Proceedings of the 11th International Applied Business and Engineering Conference, ABEC 2023, September 21st, 2023, Bengkalis, Riau, Indonesia

Research Article

Mobile-based Stress Level Detection using Tree-Based Machine Learning Algorithms

Download48 downloads
  • @INPROCEEDINGS{10.4108/eai.21-9-2023.2342968,
        author={Kartina Diah Kesuma Wardani and Tony Wijaya and Putri Madona and Juni Nurma Sari and Yuliska Yuliska},
        title={Mobile-based Stress Level Detection using Tree-Based Machine Learning Algorithms},
        proceedings={Proceedings of the 11th International Applied Business and Engineering Conference, ABEC 2023, September 21st, 2023, Bengkalis, Riau, Indonesia},
        publisher={EAI},
        proceedings_a={ABEC},
        year={2024},
        month={2},
        keywords={xgboost stress prediction mobile},
        doi={10.4108/eai.21-9-2023.2342968}
    }
    
  • Kartina Diah Kesuma Wardani
    Tony Wijaya
    Putri Madona
    Juni Nurma Sari
    Yuliska Yuliska
    Year: 2024
    Mobile-based Stress Level Detection using Tree-Based Machine Learning Algorithms
    ABEC
    EAI
    DOI: 10.4108/eai.21-9-2023.2342968
Kartina Diah Kesuma Wardani1,*, Tony Wijaya1, Putri Madona1, Juni Nurma Sari1, Yuliska Yuliska1
  • 1: Politeknik Caltex Riau
*Contact email: diah@pcr.ac.id

Abstract

Healthcare and physiological specialists have the capability to ascertain if an individual is undergoing a state of stress or not. Automatic detection of stress can minimize the risk of health problems and improve people's well-being because by detecting stress automatically allows for early intervention and prevention of more serious health problems. It’s utilize physiological signals and machine learning algorithm to automate the detection of stress levels in individuals. The current research is developing stress level prediction using the XGBoost algorithm in mobile-based stress detection based on four variables: blood pressure, heart rate per minute, body temperature, and Galvanic Skin Resistance (GSR). The mobile application receives signals from these four variables through data published using an MQTT broker from the stress detection device. Using XGBoost algorithm, the data then automatically predicts the stress level and displays the results on the mobile application interface. The model accuracy of XGBoost algorithm used in the aplication is 98% with an f1-score of 97%. Futhermore the results of usability testing of the application carried out on 35 people obtained 91% usability, 93.33% ease of use, 91% user interface, and 91% satisfaction.