A BIBLIOMETRIC ANALYSIS OF DIGITAL TECHNOLOGIES USE IN CONSTRUCTION HEALTH AND SAFETY

PurposeConstruction is a risky industry. Therefore, organizations are seeking ways towards improving their safety performance. Among these, the integration of technology into health and safety leads to enhanced safety performance. Considering the benefits observed in using technology in safety, this study aims to explore digital technologies' use and potential benefits in construction health and safety.Design/methodology/approachAn extensive bibliometrics analysis was conducted to reveal which technologies are at the forefront of others and how these technologies are used in safety operations. The study used two different databases, Web of Science (WoS) and Scopus, to scan the literature in a systemic way.FindingsThe systemic analysis of several studies showed that the digital technologies use in construction are still a niche theme and need more assessment. The study provided that sensors and wireless technology are of utmost importance in terms of construction safety. Moreover, the study revealed that artificial intelligence, machine learning, building information modeling (BIM), sensors and wireless technologies are trending technologies compared to unmanned aerial vehicles, serious games and the Internet of things. On the other hand, the study provided that the technologies are even more effective with integrated use like in the case of BIM and sensors or unmanned aerial vehicles. It was observed that the use of these technologies varies with respect to studies conducted in different countries. The study further revealed that the studies conducted on this topic are mostly published in some selected journals and international collaboration efforts in terms of researching the topic have been observed.Originality/valueThis study provides an extensive analysis of WoS and Scopus databases and an in-depth review of the use of digital technologies in construction safety. The review consists of the most recent studies showing the benefits of using such technologies and showing the usage on a systemic level from which both scientists and practitioners can benefit to devise new strategies in technology usage.

3 technology in construction safety has not been assessed well in many of the construction safety studies. However, the industry is dynamic and there is a growing need for transformation and adoption of new technologies, which is further articulated as Construction 4.0. Therefore, this study provides a bibliometrics review of existing studies on the topic of digital technologies in construction health and safety. The background of the research and methodology used is explained in the following sections.

RESEARCH BACKGROUND
The focus on construction safety research has considerably increased in recent years (Jin et al., 2019). Hence, a systematic literature review becomes vital to reveal novel findings, identify research areas and assist construction stakeholders to become familiar with the recent trends and implementations (Liang et al., 2020). Liang et al. (2020) conducted a bibliometric analysis to investigate construction safety management. They utilized data from the Web of Science database, including the time range from 1991 to 2016, and utilized CiteSpace software. Based on their analysis, they reported that one of the main topics within the construction safety management area is technology-driven management, which focuses on safety training, planning, and monitoring subtopics. Jin et al. (2019) reviewed construction safety through bibliometric and scientometric analyses. They utilized the Scopus database and practiced VOSViewer software. Their research concluded that the use of information technology for construction safety is a fast-growing and progressing area for research. They further implied that Building Information Modeling (BIM), Virtual Reality (VR), Augmented Reality (AR), and data analysis have recently attracted significant attention compared to other technologies (Jin et al., 2019).
Following up the technological developments and adopting innovative approaches is one of the most effective methods to improve safety performance in the construction industry (Skibniewski, 2014). In the last 20 years, the use of digital technologies to reduce worker injuries and fatalities has received considerable attention . As the number of  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58 59 60 e e r i n g , C o n s t r u c t i o n a n d A r c h i t e c t u r a l M a n a g e m e 4 different safety applications in the construction industry has taken place in recent years, it has been observed that safety performance has increased considerably (Hinze et al., 2013). The use of information technologies provides many advantages such as data acquisition, fast access, active transfer, full presentation, and convenient information storage. Digital technologies help to improve safety planning, real-time hazard detection and warning, and management of accumulated safety knowledge (B. H. W. . Many researchers have already investigated the application of these technologies before, during, and after construction accidents and they mentioned the successful results (Tsai, 2016). This brings the need for technology use to eliminate issues associated with the construction health and safety (Skibniewski, 2014). A systematic review of the literature on the use of these technological applications in construction health and safety is important in terms of revealing current research areas and related research findings and catching new trends (Akinlolu et al., 2020;Liang et al., 2018;Zhou et al., 2013).
There is a limited number of studies reviewing the use of technologies in the construction health and safety domain. Among these, Zhou et al. (2013) conducted an extensive literature review of construction safety management for approximately 30 digital technologies. Skibniewski (2014) studied information technology in construction safety.  provided a literature review investigating 15 digital technologies used for construction safety management. Akinlolu et al. (2020) performed a bibliometric analysis analyzing digital technologies in general with the help of VOSViewer software.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h  i  t  e  c  t  u  r  a  l  M  a  n  a  g  e  m  e 7 restriction, it uses both databases for review (WoS and Scopus).
The discussion section of this study is enriched with many resources along with providing directions for future research. In this study,  Figure  10). Moreover, in this study, a very broad review of mostly used technologies is provided (Table 5).  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h  i  t  e  c  t  u  r  a  l  M  a  n  a  g  e  m  e   8 Even though there are some other studies examining construction technologies used in the construction health and safety (Akinlolu et al., 2020;B. H. W. Guo et al., 2017;Skibniewski, 2014;Zhou et al., 2013), the number of studies conducted research in this area is scarce. These studies are listed along with their data collection and analysis methodology, main findings, and differences between this research and previous similar review are shown in Table 1. The existing studies have some limitations in terms of time range coverage and scope. One major drawback is that most of the existing reviews do not cover recent years in the analysis. Moreover, the existing reviews have a narrow scope in terms of keywords. For example, several studies investigated the use of specific technologies such as BIM (Hire et al., 2022;Akram et al., 2019) and sensing and warning technologies (Antwi-Afari et al., 2019) but a broader array of technologies has not been thoroughly investigated. To fill this gap, this study goes a step further by covering studies investigating a wide range of technologies in construction health and safety up to 2021. In this respect, this study aims to review and analyze the existing literature and to discuss and reveal the characteristics of the previous studies. In this context, the main objectives are: (1) to identify digital technologies used for construction health and safety (2) to identify the most frequently used keywords, the most influential publication sources, academics, documents, countries, and institutions (3) to reveal the relationships of keywords, scholars, and documents with the help of bibliometrics analysis To achieve the above-mentioned objectives, the study is structured as follows. This study uses a scientific mapping approach by examining existing studies as part of the research methodology that can be seen in section three. The fourth section shows analysis along with results. The fifth section details and discusses the research findings. The last section presents the conclusions of the study.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h  i  t  e  c  t  u  r  a  l  M  a  n  a  g  e  m  e   9 This study adopts a literature review-based approach to examine and summarize the studies on digital technologies used for occupational health and safety in the construction industry. Since it is a growing and multidisciplinary area, conducting a manual review becomes laborious for some.

RESEARCH METHODOLOGY
Besides, manual review tends to be more subjective and error-prone. Bibliometrics is the application of mathematical and statistical methods to investigate the literature quantitatively (Wang et al., 2021). Bibliometric analysis helps to determine the size of studies in a particular scientific field and provides a complete view of the field. It is also useful for researchers, journal editors, and referees as it transforms qualitative data into quantitative data (Baraibar-Diez et al., 2020;Wallin, 2005).
Hence, the bibliometric analysis provides more objective and scientific results compared to manual reviews (Wang et al., 2021). Accordingly, in this study, bibliometric analysis was utilized, and bibliometric research steps were conducted based on several studies (Wang et al., 2021;Liang et al., 2020;Aria and Cuccurullo, 2017). The research methodology steps followed in this study can be seen in Figure 1. 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h  i  t  e  c  t  u  r  a  l  M  a  n  a  g  e  Initially, conceptual boundaries were established by referring to search terms determined by the purpose of the study. The search query was composed based on a preliminary literature review and existing literature review studies (Boton et al., 2021;Chen et al., 2021;Subedi and Pradhananga, 2021;Akinlolu et al., 2020;Perrier et al., 2020;Wang et al., 2020;B. H. W. Guo et al., 2017;Zhou et al., 2013). The preliminary literature review was utilized to ascertain the coverage of keywords.
The search query consisted of three parts: keywords referring to (1) the construction industry, (2) occupational health and safety, and (3) digital technologies. Based on the previous literature review studies, similar terms and synonyms for the construction industry and occupational health and safety to the help of Boolean operators "AND", "OR". In the search query, * wildcard, which represents any number of characters was used to cover different spelling variations and similar words. For instance, "building information model*" was used to refer to building information model, building information modeling, and building information modeling; "sens* tech*" was used to find terms such as sensor technology, sensor technologies, sensing technology, sensing technologies, etc.
Similar to study of Zhou et al. (2013), the search query was specified to be searched in publication title, abstract, and/or keywords (which is expressed as TS=() in Web of Science, and TITLE-ABS-KEY() in Scopus). The search query can be seen in Table 2.
Table 2. The search query ("construction industry" OR "construction sector" OR "construction company" OR "construction firm" OR "construction organi*ation" OR "building industry" OR "building sector" OR "building project" OR "civil engineer*") AND (1) the construction industry ("occupational safety" OR "health and safety" OR "safety and health" OR "construction safety" OR "worker safety" OR "labo*r safety" OR "accident") AND (2) occupational health and safety ("information technolog*" OR "information and communication technology" OR "digital tech*" OR "digiti*ation" OR "industry 4.0" OR "construction 4.0" OR "construction tech*" OR "automation" OR "computer vision" OR "building automation system" OR "autonomous construction" OR "building information model*" OR "BIM tech*" OR "BIM system" OR "digital twin" OR "virtual twin" OR "robot*" OR "robotic*" OR "simulation" OR "virtual reality" OR "augmented reality" OR "augmented virtual*" OR "immersive media" OR "mixed reality" OR "virtual prototyp*" OR "gam* tech*" OR "serious gam*" OR "internet of thing*" OR "big data" OR "additive manufactur*" OR "digital prefabr*" OR "modulari*ation" OR "modular construction" OR "3D print*" OR "wearable device*" OR "wearable sensing device*" OR "wear* sens*" OR "wearable simulation*" OR "e-hard hat" OR "artificial intelligence" OR "machine learn*" OR "deep learn*" OR "cloud tech*" OR "cloud comput*" OR "blockchain" OR "e-learn*" OR "online teaching" OR "management information system*"OR "building management system" OR "e-safety management system" OR "ultra wideband" OR "ultra-wide band" OR "ultra-wideband" OR "geographic information system*" OR "global position* system*" OR "remote sens*" OR "sens* tech*" OR "sensor" OR "motion sens*" OR "ubiquitous sensor network" OR "wireless tech*" OR "wireless network" OR "wireless local area network" OR "wireless sensor network" OR "WiFi" OR "internet of service*" OR "web service*" OR "automatic identification system*" OR "3D visual* tech*" OR "4D visual* tech*" OR "photogrammetry" OR "image process*" OR "video" OR "camera" OR "vision analysis" OR "virtuali*ation" OR "barcod*" OR "mobile comput*" OR "eyetracking" OR "eye track*" OR "zigbee" OR "bluetooth" OR "accelerometer" OR "inertial measurement unit" OR "unmanned aerial" OR "drone*" OR "light detection and ranging" OR "laser detection and ranging" OR "laser scan*" OR "3D scan*" OR "point cloud" OR "radio frequency identification" OR "3D model*" OR "4D model*" OR "real-time" OR "real time" (3) digital technologies 12 OR "proximity warning" OR "early warn*" OR "ubiquitous connectivity" OR "ultrasound" OR "ultra sound" OR "infrared" OR "action recognition" OR "object recognition" OR "object identif*" OR "physiological status monitor*" OR "mobile device*" OR "smartphone" OR "industry foundation class" OR "smart build*" OR "smart city" OR "smart factory" OR "cyber physical system" OR "embedded system*" OR "human computer interaction" OR "smart construction object*" OR "personal electronic dosimeter*") The first stage of bibliometric research is to determine the data source. Among several data sources, Web of Science (WoS) and Scopus, which are multidisciplinary and selective bibliographic databases are the most reliable and valid ones (Pranckutė, 2021). WoS and Scopus were selected as data sources since they are comprehensive publication and citation databases. Then, the research boundaries were determined. In terms of publication type, the search was not restricted. It was intended to provide a comprehensive outline of the research area. Journal and conference papers convey the research based on theoretical and practical applications. They are published after peerreview and detailed examination; hence, they represent a high-quality research (Wang et al., 2021).
Books, book chapters, and reviews cover significant previous cumulative know-how. The inclusion of different types of publications helps to increase the value of the study. The time range of the study was selected as up to 2021 to provide the development of the area from past to present and to represent the most recent studies. According to the bibliometric search, the earliest paper on digital technologies in construction health and safety was published in 1986. Hence, the time range was set from 1986 to 2021. The language of the publications was not restricted to prevent bias caused by the exclusion of studies that were not written in English according to suggestion of Akinlolu et al. (2020). Moreover, there were some other exclusion criteria such as newspapers, editorial materials, and books were not used in the search. All documents with a core subject of safety but not relevant to emerging technologies were left out of the review.
After the determination of selection criteria, the search was conducted on 6 th December 2021.
It resulted in 462 studies from WoS and 363 studies from the Scopus database. Due to use of two databases, there were duplicate documents that had to be removed (Zhou et al., 2013). In a total of 13 825 documents, 100 duplicates were detected and removed. To determine whether the documents obtained as a result of the search are within the scope of the study, the titles and abstracts of the documents were reviewed by five authors. If the abstract was not satisfactory, the manuscript was also inspected. The main inclusion criteria were as follows (similar to study of Zhou et al. (2013)): The study should be conducted in the construction industry. The industry covers a broad range of activities such as transportation, bridge, and dam construction, and they were included. The study should emphasize health and safety. Some studies were related to structural health or structural safety of the construction, and they were not included. The study should also emphasize the use of digital technologies. Studies that utilize technologies as the main methodology, show an implementation of them or investigate them conceptually were included. Some methods like analytical network process, analytical hierarchy process, fuzzy logic, system dynamics, structural equation model, etc. were not included since they are mostly statistical analysis methods rather than advanced technologies. After a review of the documents, 491 publications remained on the final list.
One of the most important steps in the bibliometrics analysis is the creation of science maps.
Science mapping helps to create visual representation among different factors such as authors, sources, documents, and countries, and to improve the bibliometrics analysis (Moral-Muñoz et al., 2020). Among several related software, VOSViewer and Bibliometrix were employed in this study.
This software can create and view bibliometric maps and is capable of using data from WoS and Scopus databases in addition to being freely available. They are advantageous compared to others, VOSViewer provides advanced graphical visualization and Bibliometrix presents a comprehensive group of methods (Moral-Muñoz et al., 2020;van Eck and Waltman, 2010).
Before the analysis, data were preprocessed. The data files were modified to be suitable for bibliometrics software (ris file for VOSViewer and excel file for Bibliometrix). The keywords and author names were standardized to obtain clearer results. Because, in similar articles, the same concepts can be defined with different spelling variations and similar words. For this reason, it is 14 necessary to standardize the keywords before conducting the analysis (Su and Lee, 2010). For example, "building information model", "building information modeling", "building information modelling" and "BIM" referring to the same technology were standardized as "building information model (bim)". After data preprocessing, the final bibliometric files were transferred to VOSViewer and Bibliometrix software. The information obtained from the analysis was visualized and several science maps were created to present an understandable picture of the previous studies on digital technologies used in construction health and safety. The next section details the analysis and results.

ANALYSIS AND RESULTS
According to the results of this review, it was seen that the studies related to digital technologies in construction health and safety have been published since 1986. As digital technologies emerged and enhanced throughout the years, the number of studies increased continuously. Even though there are a few studies in the 1990s, the research area aroused interest in the last 10 years and the number of research studies has an increasing trend as can be seen in Figure   2. The trend of the increase can be said to be parallel with the development and adoption of technologies in the construction industry.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g In total, 491 documents were included in the bibliometrics analysis. Most of the documents were articles (279), which were followed by conference and proceedings papers (180). Other documents were reviews (28), book chapters (2), editorial material (1), and a short survey (1).

Publication source analysis
In total, 209 publication sources were identified in which the documents were published in the form of journals, books, etc. Most influential sources based on the number of documents published were shown in Figure 3. Considering the source impact based on the number of total citations, the top ten sources were Automation in Construction (1918 total citations), Journal of Construction Engineering and Management (1115), Safety Science (806) 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g 16 analysis of keywords shows the relationship of keywords according to the number of documents that appear together. In the visualization, the size of the nodes represents the occurrences of keywords, and the links between keywords represent connections. Label and circle size increase in direct proportion to the occurrence frequency of keywords. The smaller the distance in between, the stronger the relationship between them (van Eck and Waltman, 2010). In the analysis, a total of 2201 keywords were identified. Keywords that appear at least 5 times were selected, and 202 keywords met the threshold. In Figure 4, the keyword co-occurrence map was constructed by using VOSViewer. The most frequently occurring keywords were "health and safety" (149 occurrences), "construction industry" (132), "construction health and safety" (125), "building information model (bim)" (89), "management" (73), "accident prevention" (65). A total of 7 clusters were identified that were represented with different colors.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g 17 21 of the scholars produced single-authored documents. This may indicate that the studies on digital technologies in construction health and safety require collaborative efforts. Authors who produced the highest number of documents and had the highest citations were Jochen Teizer, Heng Li, and SangHyun Lee as can be seen in Table 3. Scholar's production was also investigated with a figure which was created by Bibliometrix based on the number of documents produced per year and total citations per year (See Figure 5).
The size of the nodes represents the number of publications by the author. The color of the nodes shows the total citations per year. According to the figure, it can be said that as a growing area,  based on their number of co-authored documents was performed in VOSViewer. The threshold for a minimum number of documents was selected as 1 to include all authors. The minimum cluster size was determined as 2, which helps to show small collaborations between authors as well. Figure   6 shows the co-authorship map of scholars where the size of the nodes represents the number of documents by scholars. It should be noted that only connected authors were shown in the figure for

Country and institution analysis
The production of countries and institutions was evaluated based on the corresponding author's affiliation. The findings revealed that the most productive 10 countries were the United States (273   documents (11) and Oregon State University (11) according to a number of documents produced in these institutions.

Document analysis
The most influential papers that affected other studies are important to understanding the development of the research area. The most cited 20 documents were shown in   (Li, Lu, Hsu, et al., 2015) Proactive behavior-based safety management for construction safety improvement Saf. Sci.
Investigates proactive behavior-based safety in terms of improving construction safety. Proactive behavior-based safety is simply referred to behavior-based safety with the technology integration.
Investigates an ontology based semantic model to organize, reuse, and store construction safety knowledge by formalizing this approach.  (Bansal, 2011) Application of geographic information systems in construction safety planning Int. J. Project Manage.
Investigates the use of geographic information systems, 3D model with its surrounding topography, and schedule for the development of safe execution sequence in terms of identifying hazards in the construction safety planning process. 92 8.4  Risk assessment of work-related musculoskeletal disorders in construction: Integration of virtually real construction model and designfor-safety-process database Autom. Cons.
Investigates the visualization of the construction processes to evaluate safety hazards. The study further proposes a design for safety process tool.

Cluster Analysis
A total of 7 clusters were identified that were represented with different colors. The first cluster included keywords such as "construction industry", "accident prevention", "accidents, injuries, and fatalities", "construction site", "occupational risks", and "construction equipment". This cluster mostly focused on the accident prevention context. In the second cluster, some of the keywords were "construction health and safety", "building information model (bim)", "safety management", "hazard identification", "design", "augmented reality", "visualization technologies", "internet of things (iot)", and "eye-tracking". This cluster mostly included visualization technologies in construction safety. The third cluster covered some words such as "construction workers", "machine learning", "computer vision", "artificial intelligence", "automation", and "scaffolding". In the fourth cluster, some keywords were "construction", "ergonomics", "sensor", "building industry", "human", "safety performance", "prevention through design (ptd)", and "musculoskeletal disorders". This cluster was more focused on human-safety relation. The fifth cluster involved "radio frequency identification (rfid)", "real-time location system (rtls)", "information communication technology (ict)". In the sixth cluster, there were "health and safety", "virtual reality (vr)", "health and safety training", "training", "education". The seventh cluster involved keywords such as "unmanned aerial system (uas)", and "digital technologies". Clusters represent keywords that tend to appear together. Four zones in the thematic map are shown based on the relevance degree (centrality) and development degree (density) of the clusters in Figure 9. Cluster analysis was conducted based on the co-occurrence of author keywords via Bibliometrix. The number of words that were used in the analysis was selected as 230, and the minimum cluster frequency was determined as 5 per thousand documents. Clusters were labeled based on the most frequent 3 keywords in the cluster. Motor themes zone includes co-occurrences that have high centrality and high density, which means that they are well-developed and important themes in the field (Aria et al., 2020). Being shown as Cluster 1 under motor themes, three keywords namely safety, included studies investigating construction health and safety, building information modelling, simulation, visualization, and risk management. Basic themes zone covers co-occurrences with high centrality but low density, which means that they include general topics across different research areas in the field and they are important. Clusters 2, 3, and 4 were positioned under basic themes having relatively high centrality but low density. Cluster 2 included topics such as construction safety, safety training, hazard identification, and eye tracking. Cluster 3 encompassed mainly occupational health and safety, accident prevention, digital technologies, and the internet of things.
Cluster 4 mainly focused on occupational health and safety, virtual reality, augmented reality, and training. Cluster 5 came up with keywords such as machine learning, deep learning, computer vision, and ergonomics having moderate density and centrality, and it belongs to both motor and basic themes. Emerging or declining themes indicate low centrality and low density, meaning that they are weakly developed and marginal (Aria et al., 2020). Cluster 6 came up with construction safety management, tracking, and real-time locating as frequent keywords and analyzed under emerging or declining themes. Niche themes show high density but low centrality which means that they are well-developed, but their importance is limited (Aria et al., 2020). Clusters 7, 8, 9, and 10 were revealed under niche themes having high density but low centrality. In these clusters, some common keywords appeared as the construction sector, personal protective equipment, digitalization, and innovation. Cluster 7 enclosed prevention through design and construction equipment in construction health and safety. In cluster 8, radio frequency identification and personal protective equipment keywords were prevalent. Clusters 9 and 10 focused on innovation, and gamebased learning, respectively.  Figure 9. Cluster analysis of author keywords

DISCUSSION
This study conducted a bibliometric analysis of the technology used in construction safety. In this context, interesting findings were reached in terms of the analysis of past studies. In addition to scanning a broad range of papers, the study further revealed the most beneficiary and common technologies used for construction health and safety. The study further discusses the use of these technologies under specific categories.

Digital Technologies used for Construction Health and Safety
Worker safety is one of the most important issues for the sustainability and development of the construction industry (Bhandari et al., 2020). Although each construction accident seems to have different characteristics, detailed studies show that the causes of accidents are common patterns.
Identifying these common patterns makes accidents foreseeable and may lead to reducing the number of future accidents (Mohammadi and Tavakolan, 2020).
Unidentified hazards create unmanageable risks. Therefore, hazard identification is an important component in the construction safety management (Carter and Smith, 2006). To identify Methods such as decision support systems, radiofrequency identification ( There is a need for a knowledge base capable of providing real-time information for the entire construction industry (Aguilar and Hewage, 2013). The interoperability of different IT tools can be examined to enable knowledge acquisition in the real-time data collection and processing (Jin et al., 2019). The below sections explain the digital technologies currently used for construction health and safety and highlight their relevance in terms of construction health and safety practices.

BIM in construction safety
Building information modeling (BIM) has been considered an innovative technology that aims to improve construction projects in terms of safety, efficiency, etc. (Wang and Chong, 2015).
BIM enables practitioners to better understand the project and helps reduce the likelihood of accidents (Ganah and John, 2015). BIM helps to reduce the probability of accidents (Ganah and John, 2015) and, thanks to its 3D visualization, also eliminates problems such as lack of foresight in traditional management. Thus, unforeseen hazards in the construction industry can be prevented and "people-oriented" safety management can be created by improving the safety management (Li et al., 2020).
Many researchers in the literature stated that the use of BIM positively affects the safety of construction projects (Park et al., 2017;Kim et al., 2016;Zhang, Sulankivi, et al., 2015;Qi et al., 2014). In addition, the European Union supports the use of BIM in the development of projects (Cortés-Pérez et al., 2020). Projects that use BIM for visualization save costs while determining the accuracy of their data (Datta et al., 2020). Pham et al. (2020), and Swallow and Zulu (2020)    It showed a significant potential for safety control and monitoring.
However, the implementation of BIM in safety plans and construction safety management is still insufficient (Rodrigues et al., 2021). The main reason for this is insufficient training and lack of wide-ranging use (Enshassi et al., 2016). Malomane et al. (2022) implied that there are critical challenges preventing the implementation of emerging technologies in the construction industry.
The most important threats to implementation were shown as a lack of adequate skills and training.
The reason behind the low adoption of BIM is further explained as a lack of governmental mandates for BIM use. For example, Hire et al. (2022) mentioned that BIM is still in the early stages of implementation in the Indian construction industry. However, they mentioned that there is great potential for using BIM to adopt it for better safety management. The study further implied that the USA, China, UK are leading countries in BIM use, and they are early adopters of BIM.

Simulations and visualization in construction safety
Construction projects differ from each other with their unique structures. It is therefore difficult to identify all possible safety risks and hazards (Azhar, 2017). According to Rubaiyat et al. (2016), innovative methods should be developed to automatically monitor worker safety at the construction  (Datta et al., 2020).
Researchers have investigated the use of simulation in construction safety management and training and obtained successful results. Several studies have explored the use of visualization in the safety of construction projects (Bhandari et al., 2020;Eiris et al., 2020;Lee et al., 2020;Azhar, 2017;Hallowell et al., 2016;Siddula et al., 2016;Li, Lu, Chan, et al., 2015;Le et al., 2014;Tsai, 2014;AlBahnassi and Hammad, 2012). Besides, the use of visualizations in safety training was investigated by some studies (Bhandari et al., 2019;Peña and Ragan, 2017). Bhandari and Hallowell, (2017) proposed a multimedia learning simulation-based training program called naturalistic injury simulation (NIS). The simulation shows the cause and effect of hand injuries with a realistic human hand replica in different scenarios. The study found that workers experienced negative emotions which may help to better perception of risks.

Sensors and wireless technologies in construction safety
The rapid development of motion sensors provides the opportunity to be used in areas such as monitoring worker behavior for safety and worker health in construction, automatically detecting unsafe actions of workers, and receiving feedback. Thus, sensor technology reduces the number of unsafe actions and prevents possible accidents (Han et al., 2014). However, according to Teizer and Cheng, (2015) site safety monitoring in the construction industry is not automatic. Usually, these monitoring are done and evaluated manually (Teizer and Cheng, 2015). For this reason, several researchers have worked on the development of sensor technologies for occupational safety in the construction industry (Hasanzadeh et al., 2020;Lee et al., 2020;Hwang and Lee, 2017;Teizer and Cheng, 2015;Riaz et al., 2014). is a lot of research on wearable sensing devices (WSD) for personal use, the information is limited in the construction industry (Awolusi et al., 2020). J. Chen et al. (2016) and Wang et al. (2017) developed a wearable electroencephalography (EEG) system. J. Chen et al. (2016) inspected inattentional blindness which is a failure to recognize unexpected hazards due to mental workload of workers during complicated construction tasks. They measured neural signals with EEG and assessed hazards with neural time-frequency analysis. The study further proposed the development of a wearable EEG safety helmet to estimate construction safety. Wang et al. (2017) proposed a wireless and wearable electroencephalography (EEG) system to evaluate the attention level of construction workers by measuring human brain signals. The results indicated that EEG signals can be used to predict the perceived risk level of construction workers.  showed that physical data obtained with wearable devices can be used to detect the psychological state of construction workers. They conducted a study to examine the relationship between the physical and psychological status of workers. They employed wearable devices to collect physical data and questionnaire survey to collect psychological data. The findings implied that physical data is available to measure workers' psychological status. Accordingly, unsafe behaviors can be identified and investigated.
Jebelli et al., (2016) investigated the fall-risk of construction workers. They used inertial measurement unit to collect stability metrics data and compared with the postural stability observed.
They implied that collected data helped to predict fall risk of construction workers. Fardhosseini et al., (2020) collected gait pattern data of construction workers using accelerometer embedded in smartphone. Based on the data, they developed supervised-learning algorithms to predict physical fatigue of workers in different working conditions. Tsai (2014) emphasized the risk of fall accidents among engineering students participating in field surveying. The study utilized accelerometer to  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h  i  t  e  c  t  u  r  a  l  M  a  n  a  g  e  m  e 32 track behavior of students and image-recognition system for surveillance. The use of these technologies helped promoting safety in education and training. Zhang and Liao (2020) analyzed the brain activity of participants during hazard recognition with wearable near-infrared spectroscopy (NIRS) system to investigate the influence variables.
They proposed and tested site familiarity, safety knowledge, and risk tolerance as the predictors of the change in the brain. Hasanzadeh and Garza (2020) collected data using real-time location tracking sensors, physiological sensor, survey, and interviews to study workers' risk-taking behavior. They found that higher risk tolerance level increases the likelihood of being involved in risky activities and taking risks. Ding et al., (2020) proposed a solution for safety in tunnel construction by developing an intelligent early warning system and a wearable device. They implemented the proposed system in multiple tunnels and produced good safety visualization and monitoring.
Wireless remote sensing technologies provide the real-time location of workers, equipment and materials (Skibniewski, 2014). RFID sensors help collect real-time information and inform construction workers in case of potential risks (Skibniewski, 2014).  identified a proactive construction management system for training of precast installation workers. The system builds a training environment using real time location technology and data visualization technology to monitor workers, equipment, and material, give real-time feedback and visualization.  developed an intelligent safety helmet by using RFID technology and wireless communication, considering ergonomics. Intelligent safety helmet provides attendance and location of the on-site workers, updated information and in-time communication.
Several studies used an eye-tracking system to ensure construction safety (Han et al., 2020;Hasanzadeh et al., 2017Hasanzadeh et al., , 2018Dzeng et al., 2016). In the study of Hasanzadeh et al. (2017), eyetracking technology was utilized, and eye movements of construction workers were recorded. The results suggest that safety knowledge enhances hazard recognition and visual search strategy, and visual attention based on eye-tracking data, and metacognitive stimulations. According to empirical findings, the proposed strategy helped to improve hazard recognition levels of construction workers.

Robotic in construction safety
Construction industry requires working in dangerous environments which creates unfavorable conditions for construction workers (Skibniewski, 2014). Hence, robotic technologies are attracting attention as an effective solution for ensuring safety in the construction industry (Kim et al., 2020).
Some studies investigated the usability of robotic systems in eliminating different risks that threaten worker health and safety in the construction industry. Bai (2007)

Virtual reality, Augmented reality and Mixed Reality in construction safety
Guray and Kismet (2022) 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h i t e c t u r a l M a n a g e m e 34 reality system to investigate the effect of social influence on workers' safety behaviors in hazardous conditions. It was found that social influence plays a significant role in safety behaviors.

Unmanned aerial vehicles (UAV)
UAV systems can be used in construction safety improvement by monitoring activities of boom vehicles and cranes on the construction site and monitoring unprotected openings (Ghesairi and Esmaili, 2019). On the other hand, UAV systems have some challenges in liability and legal concerns and safety issues (Ghesairi and Esmaili, 2019).

Artificial Intelligence and Machine Learning
Mansoor et al., (2020) developed a conceptual framework for mobile cranes to improve safety. They utilized a camera which records a video that is interpreted with a mathematical algorithm and is transcribed into commands. This procedure increases the efficiency and safety of the crane operation by eliminating the communication between the signalman and the operator. Especially, when used together, the technologies are even more effective as implied in previous studies.   -Identification (hazard identification, risk register system) -Planning (safety planning) -Design (prevention through design) -Training and education -Prevention (risk prevention) -Analysis (safety risk assessment, object-oriented and process-oriented job hazard analysis) -Visualization (near-miss information visualization) -Simulation (safety risk simulation) -Monitoring (automated safety checking system, safety risk reviews, safety audit, automated safety rule checking) -Management (construction safety management, occupational hazard management) Artificial intelligence (AI) and Machine learning (ML) -Identification (workers, safety hazards and accidents, safety risks, safety leading indicators, personal protective equipment, spatial safety risk identification of earthmoving and surface mining activities), -Classification (safety accidents, accident causes, scaffolding, construction site photos for roof detection) -Detection (real-time worker safety detection, safety characteristics, workers' actions, real-time hardhat wearing, non-certified work, safety guardrail) -Prediction (health and safety risk, safety hazards and accidents, safety outcome, post-accident disability status of workers, severity prediction models of falling risk for workers at height) -Measurement (safety performance on roof sites, dynamic blind spots for construction equipment operators) -Prevention (occupational injuries, collisions between heavy equipment vehicles and other on-site objects) -Analysis (predictive safety risk analysis, safety status of construction workers, level of collision safety, risk factors for work-related musculoskeletal disorders, physical fatigue, posture analysis, ergonomic assessment, behavior-based safety, causality patterns of unsafe behavior leading to fall hazards) -Monitoring (safety surveillance, hardhat wearing, early-warning system for site work in hot and humid environments) -Real-time occupational health and safety management system -Training and education (construction safety training for construction workers, safety education for engineering students) -Identification (hazard identification, hazard recognition) -Planning (safety planning for construction activities performed at high elevation) -Visualization (visualization of danger factors) -Simulation (safety system and simulation) -Communication (safety risk communication, safety culture integration)

Sensors and Wireless Technologies
-Identification (hazard identification and recognition, construction activity recognition, automated action recognition, workers' unsafe behaviors) -Detection (hazards, hard hat wearing, detection of fall portents, detection of near-miss accidents and falls, collision detection, obstacle detection) -Classification (classification of unsafe actions) -Measurement (construction workers' physical demands, physical fatigue, construction worker's musculoskeletal disorder risks, visual attention and situation awareness, inattentional blindness, psychological status, attitude estimation and stability measurement of articulated heavy vehicles in the construction industry) -Prevention (proactive accident prevention, prevention of struck-by-falling-object accidents, collision avoidance) -Analysis (safety risk, workers' behaviour analysis) -Visualization (real time tracking and visualization of resources) -Simulation (wearable simulations) -Monitoring (health and safety monitoring, ergonomic analysis, worker movements, hazardous area intrusion, environmental monitoring on construction sites, exposure to solar uv radiation, proximity warnings of construction safety hazards, autonomous crane safety monitoring, workers and scaffolds monitoring) -Management (construction worker safety management system) Simulation and Visualization -Design and planning of a healthy construction workplace -Detection (hazard detection) -Safety training -Visualization (visibility and proximity based risk map of earthwork site -Simulation (earthmoving projects, near real-time motion planning and simulation of cranes in construction) -Analysis (ergonomic job analysis for workplace design, job hazard area identification) -Modelling -Monitoring (on-site safety monitoring and warnings

Future Recommendations
This study provides a future direction for researchers to categorize and focus on the niche themes of construction safety technology. In this respect, Figure 10 was created to categorize these technologies concerning construction project phases. For example, BIM and visualization were identified as the key technologies for prevention through design. Moreover, VR, AR, serious games, and eye tracking were determined as the most influential technologies for training and education.
For the equipment during the construction phase, GIS, GPS, and robotics were implied as the essential themes that one needs to focus on when integrating technology into equipment. Finally, it

CONCLUSIONS
The bibliometric analysis includes evaluations regarding publication sources, scholars, keywords, and documents aimed to identify key issues related to the use of digital technologies in construction health and safety and provide a research framework. This study investigated the use of digital technologies in construction health and safety. In this respect, a broad range of studies was scanned.
This study provides a comprehensive analysis including all available studies without restricting document type, publication year, and language. Further, two bibliometric software was utilized to better visualize the existing studies.
When conducting the study, some challenges were encountered. Especially, data exchange between two different software (WoS and Scopus) was a significant challenge to generate the  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h  i  t  e  c  t  u  r  a  l  M  a  n  a  g  e  m  e 38 bibliometrics analysis. Moreover, new research is on the way for similar topics, which led researchers to conduct a timely analysis and proper investigation of past studies. Studies that were published in a different language were another challenge to overcome since the researchers aimed to include all studies for the specified period of the bibliometrics search. Finally, the keywords used were carefully selected since some keywords were very specific to finding target research. Several keywords were scanned through to best represent the focus of the study. Hence, future research on similar topics shall consider the importance of using the proper keywords to come up with a more reliable analysis. This study provides directions for researchers to (1) find the proper sources for bibliometrics analysis (2) search databases with proper keywords, and (3) generate a systemic review for the emerging technologies in safety. Hence, the study encourages researchers to better visualize the data regarding past studies and the impact of those as well as leading industry practitioners to consider what influential technologies can be used to improve safety performance.
Although information technology in construction safety has been one of the popular topics among researchers in the last decade, the study revealed that there are still research gaps in this field.
The analysis indicated that there is still a growing need for promoting artificial intelligence methods in terms of construction safety. The study further showed that a human-oriented safety management system should be established to best address safety concerns. Real-time knowledge acquisition was also detected as a lacking item in most construction projects. The detailed analysis of past studies further implied that there is a growing need for digitalization in construction safety along with promoting personal protective equipment. This study recommends that prevention guides for construction accidents must be in place to protect construction workers from being injured.
Moreover, the establishment of application guides for countries that will adopt BIM shall be put in place. Another recommendation is to design studies that combine multidisciplinary areas. The use of digital technologies in construction health and safety covers several disciplines. The field of computer science and information technology is necessary for the development and implementation On the other hand, the study had some limitations such as the use of subjective procedures in the title and abstract review and keyword standardization and selection of thresholds which may result in older keywords and underrepresentation of novel areas.
Despite certain challenges, this study has considerable implications for researchers, industry practitioners, and policymakers. First and foremost, this study points out a very important topic, which is the safety of construction workers. Considering the impact of work-related hazards and accidents upon society, this study has the potential to help construction people to develop new safety measures and 'stop the line' when necessary. This is because this study provides the most trending and frequently used technologies of which the effectiveness has already been proven. Workers can benefit from the finding of this research to benefit from the technologies when working in hazardous conditions. The risk processes can be replaced with automated processes. This in turn could reverse several fatal cases and protect many who would find themselves in hazardous situations. Second, the study encourages researchers to conduct a similar search with more recent data, where the proven benefits of technology integration in safety processes are apparent. This way, construction researchers could focus on the niche themes and develop research accordingly to best favor the use of technologies in safety research. Industry practitioners may further benefit from the findings to revise and revisit their safety programs. They could potentially develop a 'zero mistake' policy at work, which could save many lives using such emerging technology trends as VR, AR, BIM, and so on. Finally, policymakers could develop new strategies to improve safety programs with the smooth and transparent integration of technology into processes with the proven benefits of technologies used in previous implementations. Since this study provides a very broad analysis of  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   e  e  r  i  n  g  ,  C  o  n  s  t  r  u  c  t  i  o  n  a  n  d  A  r  c  h  i  t  e  c  t  u  r  a  l  M  a  n  a  g  e  m  e  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60