ABSTRACT
Artificial intelligence provides a promising solution for streamlining COVID-19 diagnoses;however, concerns surrounding security and trustworthiness impede the collection of large-scale representative medical data, posing a considerable challenge for training a well-generalized model in clinical practices. To address this, we launch the Unified CT-COVID AI Diagnostic Initiative (UCADI), where the artificial intelligence (AI) model can be distributedly trained and independently executed at each host institution under a federated learning framework without data sharing. Here we show that our federated learning framework model considerably outperformed all of the local models (with a test sensitivity/specificity of 0.973/0.951 in China and 0.730/0.942 in the United Kingdom), achieving comparable performance with a panel of professional radiologists. We further evaluated the model on the hold-out (collected from another two hospitals without the federated learning framework) and heterogeneous (acquired with contrast materials) data, provided visual explanations for decisions made by the model, and analysed the trade-offs between the model performance and the communication costs in the federated training process. Our study is based on 9,573 chest computed tomography scans from 3,336 patients collected from 23 hospitals located in China and the United Kingdom. Collectively, our work advanced the prospects of utilizing federated learning for privacy-preserving AI in digital health. The COVID-19 pandemic sparked the need for international collaboration in using clinical data for rapid development of diagnosis and treatment methods. But the sensitive nature of medical data requires special care and ideally potentially sensitive data would not leave the organization which collected it. Xiang Bai and colleagues present a privacy-preserving AI framework for CT-based COVID-19 diagnosis and demonstrate it on data from 23 hospitals in China and the United Kingdom.
ABSTRACT
OBJECTIVE: To summarize the experience of three Chinese cities (Wuhan, Shanghai and Haikou) and provide a reference for global efforts to combat COVID-19 spread among children. MATERIALS AND METHODS: Through collecting the measures and outcomes of preventing and controlling COVID-19 in China's three hospitals, we compared the effect of different strategies. RESULTS: From January to March 2020, the number of suspected and confirmed COVID-19 cases in Wuhan increased exponentially, and Wuhan Children's Hospital as a whole was transformed into a designated quarantine and treatment facility, which is the "Wuhan Model". Shanghai has more children's hospitals with better capabilities to tackle public health emergency. Besides, it is far away from Wuhan and had a small caseload. Children's Hospital of Fudan University, a facility in Shanghai to treat pediatric infectious diseases, is famous for its well-equipped building for infectious disease treatment and professional medical team, and therefore no major transformation was required. That is the "Shanghai Model". Haikou is located on an island. Amid the outbreak, large numbers of tourists and travelers from Hubei had already arrived in Haikou. Hainan Women and Children's Medical Center, as the only pediatric care hospital in Hainan Province, did not have a separate building for infectious disease treatment. After a citywide survey of the medical resources and facilities available, a temporarily idle hospital 3 kilometers away from Hainan Women and Children's Medical Center was requisitioned as the quarantine and treatment facility for pediatric cases. That is the "Hainan Model". The three models enabled the treatment of all suspected and confirmed cases and no fatality was reported. CONCLUSIONS: The COVID-19 coping strategies for children should be designed according to the existing conditions of the local children's hospitals and the risk levels of the epidemic.