Project | 05: Physiology-Based Dynamic Fatigue Modeling to Measure Physical Stress
This project aims to introduce a method for quantifying muscle fatigue, which impacts the health of many construction workers. Construction work requires unstructured muscle movements that escalate muscle fatigue, which leads to musculoskeletal disorders, accidents, and loss of productivity. We have been working to develop a muscle fatigue estimation model that incorporates data from a focus group of construction workers concerning metabolic muscle activity during specific upper-body movements. We already demonstrated that this model for elbow fatigue has a high correlation with experimental data, and we continue to develop a model for our muscles fatigue. These models will be designed to identify muscle fatigue in construction workers before it reaches a critical level of debilitation.
Project | 04: Assessing Workers’ Mental and Physical Status Using their Physiological Signals Collected from a Wristband-type Wearable Sensor
In addition to the EEG, we have investigated the feasibility of another three physiological signals (i.e., EDA, ST, and PPG) that are collected from a wristband-type biosensor to recognize workers’ mental and physical stress. From these signals, we calculated diverse metrics (i.e., Heart Rate (HR), Heart Rate Variability (HRV), Inter-Beat-Interval (IBI), Electrodermal Level (EDL), and Electrodermal Response (EDR)) to understand the potential of these signals to assess workers’ stress. Then, we developed a framework to learn the patterns of physiological signals while subjects are exposed to various stressors by applying a supervised learning algorithm (e.g., Gaussian Kernel Support Vector Machine). To examine the performance of the proposed framework, we collected physiological signal from several construction workers in the field. The proposed framework in this project was applied to this dataset and resulted in a mental stress-prediction accuracy of around 85% and physical stress recognition accuracy of around 90%. The results confirmed the potential of the proposed framework for enhancing workers’ health, safety, and productivity through early detection of occupational stressors at actual sites.
Project | 03: Quantifying Construction Workers’ Emotion and Mental Stress Using their Brain Waves
After capturing high-quality EEG signals in the field. We expanded our study of EEG monitoring on construction sites to include the development of several algorithms for identifying the type of stress being experienced by construction workers. Measuring EEG data through specialized sensors installed in construction hats, we developed a bipolar emotion model that analyzes degrees of arousal and happiness based on the EEG data. In addition, we devised a framework that recognizes stress signals in real time through the continual update of emotion recognition algorithms, which was accomplished through online Multi-Task Learning (OMTL). This framework is a novel because it recognizes stress signals beyond those that have been predetermined. Thus, this project introduces a novel method for identifying and proactively addressing construction workers’ stress before it results in workplace accidents.
Project | 02: Capturing High-Quality EEG Signals On-Site Using a Wearable EEG Device
In this project, we turned our attention to monitoring the psychological repercussions of construction work, another important aspect of construction safety. On-the-job exhaustion and stress make workers more susceptible to accidents, so monitoring workers’ brain waves for signs of distress contributes fundamentally to their safety. However, prior to this research, EEG monitoring had not been applied in the construction sector because EEG signals are easily disrupted by the high mobility of construction tasks and different signal artifacts. We developed a method that reduces the interference caused by a moving subject. We designed filters to eliminate extrinsic and intrinsic artifacts, allowing the seclusion of EEG signals. After testing this method on construction workers around the U.S., we confirmed that it produces high-quality EEG waves even in noisy environments.
Project | 01: Quantifying Construction Workers’ Gait and Postural Stability
Falls are one of the most common causes of injury and death on construction sites, as discussed above, and while safety regulations have reduced fatalities, the statistics remain unacceptably high. To understand more fully which movements result in high fall risks, In this project, we developed a framework that measures workers’ gait patterns through a wearable Inertial Measurement Unit (IMU) and then mathematically analyzes and recognizes the abnormal gait patterns and near-miss fall incidents through calculating the Maximum Lyapunov Exponents (Max LE) before the actual fall incidents occur.
In this project, we extended the fall risk assessment methodology that was developed for predicting fall risk to the measurement and analysis of construction workers’ postural stability. Because stationary movement also results in falls on construction sites, we created a framework that objectively measures the fall risk associated with inert tasks through 2 distinct metrics. These metrics together determine when workers’ posture becomes unstable and correlate the results with worksite tasks, enabling the enforcement of selective safety procedures. We evaluated this framework by testing it on a large sample of diverse construction workers. As part of this evaluation, we identified stationary tool loading via a safety harness as a high-risk task. This framework is one of the first to measure the postural stability of construction workers on-site since previous attempts required the use of a force plate, a device employed in the medical sector that is prohibitively heavy and expensive.