Estimating the amount of energy spent by physiological measurement methods

Estimating employees' break times is crucial for increasing workforce productivity and achieving efficient personnel management. The evaluation of these periods varies depending on the dynamics of the work environment and the employees. However, generally, employees' daily break times fall within the framework determined by the employer and labor laws. In this thesis, machine learning models are employed to determine the adequacy of existing break times and predict the necessary break times by considering employees' physiological measurements, perceived subjective effort, and ergonomic risk assessment scores. In this context, data were collected from 110 employees performing tasks in production departments with different levels of difficulty. During a 30-minute working period, values of Heart Rate (HR) and energy consumption (Kcal) were obtained using a smartwatch and its application, along with blood oxygen level (SpO₂) values via a pulse oximeter. These data were included in the dataset as employees' physiological evaluations during work. To identify the physical workload from an ergonomic perspective, measurements were conducted using the Rapid Entire Body Assessment (REBA) method. Additionally, perceived subjective effort levels were measured using the Modified Borg Scale. These data collected during work reflect the effort and stress levels expended by the individual during working hours. However, to calculate break times, basal metabolism data during rest periods are also necessary. Therefore, energy expenditure (Kcal) and HR values during rest periods were measured and recorded. Based on the data obtained from 110 different tasks and employees, new break times for light, moderate, and heavy work groups were determined using various machine learning models as 21.42 mins, 21.52 mins, and 22.87 mins, respectively. It is recommended to reconsider the currently implemented 15-minute break time based on the results of this study.