Background Shortcomings of the current school-based infectious disease syndromic surveillance system (SSS) in China include relying on school physicians to collect data manually and ignoring the health information of students in attendance. Objective This study aimed to design and implement an influenza SSS based on the absenteeism (collected by face recognition) and temperature of attending students (measured by thermal imaging). Methods An SSS was implemented by extending the functionality of an existing application. The system was implemented in 2 primary schools and 1 junior high school in the Yangtze River Delta, with a total of 3535 students. The examination period was from March 1, 2021, to January 14, 2022, with 174 effective days. The daily and weekly absenteeism and fever rates reported by the system (DAR1 and DFR; WAR1 and WFR) were calculated. The daily and weekly absenteeism rates reported by school physicians (DAR2 and WAR2) and the weekly positive rate of influenza virus (WPRIV, released by the Chinese National Influenza Center) were used as standards to evaluate the quality of the data reported by the system. Results Absenteeism reported by school physicians (completeness 86.7%) was 36.5% of that reported by this system (completeness 100%), and a significant positive correlation between them was detected (r=0.372, P=.002). When the influenza activity level was moderate, DAR1s were significantly positively correlated among schools (rab=0.508, P=.004; rbc=0.427, P=.02; rac=0.447, P=.01). During the influenza breakout, the gap of DAR1s widened. WAR1 peaked 2 weeks earlier in schools A and B than in school C. Variables significantly positively correlated with the WPRIV were the WAR1 and WAR2 of school A, WAR1 of school C, and WFR of school B. The correlation between the WAR1 and WPRIV was greater than that between the WAR2 and WPRIV in school A. Addition of the WFR to the WAR1 of school B increased the correlation between the WAR1 and WPRIV. Conclusions Data demonstrated that absenteeism calculation based on face recognition was reliable, but the accuracy of the temperature recorded by the infrared thermometer should be enhanced. Compared with similar SSSs, this system has superior simplicity, cost-effectiveness, data quality, sensitivity, and timeliness.

BACKGROUND Absenteeism has been shown to be a valid indicator for influenza surveillance. In China, it is common for students to attend school while ill, so a surveillance system that focuses only on absentees may lead to error. Therefore, the development of a system for both absenteeism and attendance is necessary. OBJECTIVE The objective is to design and implement an influenza symptom surveillance system (SSS) based on absenteeism (collected by face recognition) and temperature of attending students (measured by infrared thermometer), and to evaluate the effectiveness of the system. METHODS An influenza SSS was deployed by extending the functionality of an existing application for student health management. The system used infrared thermometers for temperature screening while counting absent students by facial recognition. The operation of the system was investigated in participating schools in China, the weekly absenteeism and fever rates (WAR and WFR) were calculated and compared with the weekly positive rate of influenza virus (WPRIV) released by the China National Influenza Center to inspect the data reliability and operation feasibility of this system. The system was implemented in two primary schools and one junior high school in the Yangtze River Delta, with a total of approximately 3,500 students participating. The period was from March 1, 2021 to January 14, 2022, with 174 effective days. RESULTS A significant positive correlation between WAR and WPRIV (r=0.868, p<0.001) was detected. When the influenza activity level was low, the WAR was significantly positively correlated among schools. As the influenza activity level increased, the WAR gap among schools gradually increased, and the peak was reached about one week earlier in primary schools than in the junior high school. A significant positive correlation between WFR and WPRIV (r=0.532, p<0.05) was also detected. When WFR was included, the proposed system detected influenza outbreaks up to three weeks earlier than traditional surveillance systems. CONCLUSIONS The feasibility of the proposed system, which calculates absenteeism through face recognition and the temperature of attending students with infrared thermometers, was shown in the present paper. Compared with similar existing systems, this system has advantages of simplicity, acceptability, security, sensitivity and timeliness. CLINICALTRIAL The data used in this study were anonymized, so the Tongji University Review Board designated this study as non-human subject research.