywhere, including disease screening, drug discovery, genomics, robotic surgery, and digital pathology (i.e., scanning traditional slides into digital image files, and then using computer technology and AI systems to acquire, manage, and interpret pathology information therein).
A comprehensive implementation of AI has enabled the rapid processing of large volumes of medical data as well as the improvement of cost-effectiveness in drug development. In the near future, technologies such as data visualization and image recognition are expected to alleviate the global shortage of healthcare workers and unmet clinical needs. Also, advances in natural language processing (NLP) and large language models (LLM) are enabling physicians to quickly generate medical records and utilize the time saved to focus on providing better care for patients.
Transformation of the Medical Experience Powered by AI PC
Torres also mentioned that with the imminent introduction of AI PCs, generative AI will be introduced into personal computers for the first time in history, with shipments expected to exceed 100 million units by 2025, and the application of AI in healthcare is also expected to develop at an unprecedented pace. He pointed out that the Intel Core Ultra processor includes not only CPU chips for low-latency AI workloads, but also a graphics processing unit (GPU) and a neural-network processing unit (NPU) to optimize the performance and power efficiency of AI software.
According to Torres, a GPU could allow rapid interpretation of medical images such as X-rays, MRIs, and CTs, reducing the dose of radiation required for CT scans and speeding up scanning efficiency, thereby reducing an otherwise time-consuming process to less than 30 minutes, while an NPU could be applied to low-power, sustainable AI systems, which would be useful in scenarios that require continuous monitoring of patients’ conditions.
Facilitating Brain Tumor Detection by Federated Learning
René Torres also mentioned the concept of Open Federated Learning (OpenFL) in his speech, suggesting that the expansion of healthcare AI applications would not be possible through Intel alone, but rather requires a collaborative ecosystem where the technology industry and the healthcare system work together as a team. For instance, due to legal constraints and patient privacy considerations, patients’ medical records and clinical data are not permitted to be shared outside of the hospital, but through Open FL, researchers can access the data through open-source projects and confidential computing, without compromising data integrity or patient privacy.
Torres cited a study conducted by Intel Labs in partnership with the Perelman School of Medicine at the University of Pennsylvania to illustrate the enormous potential of this novel machine learning approach. The project is the world’s largest medical federated learning study to date, collecting MRI scans from 6,314 patients at 71 sites across 6 continents, and training AI models through deep learning to aid in the diagnosis of glioblastoma, a rare form of brain cancer. The results published in December 2022 showed a 33% improvement in glioblastoma detection.
Applications of Medical AI in Taiwan – Far Eastern Memorial Hospital as an Example
In addition to Torres’ presentation, he also invited Dr. Kuan-Ming Chiu, President of the Far Eastern Memorial Hospital (Far Eastern Hospital), to speak about the application of AI and ICT at the hospital level. Dr. Chiu mentioned that Far Eastern Hospital was the first one to implement home care in Taiwan during the COVID-19 pandemic, and handled about 10,000 cases daily with the help of interactive voice response, mobile apps, and AI chatbots. The hospital later introduced a combination of virtual clinics, telemedicine, and telepharmacy to provide rapid access to antiviral medication for eligible confirmed COVID patients.
As one of Intel’s partners in Taiwan, Far Eastern Hospital has made significant breakthroughs in AI applications in 2023. Using an AI model that analyzes voice to identify potential signs of laryngeal cancer as an example, Dr. Chiu shared how AI can help physicians achieve both timely diagnosis and early intervention. The model combines the 5G telemedicine platform of Far EasTone Telecom, Intel’s OpenFL technology, Far Eastern Hospital’s voice database, and JelloX Biotech’s MetaLite, a 3D pathology analysis software equipped with an AI module. The model can predict laryngeal cancer by simply collecting the patient’s voice for 2-3 seconds through the microphone of a mobile phone, allowing people in remote areas to make use of telemedicine to identify symptoms of laryngeal cancer and seize the golden time for treatment.
Israel’s Largest Hospital Ranks in World’s Top 10 through Smart Healthcare
The fireside chat was moderated by Yu-Chuan Jack Li, Director of the International Center for Health Information Technology, Taipei Medical University. Professor Eyal Zimlichman, Chief Transformation Officer and Chief Innovation Officer of Sheba Medical Center, was invited as the guest speaker to share the future development trend of AI in healthcare and to explain the strategic planning for the implementation of AI in the world’s top smart hospitals.
Located in the Tel Aviv region of Israel, Sheba Medical Center is the largest hospital in the country and the Middle East, treating more than 1.6 million patients annually. The hospital has gone completely paperless since 2004 and remains at the forefront of the digital healthcare transformation, having accumulated data on approximately 2 million patients and a database of 8 million medical images. With its extensive and intensive use of ICT in the healthcare arena, Sheba has been ranked by Newsweek as one of the top 20 best hospitals in the world for five consecutive years starting in 2019 (No. 11 in 2023), and among the top 20 smart hospitals in the world in 2023 (ranked No. 13).
Prof. Zimlichman mentioned that Sheba Medical Center established the ARC Center (Accelerate, Redesign, Collaborate) for Digital Innovation in November 2019, which is dedicated to accelerating digital transformation and building an ecosystem where Israeli digital healthcare start-ups can collaborate with hospitals to develop solutions. The ARC Center is based on four core concepts: digital health, open and organic innovation, international collaboration, and a structured platform, and aims to revolutionize the global healthcare system by 2030 by targeting five areas: precision medicine, extended reality, artificial intelligence, biomedical engineering, and remote care.
Sheba’s Breakthroughs in AI-driven Diagnostics and Digital Pathology
Regarding the applications of AI in healthcare, Prof. Zimlichman pointed out that Sheba has been focusing on areas such as predictive analytics, clinical decision-making, and digital image processing. He also highlighted Aidoc Medical, an AI medical imaging software that was spun off from Sheba Medical Center in 2017. Aidoc’s innovative AI products have been deployed in more than 1,200 hospitals around the world, particularly in the emergency department to identify critical conditions such as stroke and pulmonary embolism. Recently, Sheba participated in a retrospective cohort study using Aidoc’s solutions, the results of which were published in August 2023, and demonstrated that AI-assisted diagnosis and patient triage in the emergency department could reduce the mortality rate of intracranial hemorrhage by up to 30%.
Prof. Zimlichman also introduced a new discipline called Pathoradiomics. Radiomics itself involves the conversion of a large number of medical diagnostic images (including MRI, CT, and PET scans) into quantitative data that can be integrated and analyzed with the power of AI to identify and characterize tumor shapes, sizes, and other features of cancers.
Pathoradiomics extends these applications to images of pathology slides, allowing physicians to identify genetic mutations and biomarkers associated with cancer through image interpretation. Taking non-small cell lung cancer as an example, a Sheba’s study has shown that deep learning algorithms can be used to identify ALK and ROS1 fusion mutations directly from digital pathology slide images, without performing gene sequencing. If such a technology matures in the future, it will be possible for doctors to identify biomarkers carried by patients and then make appropriate diagnoses by analyzing medical images without the need for biopsies.
An “Evolution in Motion” and a “Revolution Waiting to Happen”
In spite of limitations in the quality and accessibility of clinical data, which may pose some obstacles to the expansion and generalization of healthcare AI, Prof. Zimlichman expressed his overall optimism about the development of AI in the biomedical field. He described non-generative AI as “an evolution in motion”, and generative AI as “a revolution waiting to happen,” emphasizing that the advancements of AI in healthcare will bring about substantial and meaningful changes in terms of quality of service, patient safety, health outcomes, healthcare costs, patient experience, and health equity.©www.geneonline.com All rights reserved. Collaborate with us: email@example.com