Artificial intelligence (AI) has already started to disrupt the traditional medical model. While a new medical model is being incubated, there is still long way to go for it to be fully established.
As early as 2008, prediction models based on big data started being used to predict influenza epidemics and inform public health.1 Since then many more AI models have been developed in healthcare. With continuous network technology updgrades, explosive growth of data, and enhanced computing power, in some cases AI models are starting to show excellent performance comparable to experienced doctors, especially in disease diagnosis.2,3 This represents an important potential impact on human health.
AI made a huge contribution in responding to coronavirus disease 2019 (covid-19)4 both in Wuhan city, China, where I live and work, and elsewhere in the world. There were two typical scenarios where AI has been applied against covid-19.
First, big data analytics have been applied in surveillance, prevention and control of the pandemic. Digital QR codes were assigned to individual citizens to record travel history and self-reported health status. AI algorithms process the data and display people’s real-time health status using three colors (green, yellow and red), in which the red indicates high risk and green normal status. The QR code-based system can quickly identify suspected patients and accurately track covid-19 patients and their close contacts.5
Secondly, AI has the potential to solve medical problems when healthcare workers’ capacity is constrained. AI-assisted CT and radiographic examinations have been used to aid diagnosis of people with covid-19, improving diagnostic accuracy of primary medical institutions; increasing the work efficiency of radiologists; and enabling senior doctors to guide junior doctors via telemedicine.6
In early January 2021, a vehicle-carried mobile CT scanner was sent to Hebei Province in China from Zhongnan Hospital of Wuhan University, to assist with a covid-19 outbreak. AI-assisted CT was integrated into the mobile CT scanner and images could be quickly transmitted via a 5G network back to medical experts in Wuhan for diagnoses. This AI-assisted system helps any hospital in epidemic center quickly get additional clinical support from external radiologists and experts, which can greatly improve the accuracy of diagnosis.7
For AI, data is life. However, data barriers exist in every corner of the world, blocking the flow of data among medical institutions, administrative regions, and countries. The laws and ethics of data security need to be considered before implementing any AI technology. During the pandemic, balancing laws, ethics, and trust has been an unprecedented common challenge for all humankind.
A recent review analysed the published AI tools related to the diagnosis and prognosis of covid-19.8 In this review, information bias was observed in data set construction, goal setting and feature selection, and data labeling, resulting in algorithms that performed poorly. Moreover, the urgency of the pandemic has meant most of the AI tools, which were developed rapidly, have not performed satisfactorily in external validation leading to insufficient replication of predictions. The existence of bias and a lack of replicability has resulted in limited of clinical practical value for these developed tools. I believe encouraging data sharing while protecting data security is the key to overcoming these problems, increasing the diversity of data and both the reliability and credibility of artificial intelligence tools.
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)4 is common challenge for all humankind. In facing this global health crisis, solidarity is a source of strength and trust is the foundation of solidarity. Establishing a mechanism to improve trust and solidarity is a huge challenge that isn’t just technical. China’s approach has been to realise comprehensive data sharing in its successful fight against the covid-19. In addition, universal participation in the lockdown disrupted the virus transmission to the greatest extent at an early stage in China.
No country can defeat SARS-CoV-2 alone, and maintaining close communication could be a step in the right direction. Scientists in various fields need to communicate and cooperate more closely ensure scientific research achieves its goals. For example, the development and validation of artificial intelligence diagnostic models requires the full participation of clinicians to ensure the clinical applicability of models. In the future, more communication channels need to be opened up, including global covid-19 origin tracing, optimization and upgrading of artificial intelligence systems, and vaccine development and production.
Drawing lessons from this pandemic, I believe that improving trust and solidarity could help release the full power and potential of AI. In turn this could support our shared aspiration of protecting people’s lives.
Xing-Huan Wang is Director of Leishenshan Hospital in Wuhan and Center for Evidence-Based and Translational Medicine, Zhongnan Hospital of Wuhan University, China.
Competing interests: I have read and understood BMJ policy on declaration of interests and have nothing to declare.
This article is part of our Artificial Intelligence and covid-19 collection.
- Ginsberg J, Mohebbi MH, Patel RS, et al. Detecting influenza epidemics using search engine query data. Nature 2009;457(7232):1012-14. doi: 10.1038/nature07634 pmid: 19020500
- Nagendran M, Chen Y, Lovejoy CA, et al. Artificial intelligence versus clinicians: systematic review of design, reporting standards, and claims of deep learning studies. BMJ 2020;368:m689. doi: 10.1136/bmj.m689 pmid: 32213531
- Liu X, Faes L, Kale AU, et al. A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: a systematic review and meta-analysis. The Lancet Digital Health 2019;1(6):e271-e97. doi: 10.1016/S2589-7500(19)30123-2 pmid: 33323251
- Jin YH, Zhan QY, Peng ZY, et al. Chemoprophylaxis, diagnosis, treatments, and discharge management of COVID-19: An evidence-based clinical practice guideline (updated version). Mil Med Res 2020; 7:41. doi: 10.1186/s40779-020-00270-8.
- Nakamoto I, Wang S, Guo Y, et al. A QR Code-Based Contact Tracing Framework for Sustainable Containment of COVID-19: Evaluation of an Approach to Assist the Return to Normal Activity. JMIR mHealth and uHealth 2020;8(9):e22321. doi: 10.2196/22321pmid: 32841151
- Wang B, Jin S, Yan Q, et al. AI-assisted CT imaging analysis for COVID-19 screening: Building and deploying a medical AI system. Appl Soft Comput 2021;98:106897-97. doi: 10.1016/j.asoc.2020.106897 pmid: 33199977
- Xinhua. Zhongnan Hospital of Wuhan University sends vehicle-carried mobile CT scanner to Hebei. 2021. http://en.people.cn/n3/2021/0111/c90000-9808012.html (April 1, 2021 accessed)
- Roberts M, Driggs D, Thorpe M, et al. Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19 using chest radiographs and CT scans. Nature Machine Intelligence 2021;3(3):199-217. doi: 10.1038/s42256-021-00307-0