Abstract
This study presents a PRISMA-based systematic review and meta-analysis of HPE applications for ergonomic risk assessment and workplace safety. The findings indicate that HPE models offer a non-invasive, scalable solution for monitoring worker posture and identifying ergonomic risks. The systematic review highlighted the potential of HPE models in improving workplace safety by providing real-time posture monitoring and ergonomic risk assessment. Studies using OpenPose, MediaPipe, and AlphaPose demonstrated high accuracy in detecting posture deviations, with mean errors ranging from less than 2 mm RMSE to over 70 mm PA-MPJPE, depending on environmental conditions and body regions tracked. The meta-analysis revealed a significant reduction in MSD prevalence due to ergonomic interventions, with a standardized mean difference of −0.71 in pre-post studies and −0.91 in controlled trials. Studies implementing workstation adjustments, ergonomic training, and real-time feedback, showed the highest improvements in posture correction and risk mitigation, consistently reducing musculoskeletal symptoms across workplace settings.
Keywords
Introduction
Workplace safety is a critical concern across various industries, where repetitive tasks, heavy lifting, and prolonged awkward postures contribute to musculoskeletal disorders (MSDs) (Iyer, Macwan. et al., 2024; Iyer, Reynolds, et al., 2024; Odebiyi & Okafor, 2023). These disorders significantly impact worker well-being, productivity, and healthcare costs (Bevan, 2015). Addressing these ergonomic risks requires effective monitoring systems capable of real-time assessment and intervention. Traditional ergonomics assessments often rely on observational techniques or wearable sensors, which can be intrusive and impractical for continuous monitoring in industrial settings (Hilmi et al., 2024; Zhao et al., 2021). Advancements in human pose estimation (HPE) have enabled the development of non-invasive solutions for evaluating workplace ergonomics (Iyer, Macwan. et al., 2024; Iyer, Reynolds, et al., 2024; Zheng et al., 2023). Using computer vision and deep learning techniques, HPE models analyze skeletal movements and posture variations to identify ergonomic risks. Compared to sensor-based methods, image-based HPE allows for unobtrusive monitoring, improving scalability and usability in dynamic work environments (Iyer & Jeong, 2024; Macwan et al., 2024). However, challenges such as occlusions and model generalizability affect accuracy and reliability (Angelini et al., 2019). This study conducts a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework, analyzing 38 research articles on HPE applications for ergonomic risk assessment and workplace safety (Page et al., 2021). Additionally, a meta-analysis of 22 studies examines the impact of ergonomic interventions on reducing MSD prevalence. The objectives of this study are to (1) assess the effectiveness of HPE models in workplace settings, (2) explore usability challenges, and (3) evaluate their accuracy in detecting posture-related risks.
Background
HPE has evolved from marker-based motion capture systems to deep learning-driven approaches capable of detecting human posture from image and video data (Wang et al., 2021). Early methods relied on attaching markers to the body for precise tracking, which, while accurate, was unsuitable for real-world applications. Modern approaches use deep learning architectures to extract skeletal landmarks without requiring physical markers (Iyer & Jeong, 2024). Several widely used HPE models, such as OpenPose, MediaPipe, and AlphaPose, have demonstrated effectiveness in tracking human movement (Cao et al., 2019; Fang et al., 2022; Lugaresi et al., 2019). These models estimate joint positions and analyze body orientation to assess postural risks in industrial environments. Their application in workplace safety includes identifying improper lifting techniques, prolonged static postures, and repetitive motion patterns linked to MSD risks (Iyer, Macwan. et al., 2024; Iyer, Reynolds, et al., 2024; Mehrizi et al., 2019). Despite these advancements, practical deployment in industrial settings presents challenges. While HPE can detect postural deviations, integrating these outputs into ergonomic assessment frameworks like the Rapid Upper Limb Assessment or the Rapid Entire Body Assessment remains an area of ongoing research (Hignett & McAtamney, 2000; McAtamney & Corlett, 1993). This review explores how HPE technologies have been applied for ergonomic analysis and examines their role in enhancing workplace safety. By addressing key challenges in usability and accuracy, the study identifies opportunities for improving HPE applications in real-world settings.
Approach
The study followed the PRISMA-based systematic review to evaluate existing research on HPE for ergonomic risk assessment. A comprehensive literature search was conducted across major academic databases, including Google Scholar, IEEE Xplore, and PubMed, to identify relevant studies published between 2010 and 2025. The inclusion criteria focused on studies that applied HPE for assessing posture, ergonomic risks, and MSD prevention in occupational settings. A total of 341 articles were initially retrieved, which were then screened for duplicates and relevance to the research questions. After applying exclusion criteria, 38 studies were selected for systematic review. The selected studies were analyzed based on their methodology, accuracy, usability challenges, and integration with ergonomic assessment frameworks. In addition to the systematic review, a meta-analysis of 22 studies was conducted to assess the effectiveness of ergonomic interventions in reducing MSD risks. These studies reported quantitative outcomes related to posture correction, musculoskeletal strain reduction, and ergonomic improvements. Statistical measures such as effect sizes and confidence intervals were used to evaluate intervention efficacy.
Outcome
The systematic review highlighted the potential of HPE models in improving workplace safety by providing real-time posture monitoring and ergonomic risk assessment. Studies using OpenPose, MediaPipe, and AlphaPose demonstrated high accuracy in detecting posture deviations, with mean errors ranging from less than 2 mm root mean square error (RMSE) to over 70 mm Procrustes-aligned mean per joint position error (PA-MPJPE), depending on environmental conditions and body regions tracked. The review identified key challenges affecting HPE deployment in industrial settings. Accuracy variability was noted between different body parts, with larger joints such as shoulders and knees exhibiting higher precision than smaller joints like wrists and ankles. Environmental occlusions and lighting inconsistencies further impacted model performance, particularly in dynamic work environments. The meta-analysis revealed a significant reduction in MSD prevalence due to ergonomic interventions, with a standardized mean difference of −0.71 in pre-post studies and −0.91 in controlled trials. These reflect moderate to large effects, indicating that ergonomic interventions consistently reduced musculoskeletal symptoms across workplace settings. Studies implementing workstation adjustments, ergonomic training, and real-time feedback systems showed the highest improvements in posture correction and risk mitigation. The findings suggest that integrating HPE with ergonomic interventions can improve workplace safety by facilitating early detection of high-risk movements and reducing physical strain.
Conclusion
This study presents a PRISMA-based systematic review and meta-analysis of HPE applications for ergonomic risk assessment and workplace safety. The findings indicate that HPE models offer a non-invasive, scalable solution for monitoring worker posture and identifying ergonomic risks. However, deployment challenges such as occlusions, lighting variability, and computational demands must be addressed to improve real-world applicability. Future research should focus on enhancing model robustness, improving generalizability across different workplace conditions, and integrating HPE with automated ergonomic assessment frameworks. The development of real-time feedback systems powered by HPE could further support proactive ergonomic interventions, reducing MSD risks and improving worker well-being. Continued validation of HPE technologies in industrial settings will help their effectiveness for occupational health and safety applications.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
