A neural network-based infection screening system that uses vital signs and percutaneous oxygen saturation for rapid screening of patients with influenza

Objective: Influenza is a highly infectious viral disease, which occurs epidemically almost every winter in Japan. Rapid screening of patients with suspected influenza in places of mass gathering is important to delay or prevent transmission of the infection. The aim of this study was to assess the effectiveness of our newly developed infection screening system that employed vital signs and percutaneous oxygen saturation (SpO2) as parameters in a clinical setting. Methods: Since SpO2 accurately reflects respiratory status during influenza virus infection, we upgraded our previous system by adding SpO2 as a new parameter to improve the screening accuracy. This system instantly measures SpO2 and vital signs (i.e., heart rate, respiration rate, and facial temperature), which automatically detects infected individuals via a neural network-based nonlinear discriminant function using these derived parameters. We tested the system on 45 patients with seasonal influenza (35.8 ̊C < body temperature < 40.0 ̊C, 18 35 years) and 64 normal control subjects (35.0 ̊C < body temperature < 37.5 ̊C, 18 30 years) at Japan Self-Defense Central Hospital in 2012. Results: The system identified 40/45 patients with influenza and 60/64 normal control subjects, and provided sensitivity, specificity, and positive and negative predictive value (PPV, NPV) of 88.8%, 93.8%, 90.9%, and 92.3%, respectively. By including SpO2 as a screening parameter, we achieved superior sensitivity and NPV compared to that reported in our previous paper (sensitivity = 88%; NPV = 82%). Conclusions: Our results suggest that SpO2 is a good screening parameter that improves the accuracy of infection screening. The proposed system has the potential to efficiently identify infected individuals, thereby delaying or preventing the spread of infection during epidemic seasons.