Machine Learning-Based Fatigue Prediction in Construction Workers Using Autonomous Cardiovascular Efficiency and Wearable Sensor Data
Faculty Mentor
Dr. Mohammadsoroush Tafazzoli
Location
Russell Union Ballroom
If Other was choses above, please indicate your topic area here:
Construction Safety Management
Type of Research
On-going
Session Format
Poster Presentation
College
Allen E. Paulson College of Engineering & Computing
Department
Civil Engineering
Abstract
Construction workers are frequently exposed to high physical and thermal loads, leading to fatigue-related safety risks. While traditional safety protocols rely on subjective reporting, this study utilizes wearable sensors to objectively quantify the impact of rest-break strategies on 34 workers. We analyzed Heart Rate (HR) and Heart Rate Variability (HRV), specifically RMSSD and SDNN, alongside a derived metric: Autonomous Cardiovascular Efficiency (ACE). Our results indicate that a Multiple Short Break (MSB) strategy significantly preserves physiological homeostasis (p < 0.001) compared to a One Long Break (OLB) approach, which showed a 48% decline in ACE. Furthermore, we developed a Random Forest predictive model that achieved an accuracy of 71% (R^2 = 0.993) in classifying fatigue states. These findings demonstrate the feasibility of integrating wearable biometrics with machine learning to provide real-time, objective safety monitoring on-site.
Program Description
.
Start Date
4-23-2026 10:00 AM
End Date
4-23-2026 12:00 PM
Recommended Citation
Haq, Iffat, "Machine Learning-Based Fatigue Prediction in Construction Workers Using Autonomous Cardiovascular Efficiency and Wearable Sensor Data" (2026). GS4 Student Scholars Symposium. 66.
https://digitalcommons.georgiasouthern.edu/research_symposium/2026/2026/66
Machine Learning-Based Fatigue Prediction in Construction Workers Using Autonomous Cardiovascular Efficiency and Wearable Sensor Data
Russell Union Ballroom
Construction workers are frequently exposed to high physical and thermal loads, leading to fatigue-related safety risks. While traditional safety protocols rely on subjective reporting, this study utilizes wearable sensors to objectively quantify the impact of rest-break strategies on 34 workers. We analyzed Heart Rate (HR) and Heart Rate Variability (HRV), specifically RMSSD and SDNN, alongside a derived metric: Autonomous Cardiovascular Efficiency (ACE). Our results indicate that a Multiple Short Break (MSB) strategy significantly preserves physiological homeostasis (p < 0.001) compared to a One Long Break (OLB) approach, which showed a 48% decline in ACE. Furthermore, we developed a Random Forest predictive model that achieved an accuracy of 71% (R^2 = 0.993) in classifying fatigue states. These findings demonstrate the feasibility of integrating wearable biometrics with machine learning to provide real-time, objective safety monitoring on-site.
