A Smart Health Device to Measure Stress Levels Based on the Internet of Things Using the K-Nearest Neighbor Algorithm

Abstract

Mental health plays an important role in daily life. However, many factors can affect mental health, one of which is stress. Students are one group that is prone to stress. Academic stress is common among students. Currently, physiological stress screening devices are available, but most of them still work separately and are quite expensive. Therefore, this study aims to design an IoT-based stress detection device with a KNN algorithm that can measure physiological symptoms to detect stress levels in a practical and economical manner. This research is a type of engineering research, which involves the process of designing hardware and software for the system in an IoT-based stress level detection tool. The tool is designed to measure three physiological parameters, namely skin conductance, heart rate, and body temperature. Sensor data is processed using the KNN algorithm to classify stress levels into four categories, namely normal, mild, moderate, and severe. The results are displayed on an OLED and ThingSpeak platform so that they can be accessed remotely through IoT integration. Testing was carried out by collecting stress condition data from several subjects. The test results show that the device has an average accuracy and precision value of 83.33% to 99.82%. In addition, the average prediction computation time produced by the system was only 0.44 seconds, this computation time falls within an acceptable range for real-time applications. Thus, this stress level detection tool is expected to be an alternative solution and facilitate remote condition monitoring through the IoT feature.