Leveraging natural language processing models using a large volume of text data in the construction safety domain offers a unique opportunity to improve understanding of safety accidents and the ability to learn from them. However, little effort has been made to date in regard to utilizing large language models for the prediction of accident types that can help to prevent and manage potential accidents. This research aims to develop a model for predicting the six types of accidents (caught-in-between, cuts, falls, struck-by, trips, and others) by employing transfer learning with a fine-tuned generative pre-trained transformer (GPT). Additionally, to enhance the interpretability of the fine-tuned GPT model, a method for saliency visualization of input text was developed to identify words that significantly impact prediction results. The models were evaluated using a comprehensive dataset comprising 15,000 actual accident records. The results indicate that the suggested model for detecting the six accident types achieves 82% accuracy. Furthermore, it was observed that the proposed saliency visualization method can identify accident precursors from unstructured free-text data of construction accident reports. These results highlight the advancement of the generalization performance of large language processing-based accident prediction models, thereby proactively preventing construction accidents.

As the worldwide population is aging, the demands of aging-in-place are also increasing and require smarter and more connected cities to keep mobility independence of older adults. However, today's aging built environment often poses great environmental demands to older adults' mobility and causes their distresses. To better understand and help mitigating older adults' distress in their daily trips, this paper proposes constructing the digital twin city (DTC) model that integrates multimodal data (i.e., physiological sensing, visual sensing) on environmental demands in urban communities, so that such environmental demands can be considered in mobility planning of older adults. Specifically, this paper examines how data acquired from various modalities (i.e., electrodermal activity, gait patterns, visual sensing) can portray environmental demands associated with older adults' mobility. In addition, it discusses the challenges and opportunities of multimodal data fusion in capturing environmental distresses in urban communities.

Urban built environments often include many negative stimuli (e.g., unleashed dogs, dead animals, litter, graffiti, abandoned vehicles) that are linked with stress symptomatology among urban populations. Biosignals (e.g., electrodermal activity, gait patterns, and blood volume pulse) can help assess pedestrian distress levels induced by negative environmental stimuli by overcoming the measurement limitations of traditional self-reporting methods and field observations. Despite their potential, biosignals from naturalistic outdoor environments are often contaminated by uncontrollable extraneous factors (e.g., movement artifacts, physiological reactivity due to unintended stimuli, and individual variability). Thus, more quantitative evidence and novel methodological approaches are required to accurately capture pedestrian environmental distress resulting from negative environmental stimuli. In this context, we investigate the interplay between pedestrians’ biosignal data and image-based data (built environment feature information and perceptual distress levels identified from images) in a machine learning model. Results from the statistical model estimated with the biosignal data demonstrated significant physiological responses to the negative environmental stimuli. The use of the features from image-based data increased the prediction accuracy of the computational model. This method can be applied to geospatial intelligence, further advancing built environmental assessments and evidence-based approaches to promote walking and walkable communities.

Burgeoning attention has recently been paid to measuring and assessing the effects of urban built environments on pedestrians’ physiological signals (e.g., gait patterns, blood volume pulse, and electrodermal activity). Previous studies have attempted to use physiological signals obtained from naturalistic ambulatory settings to assess negative environmental stimuli (e.g., the presence of broken houses, unstable sidewalks, abandoned cars, etc.), but several unanswered questions remain regarding whether physiological signals, which include various confounding factors (e.g., noise, movement artifacts, and individual variability), can capture the impact of negative environmental stimuli on pedestrians. Additionally, our own previous research proposed a saliency detection method to capture the changes in physiological signals caused by negative environmental stimuli. However, the effect of diverse physiological signal patterns on the proposed saliency detection method is still uncertain and needs further analysis (such as analysis regarding the sensitivity of initial input data for signal segmentation, the validity of aggregation across all subjects’ data, etc.). In this context, this paper aims to (1) examine the direct association between pedestrians’ physiological signals and an isolated negative environmental stimulus and (2) test the use of the saliency detection method to crowdsource identification of pedestrians’ environmental distress by using data sets from naturalistic ambulatory settings. The experimental and statistical results, attained from physiological signals, present distinct physiological responses to negative environmental stimuli, and the saliency detection method is also effectual in capturing prominent local patterns. We envisage that the outcome of this study will provide opportunities for advancing urban built environment assessment, especially in terms of promoting neighborhood walkability, increasing feelings of comfort and satisfaction in the urban space, and augmenting residents’ quality of life.

One of the primary causes of fires at construction sites is welding sparks. Fire detection systems utilizing computer vision technology offer a unique opportunity to monitor fires in construction sites. However, little effort has been made to date in regard to real-time tracking of small sparks that can lead to major fires at construction sites. In this study, a novel method is proposed to detect welding sparks in real-time contour detection with deep learning parameter tuning. An automatic parameter tuning algorithm employing a convolutional neural network was developed to identify the optimum hue saturation value. Additional filtering methods regarding the non-welding zone and a contour area-based filter were also newly developed to enhance the accuracy of welding spark prediction. The method was evaluated using 230 welding spark images and 104 videos. The results obtained from the welding images indicate that the suggested model for detecting welding sparks achieves a precision of 74.45% and a recall of 63.50% when noise images, such as flashing and reflection light, were removed from the dataset. Furthermore, our findings demonstrate that the proposed model is effective in capturing the number of welding sparks in the video dataset, with a 95.2% accuracy in detecting the moment when the number of welding sparks reaches its peak. These results highlight the potential of automated welding spark detection to enhance fire surveillance at construction sites.

Enhancing safety communication within the construction industry is of paramount importance due to its potential in curtailing occupational injuries and improving the overall well-being of construction field workers. While the importance of improved communication is apparent, few studies have been focused on identifying the factors that positively influence communication, particularly in the context of safety. Especially in the case of foreign construction field workers (FCFWs), who often face communication challenges stemming from language and cultural differences, performing labor tasks in harsh and constantly changing environments is contributing significantly to the increasing rate of industrial accidents. Therefore, this study is aimed at investigating the vital factors that impact safety communication among FCFWs. A phenomenological qualitative method was applied to determine the vital factors influencing the safety communication among FCFWs. On applying the analytical hierarchy process, the factors and their importance were determined, and the vital factors were identified. Quantitative assessment through the analytic hierarchy process (AHP) established that extrinsic motivation (weight: 0.513), management communication style (0.264), and visible safety information (0.127) significantly overshadow other factors in safety communication effectiveness, validating their pivotal roles. However, a major limitation is that interview data were collected from workers of only four nationalities. Future studies should address this by expanding the range of nationalities included to enhance the diversity of experiences and perspectives from a broader variety of foreign construction field workers.