I. Introduction
The proper use of medical personal protective equipment (MPPE) is essential for frontline healthcare workers (HCWs) handling emerging highly infectious diseases worldwide (e.g., COVID-19 [1], Ebola [2] and Eastern Respiratory Syndrome [3]). However, studies have shown that MPPE protocol (the standard process of MPPE donning and doffing) deviations are common, which could result in self-contamination and transmission [4], [5]. According to statistics as of May 2020, more than 152,888 HCWs are infected in 130 countries during the COVID-19 pandemic [1]. Except for the shortage of MPPE, the complexity of the standard protocol and the inadequate training of MPPE users also possibly contribute to this situation [6]. In the related guidance, the Centers for Disease Control and Prevention (CDC) recommends having a trained observer or a mirror for auxiliary MPPE donning and doffing monitoring [7]. During times such as the global epidemic of COVID-19 [8], the shortage of experienced HCWs is growing [9], [10]. Correspondingly, the monitoring based on medical staff could become expensive; at the same time, the demand for teaching MPPE donning and doffing is on the rise [11]. Applying cutting-edge artificial intelligence (AI) technologies like computer vision and cognitive computing might help alleviate this issue. Briefly, the AI system can reduce the contamination risk and the workload of HCWs in two aspects: 1) being used as an accompanying assistant providing instructions and risk warnings of incorrect behavior during the MPPE donning and doffing process; 2) being used as a low labor cost coach providing updated knowledge of safety protocol for MPPE donning and doffing.