PJPaul Jenkins

Paul Jenkins

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Paul Jenkins has a total of 134 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Improved classification of agricultural injuries from free-text analysis of pre-hospital care reports under temporal drift

    by Laura E. Jones, Erika Scott, Nicole Krupa, Cristina S. Hansen-Ruiz, Megan E. Kern, Paul Jenkins

  • Farming Becomes More Precarious With Age: Injury in Maine Agricultural Communities, 2008–2022, via Time Series Analysis

    by Laura E. Jones, Erika Scott, Nicole Krupa, Cristina S. Hansen‐Ruiz, Paul Jenkins

    Background: Fatal occupational injury rates among agricultural workers are within the top 10 most dangerous civilian jobs. However, tracking and documenting non‐fatal agricultural injuries in Maine, as in other states, has proved challenging. In 2021, we developed a machine learning‐based strategy to extract injury cases from free‐text pre‐hospital care records (PCRs), which together with final human coding, produces injury time‐series records coded by type, severity, location, date, and subject industry. In this paper we explore and summarize novel time‐series records for agriculture obtained from PCR from Maine. Methods: From a fully labeled Maine dataset ( N = 57,960) comprising coded injuries, we selected only agricultural events, yielding 1604 injuries from 2008 to 2022. We summarize by year, month and age category, and establish seasonality before decomposing time series data, divided into three roughly equal 4‐year sub‐periods, into seasonal, trend and random components using a classical additive model. We investigate associations between age category and injury rate via mixed effects regression, then perform time series regression on differenced monthly injury time series and temperature records to determine if, seasonality aside, temperature extrema are responsible for increased injuries. Finally, we visualize and summarize trend and random components for each study sub‐period. Results: Injury rates show strong seasonality with a peak in July–August, and a trough in January or February. Subject age drifts slowly upwards during our study period, and there is a significant and positive association between age category and injury rate for all but the most elderly farm workers. Injury rates in the age categories of 40–81 years increase dramatically between 2016 and the 2019–2022 period, as does the moving average of the injury rates, and the variability of the random component of the time series. Conclusions: There is a significant positive association between increasing age category and injury rate across all periods. While our injury data has strong seasonality, we find no significant associations between monthly temperature extremes and injury rates. Moving average trends for injury rates in the two periods comprising 2008–2016 show little change in trend, but injury rate trend shifts upward in 2019–2022, almost doubling in mean value.


  • Open Access

    OIICS coding of agricultural injuries mined from Maine PCR records (2008-2022) reveals significant differences in injury source, event, and nature by age group and sex

    by Laura E. Jones, Megan Kern, Cristina Silvia Hansen-Ruiz, Nicole Krupa, Paul Jenkins, Erika Scott

    Objectives: Agricultural injuries are known to be under-reported in existing surveillance systems. OIICS codes are a standardized classification system developed by the Bureau of Labor Statistics (BLS) which ensure consistency in reporting and analysis of workplace incidents over time across industry sectors. Our study examines OIICS coded injuries obtained via mining emergency response (Pre-Hospital Care Report) records (PCRs) to improve tracking, documentation, and understanding of agricultural injury trends. Methods: We analyzed frequencies of OIICS subcodes for Primary Injury Source, Event/Exposure, Nature of Injury, and Body Part classifications for 1,583 injuries among agricultural workers in Maine, spanning January 2008 to December 2022. To streamline the dataset and subsequent analysis, subcodes within each category were thematically grouped. We summarized and visualized grouped code frequencies by subject sex, age category, season of injury, and study subperiod. Chi-square tests were used to assess differences in injury patterns by sex and age group. Results: Reported injuries increased over time from 420 in 2008–2011 to 631 in 2019–2022. The most frequently reported classifications were: ‘Tractors/PTOs’ (Injury Source), ‘Fall’ (Event), ‘Multiple parts’ (Body Part), and ‘Pain’ (Nature of Injury). A marked increase in ‘Nonclassifiable’ Source subcodes and ‘Fall’ Event subcodes was observed in 2019–2022 relative to earlier periods. Significant differences by sex were found for injury Event subcodes: The most frequent source of injuries for females were animals, versus objects and equipment being the most frequent source for males. Nature of Injury also varied significantly by sex. All four OIICS categories (Source, Event, Nature, Body Part) showed significant variation by age group. Older subjects reported more injuries due to falls and overexertion, while younger were more frequently subject to exposure, intentional self-injury, injury in fires, and injuries involving farm vehicles and equipment. Conclusion: Injury counts rose across each successive study period. All injury subcodes differed significantly by age category, while injury Event and Body Part codes also varied significantly by sex. This suggests that injury risks are not uniform across demographics, and tailored safety interventions by sex and age group may be more effective in reducing agricultural injuries.

  • The Effect of the trace operator on the duality of modular grids in genus zero levels

    by Archer Clayton, Paul Jenkins

    Griffin, the second author, and Molnar studied coefficient duality for canonical bases for a broad range of spaces of weakly holomorphic modular forms, showing that the Fourier coefficients of canonical basis elements appear as negatives of Fourier coefficients for elements of a canonical basis of a related space of forms. We investigate the effect of the trace operator on this duality for modular forms for [ ] of genus zero and show exactly when duality still holds after applying the trace operator.

  • Effectiveness and Safety of Mechanical Thrombectomy in Mild Stroke due to Large-Vessel Occlusion: Insights from the ASSIST Registry

    by Sophia Hohenstatt, Dominik F. Vollherbst, Salvador Miralbés, Bharath Naravetla, Alejandro Spiotta, Christian Loehr, Mario Martínez-Galdámez, Ryan McTaggart, Luc Defreyne, Pedro Vega, et al.

    BACKGROUND AND PURPOSE: Mechanical thrombectomy (MT) is effective for acute ischemic stroke, yet its indication in mild stroke remains unclear. This study evaluates MT9s effectiveness and safety in low NIHSS patients and assesses different MT strategies9 impact on procedural success and clinical outcomes. MATERIALS AND METHODS: Data from the ASSIST Registry were analyzed, categorizing patients with large vessel occlusion of the anterior circulation into mild (NIHSS≤5) and moderate-severe (NIHSS>5) stroke groups. Baseline characteristics, procedural parameters, angiographic and imaging outcomes, clinical outcomes, and safety endpoints were compared. Within the mild stroke subgroup, outcomes were compared between different MT techniques. RESULTS: Among 1360 patients with LVO, 122 had minor ischemic strokes (9%). Mild stroke patients had high rates of excellent functional outcomes (mRS 0-1) at 90 days (77.1%) and functional independence (mRS 0–2) (85.7%). Procedural success rates were similar between NIHSS groups, while safety outcomes, except mortality, were comparable. No statistically significant differences were observed in treatment techniques within the mild stroke subgroup. Significant predictors of early neurological deterioration (END) in mild stroke patients were the total number of passes (OR 1.49, 95% CI 1.01 – 2.19, P=.04) and total procedure time (OR 1.02, 95% CI 1.01 – 1.04, P=.01). Patients with END were more likely to have an unfavorable functional outcome (mRS 3-6) at 90 days (89% vs. 6%, P<0.001). CONCLUSIONS: MT is effective and safe in mild stroke patients. Procedural success did not vary among MT techniques in mild stroke. The total number of passes predicts END, which suggests a causal pathway that requires further exploration. ABBREVIATIONS: AIS = acute ischemic stroke; BMT = best medical treatment; DA = direct aspiration; END = early neurological deterioration; EVT = endovascular treatment; LVO = large vessel occlusion; MT = mechanical thrombectomy; SAEs = serious adverse events; sICH = symptomatic intracranial hemorrhage; SR = stent retriever.


  • Open Access

    Benchmarking four algorithms for improved classification of agricultural injury cases from free-text analysis of pre-hospital care reports (Preprint)

    by Laura E. Jones, Erika Scott, Nicole Krupa, Megan Kern, Cristina Silvia Hansen-Ruiz, Paul Jenkins

    BACKGROUND Fatality rates in Agriculture, Forestry, and Fishing (AgFF) industries are historically the highest of any US sector, with a combined rate of 18.6 deaths per 100,000 workers. Despite clear trends for fatal AgFF workplace injuries in federal data, challenges remain in capturing nonfatal agricultural injuries. The Northeast Center for Occupational Health and Safety (NEC) developed a naïve Bayes-based classification strategy to extract non-fatal injury cases from pre-hospital (EMS) free-text records. OBJECTIVE The aim of this paper is to improve retrieval rates, in terms of false positive rate required to obtain a true positive rate of 0.90, by benchmarking naïve Bayes against three other algorithms: elastic net regression, Support Vector Machines, and boosted decision trees (XGBoost). METHODS Using a labeled, fully one-hot coded gold-standard dataset (N=60,143) with substantial (24%) missing data, we benchmark these algorithms on complete case data (N=44,566) and imputed data from two imputation schemes: grouped hot-deck and recoding of missing units to the category “unknown,” using a 75:25 train/test split and stratified sampling. RESULTS All models produced similarly accuracies (~0.98) on complete case data, though necessary False Positive Rates (FPR*) varied from 0.055 (XGBoost) to 0.20 (naïve Bayes) on training data, and on predictions, the range was 0.10 (elastic net) to 0.22 (naïve Bayes). On imputed data, accuracies ranged from 0.96 (Bayes) to 0.98 (XGBoost) for training data, yielding false positive rates from 0.095 (XGBoost) to 0.34 (Bayes). Predictions from imputed data showed FPR* ranging from 0.12 (XGBoost) to 0.41 (Bayes) depending on imputation scheme. CONCLUSIONS While all four models perform well on complete data, missing units are substantial and can result in misclassification and in omissions, both requiring human coding. Reliance on a machine learning method that is robust to missing data and imputation method, such as XGBoost, is a reasonable approach to improving classification rates without omitting data.

  • Open Access

    A Case Study on the Practical Use of Low-Fidelity Modeling to Mitigate the Spread of COVID-19 Amongst the Underserved Farmworker Community

    by Leigh McCue-Weil, Margaret Knight, Maryellen Driscoll, Paul Jenkins, Julie Sorensen

    In March of 2020, with the full magnitude of the COVID-19 pandemic yet to be seen, Costa and Martin released a report through the Economic Policy Institute noting that “To prevent infections and the spread of COVID-19 on farms, farm employers should be planning and implementing safety measures to protect their employees” (Costa D, Martin P, Coronavirus and farmworkers: farm employment, safety issues, and the H-2A guestworker program, Economic Policy Institute, https://www.epi.org/publication/coronavirus-and-farmworkers-h-2a/, 2020). The report goes on to provide multiple observations recognizing the seasonal nature of farm work, effects increased unemployment may have on the workforce, industry dependence on H-2A visa farm workers, impact school closings would have on worker availability, and includes recommendations for safety equipment, social distancing, as well as worker housing and transportation. This paper focuses on the worker housing component of those recommendations and describes an effort to rapidly develop and deploy a computationally efficient, web-based, low-fidelity mathematical model of COVID-19 spread in dormitory style housing to support education and mitigation strategies for the historically underserved farmworker community.


  • Open Access

    COVID MED – An Early Pandemic Randomized Clinical Trial of Losartan for Hospitalized COVID-19 Patients

    by Daniel Freilich, Jennifer Victory, Paul Jenkins, James Wheeler, G. Matthew Vail, Erik Riesenfeld, Peggy Cross, Catherine Gilmore, Melissa Huckabone, Anna Schworm, et al.

    Background: ACEi/ARB medications have been hypothesized to have potential benefit in COVID-19. Despite concern for increased ACE-2 expression in some animal models, preclinical and observational-retrospective and uncontrolled trials suggested possible benefit. Two RCTs of the ARB losartan from University of Minnesota showed no benefit yet safety signals for losartan in outpatient and hospitalized COVID-19 patients. COVID MED, started early in the pandemic, also assessed losartan in a RCT in hospitalized patients with COVID-19. Methods: COVID MED was a double-blinded, placebo-controlled, multicenter, platform randomized clinical trial (RCT). Hospitalized COVID-19 patients were randomized to receive standard care and hydroxychloroquine, lopinavir/ritonavir, losartan, or placebo. Hydroxychloroquine and lopinavir/ritonavir arms were discontinued after RCTs showed no benefit. We report data from the losartan arm compared to combined (lopinavir-ritonavir and placebo) and prespecified placebo-only controls. The primary endpoint, the mean COVID-19 Ordinal Severity Score (COSS) slope of change, was compared with the Student’s t-test. Slow enrollment prompted early termination. Results: Of 448 screened patients, 15 (3.5%) were randomized/enrolled, 9 to receive losartan and 6 to receive control (lopinavir/ritonavir [N=2], placebo [N=4]); 1 patient who withdrew prior to study drug was excluded yielding 14 patients for analysis (losartan [N=9] vs. control [N=5] [lopinavir/ritonavir [N=2], placebo [N=3]]). Most baseline parameters were balanced. Treatment with losartan was not associated with a difference in mean COSS slope of change in comparison with combined control (p=0.4) or placebo-only control (p=0.05) (trend favoring placebo). 60-day mortality and overall AE and SAE rates were numerically but not significantly higher with losartan. Conclusions: In this small blinded RCT in hospitalized COVID-19 patients, losartan did not improve outcome vs. control comparisons and was associated with adverse safety signals.

  • Preoperative body-related emotional distress and culture as predictors of outcomes of bariatric surgery

    by Shulamit Geller, Sigal Levy, Ofra Hyman, Paul L. Jenkins, Subhi Abu-Abeid, Gil Goldzweig

    Purpose Findings concerning the impact of bariatric surgical intervention on both psychological variables and weight loss are often controversial and misconstrued the world over. The aim of this study was to classify bariatric surgery patients according to patterns of preoperative measures that may predict postoperative psychological and physiological outcomes and to compare these patterns between two distinct cultures. Methods Of 169 consecutive bariatric surgery candidates from Israel and 81 candidates from the United States, 73 and 35 patients, respectively consented to be included in a follow-up phase. Body image dissatisfaction, emotional eating behaviors, risk of suicide, depressive symptoms, anxious symptoms, and percent excess weight loss were measured. K-means clustering procedure was used to classify bariatric surgery patients according to their preoperative body-related emotional distress, which was composed of body image dissatisfaction and emotional eating. The joint effect of culture and body-related emotional distress cluster on psychological distress was tested. Results The cluster analysis revealed two preoperative body-related emotional distress patterns: high body-related emotional distress and low body-related emotional distress. Following surgery, US patients showed a higher risk of suicide and lower excess weight loss than Israeli patients within only the high body-related emotional distress cluster (a significant interaction effect). Conclusion Preoperative assessment of body-related emotional distress patterns among bariatric surgery candidates may enable professionals to identify potential postoperative risks of suicide, anxiety, and decreased weight loss. The relationship between the body-related emotional distress cluster and outcome measures is culture dependent. Level III Case–control analytic study.