Using Artificial Intelligence to Predict Oxygen Demand for Patients with COVID
Addenbrooke’s Hospital in Cambridge and 20 other hospitals worldwide and healthcare technology leader NVIDIA have utilized artificial intelligence (AI) to anticipate Covid patients’ oxygen needs on a global scale.
The pandemic triggered the research study and set out to develop an AI tool to forecast how much additional oxygen a Covid-19 patient may require in the initial days of hospital care, utilizing data from four continents.
The technique, called federated learning, made use of an algorithm to analyze chest x-rays and also electronic healthcare data from hospital patients with Covid symptoms.
In order to maintain strict patient confidentiality, the patient data was completely anonymized, and an algorithm was delivered to each hospital, so no data was shared or left its location.
The analysis was united to construct an AI tool that might anticipate the oxygen needs of hospital Covid patients anywhere in the globe as soon as the algorithm “learned” the data.
Published on September 15 in Nature Medicine, the study called EXAM (for EMR CXR AI Version) is among the most extensive, most varied clinical federated learning studies to date.
To check the precision of the EXAM, it was tested out in many hospitals throughout five continents, including Addenbrooke’s Hospital. The outcomes showed it predicted the oxygen required within 24-hour of a person’s arrival in the emergency department, with a level of sensitivity of 95 percent and a specificity of over 88 percent.
“Federated learning has the immense transformative power to bring AI advancement to the clinical workflow,” said Professor Fiona Gilbert, that led the research in Cambridge and is an honorary consultant radiologist at Addenbrooke’s Hospital and chair of radiology at the University of Cambridge School of Clinical Medicine.
“Our ongoing work with EXAM shows that these types of international collaborations are repeatable and also more efficient, to make sure that we can fulfill clinicians’ demands to take on complicated health obstacles and future epidemics.”
The first author on the research, Dr. Ittai Dayan, from Mass General Bingham in the US, where the EXAM algorithm was created, stated:
“Typically in AI development, when you produce an algorithm on one hospital’s data, it does not function well at any other hospital. By creating the EXAM model using federated learning and objective, multimodal information from different continents, we were able to construct a generalizable design that can assist frontline physicians worldwide.”
Combining partners across North and South America, Europe, and Asia, the EXAM research study took two weeks of AI ‘learning’ to achieve high-quality forecasts.
“Federated Learning enabled researchers to team up as well as establish a brand-new standard of what we can do around the world, using the power of AI,” stated Dr. Mona G Flores, Global Head for Medical AI at NVIDIA. “This will advance AI for medical care and across all industries intending to build durable models without compromising privacy.”
The results of around 10,000 COVID individuals from throughout the world were analyzed in the research, including 250 that involved Addenbrooke’s Hospital in the very first wave of the pandemic in March/April 2020.
The National Institute for Health Research (NIHR) and the Cambridge Biomedical Research Centre (BRC) supported the research.
Development on the Exam model proceeded. Mass General Brigham and the NIHR Cambridge BRC are currently working with NVIDIA Inception startup Rhino Health, co-founded by Dr. Dayan, to run potential research studies utilizing EXAM.
Professor Gilbert included: “Developing software to match the performance of our best radiologists is complicated, however a transformative aspiration. The more we can safely integrate information from different sources utilizing federated learning and cooperation, and have the room needed to innovate, the faster academics can make those transformative objectives a fact.”
Originally published on Sciencedaily.com. Read the original article.
Reference: Ittai Dayan, Holger R. Roth, Aoxiao Zhong, Ahmed Harouni, Amilcare Gentili, Anas Z. Abidin, Andrew Liu, Anthony Beardsworth Costa, Bradford J. Wood, Chien-Sung Tsai, Chih-Hung Wang, Chun-Nan Hsu, C. K. Lee, Peiying Ruan, Daguang Xu, Dufan Wu, Eddie Huang, Felipe Campos Kitamura, Griffin Lacey, Gustavo César de Antônio Corradi, Gustavo Nino, Hao-Hsin Shin, Hirofumi Obinata, Hui Ren, Jason C. Crane, Jesse Tetreault, Jiahui Guan, John W. Garrett, Joshua D. Kaggie, Jung Gil Park, Keith Dreyer, Krishna Juluru, Kristopher Kersten, Marcio Aloisio Bezerra Cavalcanti Rockenbach, Marius George Linguraru, Masoom A. Haider, Meena AbdelMaseeh, Nicola Rieke, Pablo F. Damasceno, Pedro Mario Cruz e Silva, Pochuan Wang, Sheng Xu, Shuichi Kawano, Sira Sriswasdi, Soo Young Park, Thomas M. Grist, Varun Buch, Watsamon Jantarabenjakul, Weichung Wang, Won Young Tak, Xiang Li, Xihong Lin, Young Joon Kwon, Abood Quraini, Andrew Feng, Andrew N. Priest, Baris Turkbey, Benjamin Glicksberg, Bernardo Bizzo, Byung Seok Kim, Carlos Tor-Díez, Chia-Cheng Lee, Chia-Jung Hsu, Chin Lin, Chiu-Ling Lai, Christopher P. Hess, Colin Compas, Deepeksha Bhatia, Eric K. Oermann, Evan Leibovitz, Hisashi Sasaki, Hitoshi Mori, Isaac Yang, Jae Ho Sohn, Krishna Nand Keshava Murthy, Li-Chen Fu, Matheus Ribeiro Furtado de Mendonça, Mike Fralick, Min Kyu Kang, Mohammad Adil, Natalie Gangai, Peerapon Vateekul, Pierre Elnajjar, Sarah Hickman, Sharmila Majumdar, Shelley L. McLeod, Sheridan Reed, Stefan Gräf, Stephanie Harmon, Tatsuya Kodama, Thanyawee Puthanakit, Tony Mazzulli, Vitor Lima de Lavor, Yothin Rakvongthai, Yu Rim Lee, Yuhong Wen, Fiona J. Gilbert, Mona G. Flores, Quanzheng Li. Federated learning for predicting clinical outcomes in patients with COVID-19. Nature Medicine, 2021; DOI: 10.1038/s41591-021-01506-3
