Research Trainee: Florence Kizza, PhD Student, Michigan State University

Principal Investigator: Dawn Misra, Department Chair and Professor, Department of Epidemiology and Biostatistics, Michigan State University

The proposed study will investigate the impact of multidimensional workplace environment exposures on preterm birth (PTB) among Black mothers. Prenatal workplace exposures and experiences have rarely been studied among Black women, who are at elevated risk of PTB. Using data from a large cohort of pregnant Black women enrolled in the NIH Parent study, “Impact of Racism on Risk of Preterm Birth in Black Women” (1R01HD058510; PI: Misra, DP), we will pursue the following: Aim 1: evaluate the associations between work conditions (i.e., work hours, job loss) and PTB; Aim 2: assess the association between maternal occupation and PTB; Aim 3: evaluate whether the associations between work exposures (work conditions, maternal occupation) and PTB differ according to depressive symptoms, perceived stress, and lifetime experiences of racism. This study will provide novel empirical evidence on the role of multi-dimensional work exposures on preterm birth among Black mothers. This may support development of workplace policies to support the health of pregnant women and infants. Florence Kizza, MPH, is currently pursuing a PhD in Epidemiology, and is uniquely positioned to benefit from the fellowship because of her strong quantitative and epidemiology skills and prior experience as a maternal and child health epidemiologist at the state health department. This pilot research training award will provide the applicant with interdisciplinary training in: 1) classifying occupational exposures, 2) methodology for analyzing occupational exposure data, 3) an in-depth understanding of the impact of work exposures on birth outcomes. Training activities will include independent study, short courses, and workshops. The project will be implemented in the Department of Epidemiology and Biostatistics at MSU, which is an ideal training environment due to the department’s strengths in perinatal health and its strong history of mentoring and interdisciplinary collaboration. This pilot research training will be mentored by: Dr. Misra (sponsor, internationally recognized perinatal and social epidemiologist with expertise in racial disparities and preterm birth and a long track record of successful mentoring) and Dr. Harduar Morano (co-sponsor, occupational and environmental epidemiology). The proposed fellowship will equip the applicant with substantive knowledge, analytic tools, and professional competencies necessary to achieve her long-term goal of becoming a perinatal epidemiologist, leading independent investigations to evaluate the impact of the workplace environment on perinatal outcomes and racial disparities in birth outcomes.

Research Trainee: Ting Hung Lin, PhD Student, University of Wisconsin-Madison

Principal Investigator: Robert Radwin, Professor, University of Wisconsin-Madison

Occupational ergonomics methods for evaluating repetitive manual tasks have traditionally employed an industrial engineering work methods analysis approach, where work activities are observed and categorized using fundamental work elements and quantified by time, frequency, exertions, and associated movements. This approach gave rise to important exposure measures such as the Hand Activity level, Duty Cycle Fatigue, the Strain Index, and other validated measures. Computer vision and AI are employed in this proposal for creating a human centric model of repetitive manual tasks for automatically evaluating work activities. Using a set of work element descriptors, Reach, Gasp, Move, Pause and Release, AI vision-language models will be trained to recognize and categorize video recordings of repetitive manual tasks. We will incorporate videos pooled from 1,649 workers across 16 industries, recorded by multiple institutions including NIOSH, the State of Washington SHARP, UC-San Francisco and the University of Wisconsin-Milwaukee upper extremity consortium studies. The video clips will be manually annotated based on the work element descriptor vocabulary. We will use fine-tuning vision-language models to produce captions that describe the contextual events in the video clips. We propose to generate a video dataset with fine-grain text annotation with the five key motion descriptors. We will test this model by applying it to calculate the ACGIH Hand Activity Level (HAL) based on frequency and duty cycle. If successful, this pilot project will provide support for the further development of AI computer vision models for automatically analyzing repetitive manual tasks for musculoskeletal injury risk factors and potentially lead to AI assisted interventions.

Research Trainee/Principal Investigator: Justin Morrow, Research Scientist, University of Cincinnati

Mentor: Jun Wang, Associate Professor, University of Cincinnati

The proliferation of nano- and microplastics (NMP) in the environment is widely known as a topic of significant research and growing concern. However, much less is known about occupational airborne NMP exposures faced by workers and how this differs from environmental exposure. Past research has confirmed that airborne NMP concentrations near plastic manufacturing processes can be orders of magnitude higher than ambient environmental airborne NMP. However, little work has been done to quantify how much airborne NMP exposure workers face and what specific health risks are posed. There are tens of common plastics used with hundreds of chemical additives, such as BPA and phthalates, which are already known to cause various adverse health effects (e.g., neurological, development, reproductive disorders, cancers), yet there is no recommended exposure limits or regulation specific to occupational airborne NMP. Therefore, there is a critical need for an airborne NMP characterization protocol that can physically capture NMP aerosols, chemically identify the plastic constituents, and quantify the concentration, leading to an accurate estimation of dosages. Existing research has shown that standard aerosol collection methods and Raman micro-spectroscopy are effective for identifying and quantifying airborne NMP greater than 1-2 micron (µm) in size through direct particle counting and analysis. However, there has been little work in collecting and characterizing the respirable fraction of airborne NMP, which is below the minimum particle size for direct Raman analysis. In summary, the current understanding of occupational exposure to respirable NMP is severely lacking. To address this need, we will use methods developed by the National Institute for Occupational Safety and Health (NIOSH) for quantitative Raman analysis of respirable crystalline silica (RCS) and modify this approach for quantification of respirable airborne NMP using best practices learned from environmental micro and nanoplastic research. We will apply this method to 3D printing as a model manufacturing process, which is well known to emit primarily ultrafine sub-100 nm aerosol particles and will limit the pilot to three commonly printed plastics (PLA, ABS, PETG). Our project aims to prove that direct-on-filter (DoF) Raman analysis can distinguish these plastics and quantify the average exposure to each plastic during a single aerosol measurement. This will be an improvement over standard particulate counting without chemical analysis. If successful, the method would enable sources of respirable airborne NMP to be isolated and quantified for future exposure studies of respirable airborne NMP in short-term acute exposure (e.g., industrial plastic manufacturing) and low-level chronic exposure (e.g., office and building environment). These studies would be well suited for follow-up funding through either NIOSH or the National Institute of Environmental Health Sciences (NIEHS) which has an active notice of special interest (NOT-ES-23-002) on NMP measurement methods and health impact research.

Research Trainee: Mousa Alsulais, PhD Student, Environmental Health Science Department, University of Michigan

Principal Investigator: Oshin Tyagi, Assistant Professor, Industrial and Operations Engineering Department, University of Michigan

Occupational musculoskeletal injuries are a significant concern for Emergency Medical Services (EMS) professionals. This study aims to investigate the factors influencing EMS professionals’ willingness to use exoskeletons during real-world emergency responses and how exoskeleton use can be integrated into existing EMS protocols to minimize disruption to response times. The research will focus on user comfort, perceived benefits/risks, integration with existing workflows, donning/doffing times, equipment accessibility, and overall impact on emergency response efficiency. Twenty EMS clinicians will participate in a laboratory experiment performing tasks with and without a low-back exoskeleton, adapted from Physical Agility Tests (PAT) for EMS workers. Individual interviews will be conducted to gather feedback on task performance, device comfort, and challenges faced. Data will be analyzed for themes and recommendations for effective exoskeleton integration into EMS practice will be developed. This study addresses critical gaps in the validation of occupational exoskeleton devices among EMS professionals and aims to reduce the back injury rate among this population

Research Trainee: Xiangyu Peng, PhD Student, Robotics Department, University of Michigan

Principal Investigator: Leia Stirling, Associate Professor, Industrial and Operations Engineering and Robotics Department, University of Michigan

Musculoskeletal disorders (MSDs) resulting from overexertion and repetitive motion have significant impacts on workers, affecting their ability to perform job-related tasks and incurring substantial costs for companies. Exoskeletons have emerged as a potential tool to reduce physical demands in labor-intensive jobs. While passive upper limb exoskeletons have shown success in specific tasks, they lack adjustability, suitability for complex tasks, and adaptation to changing conditions. Active exoskeletons offer advantages by incorporating software-based parameters for modifying human-exoskeleton interaction. However, existing wearable and portable active exoskeletons have fixed control policies, lacking adaptability or only achieving optimization after long-time evaluations before converging (can take hours). In the upper body, where task switching and object manipulation significantly impact dynamics, the adaptability of upper-extremity exoskeletons needs to address these variations and enable real-time dynamic adjustments.

Preliminary work towards this goal has explored adaptive controllers capable of dynamically updating control parameters based on real-time user behavior. Specifically, in a target-matching task involving cyclical elbow movements, these adaptive controllers exhibited significantly lower classification errors when compared to the default controller. However, these approaches have limitations, including the inability to detect object presence and weight, hindering adaptability in scenarios involving object manipulation. Additionally, prior research focused primarily on cyclical elbow movements, limiting the understanding of upper limb functionality in a wide range of daily activities. In the proposed research, the Myomo MyoPro powered orthosis will serve as the experimental platform, utilizing surface electromyography (sEMG) sensors and pressure sensors as inputs to the adaptive algorithm. The specific objectives are to incorporate object interaction into the adaptive controller and examine its generalizability to more complex, real-life tasks. By addressing these challenges and considering variations in upper-extremity tasks, this research aims to advance the field of occupational robotics research, specifically in the context of MSDs and emerging technologies like exoskeletons.