The system then makes predictions based on past data of students who stopped attending classes and shows how close a student is to the risk of chronic nonattendance. It color-codes the probability figures for each student, ranging from the highest "red" to "pink," "orange" and the lowest "yellow." During the trial run at 18 primary and junior high schools, the program concluded that a total of 1,193 students were at high risk. Based on the outcome, teachers judged that 265 of the total needed to receive priority in receiving assistance, taking into account their behavior and other factors, according to the education board. To deal with privacy concerns, the education board established rules to safeguard personal data and to prevent the AI-produced predictions from being used in any discriminatory treatment of students.