JMIR Publications

ABSTRACT

Background:

As the epidemic situation of COVID-19 worsened, the burden of Quarantine Stations (Q stations) outside of emergency departments at every hospital increased day by day. To prepare for the screen workload inside the Q station, all staff with medical licenses were required to support the working shift.

Objective:

Therefore, the need to simplify the workflow and decision-making process for physicians and surgeons from all subspecialist fields was necessary.

Methods:

We report an observational study for emerging pandemic COVID-19 disease with constitutively 643 patients. The artificial intelligence (AI) trilogy, 1) smart Q station diversion, 2) AI assisted image interpretation and 3) built-in clinical decision-making algorithm on tablet computer was applied to shorten the quarantine survey and processing time during the COVID-19 pandemic period.

Results:

This facilitated the processing of suspected cases; with or without symptoms, travel, occupation, and contact or clustering histories, all performed by a tablet computer device. A separate AI mode function that quickly recognizes pulmonary infiltrates on chest x-rays was merged into the smart clinical assisting system (SCAS), and subsequently trained this model with COVID-19 pneumonia cases from the GitHub open source dataset. The detection rates of 93.2% in posteroanterior and 45.5% in anteroposterior chest x-rays, respectively. The SCAS algorithm was adjusted continuously following the frequently updated Taiwan Center for Disease Control public safety guidelines for faster clinical decision-making. Our ex vivo study demonstrated the efficiency of 75% alcohol disinfection on the tablet computer surface for 20 μL positive SARS-CoV-2 virus solution. The initial crossing point of the cycle value by real time-polymerase chain reaction as 34 became 0 after 1 and 2 times of disinfection procedures. Compared with the conventional ER track (n = 281), the survey time at the clinical Q station (n=362) was significantly shortened [median survey time (95% C.I.) at the ER; 153 (138-163) vs. clinical Q station of 52 (46-56) minutes, p<0.0001]. Furthermore, the use of this AI application for the Q station reduced survey times significantly [median survey time (95% C.I.) without AI; 100.5 (80-120) vs. with AI; 45.5 (42-51); p<0.0001]

Conclusions:

This AI trilogy improves medical care workflow by shortening the quarantine survey and processing time, especially for an emerging epidemic infectious disease. Clinical Trial: An observation study, not clinical trial design