About

Jeffrey Stein is a Professor Emeritus in the Department of Mechanical Engineering at the University of Michigan. He holds a Ph.D. in Mechanical Engineering from the Massachusetts Institute of Technology, earned in 1983, along with a S.B., M.S., and B.S. degrees from MIT and a B.S. in Psychology from the University of Massachusetts. His research interests encompass systems and control, including machine design, control, monitoring, and diagnostics; physical system modeling; automated modeling; bond graph theory; and the proper modeling of active suspensions. He focuses on vehicle handling and ride models, high-efficiency dynamic formulations for vehicle dynamics, and the monitoring and control of thermally induced spindle bearing loads, as well as the design and control of high-speed spindles and novel milling machines.

Research topics

• Medicine
• Ophthalmology
• Optometry
• Computer science
• Internal medicine

Selected publications

• Contributors to negative biopsychosocial outcomes in rugby players (CoNBO): part 1 the systematic review

  BMJ Open Sport & Exercise Medicine · 2026-01-01

  articleOpen access

  Objective: This review aimed to establish the contributors to negative biopsychosocial outcomes in rugby, defined as unexpected adverse changes in players' physical, psychological, social or health status. Design: Systematic review. Data sources: PubMed, Scopus, MEDLINE, SPORTDiscus and CINAHL. Eligibility criteria: Studies were eligible if they investigated a physical, psychological or social factor which results in a negative biopsychosocial outcome in men's or women's rugby union, league or sevens. Studies were excluded if they did not differentiate outcome measures between rugby and other sports or did not differentiate outcome measures (ie, positive or negative) between physical/psychological/social factors and other factors. Results: 9165 studies were identified in the initial search and two studies were identified from reference lists, 151 of which met the inclusion criteria (104 rugby union, 46 rugby league, 6 rugby sevens; 141 men, 16 women; 37 youth populations). 29 contributor groups and eight negative biopsychosocial outcome groups were identified. Previous injury (n=26), physical characteristics (n=32), training and match load (n=30) and factors within the contact event (n=22) were the most identified contributor groups. The negative biopsychosocial outcome of injury was investigated by 84% of studies. Conclusion: Overall, the systematic review summarises the contributors to negative biopsychosocial outcomes within the current evidence base. There is a focus on previous injury, physical characteristics, training and match load and factors within the contact event as contributors to negative biopsychosocial outcomes. Eight studies investigated women's cohorts independently from men; this underrepresentation within the literature could lead to the potential omittance of women-specific contributors. Prospero registration: https://www.crd.york.ac.uk/PROSPERO/view/CRD42022346751.

  PublisherOA PDFDOI

• Association of Eye Drop–Treated Diseases and Conditions That Can Impair Eye Drop Self-Administration

  Ophthalmology Glaucoma · 2025-04-08 · 1 citations

  articleOpen accessCorresponding
  PublisherOA PDFDOI

• Association between sociodemographic factors and visual impairment at initial presentation: A SOURCE data repository analysis

  AJO International · 2025-12-11

  articleOpen access
  PublisherDOI

• Ensemble learning to enhance accurate identification of patients with glaucoma using electronic health records

  JAMIA Open · 2025-07-03 · 1 citations

  articleOpen access

  Abstract Objectives Existing ophthalmology studies for clinical phenotypes identification in real-world datasets (RWD) rely exclusively on structured data elements (SDE). We evaluated the performance, generalizability, and fairness of multimodal ensemble models that integrate real-world SDE and free-text data compared to SDE-only models to identify patients with glaucoma. Materials and Methods This is a retrospective cross-sectional study involving 2 health systems- University of Michigan (UoM) and Stanford University (SU). It involves 1728 patients visiting eye clinics during 2012-2021. Free-text embeddings extracted using BioClinicalBERT were combined with SDE. EditedNearestNeighbor (ENN) undersampling and Borderline-Synthetic Minority Over-sampling Technique (bSMOTE) addressed class imbalance. Lasso Regression (LR), Random Forest (RF), Support Vector Classifier (SVC) models were trained on UoM imbalanced (imb) and resampled data along with bagging ensemble method. Models were externally validated with SU data. Fairness was assessed using equalized odds difference (EOD) and Target Probability Difference (TPD). Results Among 900 and 828 patients from UoM and SU, 10% and 23% respectively had glaucoma as confirmed by ophthalmologists. At UoM, multimodal LRimb (F1 = 76.60 [61.90-88.89]; AUROC = 95.41 [87.01-99.63]) outperformed unimodal RFimb (F1 = 69.77 [52.94-83.64]; AUROC = 97.72 [95.95-99.18]) and ICD-coding method (F1 = 53.01 [39.51-65.43]; AUROC = 90.10 [84.59-93.93]). Bagging (BM = LRENN + LRbSMOTE) improved performance achieving an F1 of 83.02 [70.59-92.86] and AUROC of 97.59 [92.98-99.88]. During external validation BM achieved the highest F1 (68.47 [62.61-73.75]), outperforming unimodal (F1 = 51.26 [43.80-58.13]) and multimodal LRimb (F1 = 62.46 [55.95-68.24]). BM EOD revealed lower disparities for sex (<0.1), race (<0.5) and ethnicity (<0.5), and had least uncertainty using TDP analysis as compared to traditional models. Discussion Multimodal ensemble models integrating structured and unstructured EHR data outperformed traditional SDE models achieving fair predictions across demographic sub-groups. Among ensemble methods, bagging demonstrated better generalizability than stacking, particularly when training data is limited. Conclusion This approach can enhance phenotype discovery to enable future research studies using RWD, leading to better patient management and clinical outcomes.

  PublisherOA PDFDOI

• Enhanced Phenotype Identification of Common Ocular Diseases in Real-World Datasets

  Ophthalmology Science · 2025-01-24 · 1 citations

  articleOpen access1st authorCorresponding

  Objective: For studies using real-world data, accurately identifying patients with phenotypes of interest is challenging. To identify cohorts of interest, most studies exclusively use the International Classification of Diseases (ICD) billing codes, which can be limiting. We developed a method to accurately identify the presence or absence of 3 common ocular diseases (diabetic retinopathy [DR], age-related macular degeneration [AMD], and glaucoma) using electronic health record (EHR) data. Design: Database study. Participants: Three thousand nine hundred fourteen eyes from 1957 patients at 2 Sight OUtcomes Research CollaborativE (SOURCE) Ophthalmology Data Repository sites. Methods: We developed enhanced phenotype identification (EPI) algorithms that search EHR fields, including eye examination findings, orders, charges, medication prescriptions, and surgery data for evidence that a patient has glaucoma, DR, or AMD. We trained our EPI models using gold standard assessments of the EHR by ophthalmologists for the presence/absence of these conditions, compared the performance of our EPI models to models developed using ICD codes alone, and validated the performance of model using data from another SOURCE site. Main Outcome Measures: Area under the receiver operating curve (AUC), area under the precision-recall curve (AUPRC), and model calibration. Results: The AUCs of our EPI models were better than ICD-only models for glaucoma (0.97 vs. 0.90), DR (0.997 vs. 0.98), and AMD (0.99 vs. 0.95). The AUPRCs of our EPI models were also much better than ICD-only models for glaucoma (0.79 vs. 0.32), DR (0.96 vs. 0.84), and AMD (0.74 vs. 0.55). When testing on patients from a second SOURCE site, the AUC and AUPRC for glaucoma (0.93, 0.74), DR (0.98, 0.77), and AMD (0.96, 0.64) were slightly worse than the primary site but still quite high. However, for all 3 conditions, model calibration was worse at the second site. Conclusions: Leveraging machine learning, we developed EPI models to accurately identify most patients with glaucoma, DR, and AMD in real-world datasets. The EPI models significantly outperform ICD-only models in identifying patients confirmed to have these conditions. These findings underscore the potential of using comprehensive EHR data combined with advanced machine learning techniques to improve the accuracy of patient phenotype identification, leading to better patient management and clinical outcomes. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

  PublisherOA PDFDOI

• Risk Factors and Consequences of Lapses in Proliferative Diabetic Retinopathy Care in a National Cohort

  Diabetes Care · 2025-09-16 · 2 citations

  article

  OBJECTIVE: To identify prevalence, risk factors, and visual outcomes associated with occurrence and duration of lapses in proliferative diabetic retinopathy (PDR) care. RESEARCH DESIGN AND METHODS: This was a retrospective national cohort study (2008-2023) of adults with PDR and ≥6 months of follow-up who were participating in the Sight Outcomes Research Collaborative. We used multivariable regressions to assess factors associated with lapse occurrence and duration, and compared post-lapse visual acuity by lapse duration. RESULTS: Among 15,211 individuals, 71.8% experienced a lapse in care; 14.2% of the lapses lasted >24 months. Lapses were more common among non-Hispanic Black, younger, and individuals with disability, and less common in those with poor vision or prior PDR treatment. Older age and PDR treatment predicted shorter lapses, and residence in distressed areas predicted longer lapses. Visual acuity worsened after lapses, with greater declines after longer lapses. CONCLUSIONS: Prolonged lapses in PDR care are common, disproportionately affect vulnerable groups, and are associated with persistent vision loss.

  PublisherDOI

• Field Aging of Photovoltaic Module Packaging Materials: DuraMAT Field Module Library

  2025-01-01

  reportOpen accessSenior author

  To understand and develop models for silicon photovoltaic module degradation, accelerated testing is often used, however, outdoor field testing is necessary for validation. Outdoor field testing publications are often limited by the lack of a pristine, control module to compare the fielded module to. In this work, commercially available modules were purchased from seven different manufacturers for outdoor fielding then destructive characterization to investigate packaging material degradation on

  PublisherOA PDFDOI

• Disparities in Medicaid and Medicare physician reimbursements for ophthalmic procedures

  PLoS ONE · 2025-06-18 · 1 citations

  articleOpen accessCorresponding
  PublisherDOI

• Exercise Counteracts Inflammation And Muscle Proteolysis Mediators In Tumor-bearing Mice

  Medicine & Science in Sports & Exercise · 2025-09-16

  article

  Cancer cachexia is characterized by unexplained weight loss accompanied by muscle atrophy, which may result from abnormal increases in inflammation and E3 ubiquitin ligases. While the benefits of exercise countermeasures (EC) for enhancing muscle mass are well recognized, their potential to combat muscle loss and downregulate inflammatory and proteolysis mediators remains to be elucidated. PURPOSE: To determine the clinical efficacy of different types of EC on alterations in muscle wet weight and inflammatory mediators and E3 ubiquitin ligases in ApcMin/+ mice, a model of intestinal neoplasia. METHODS: Forty ApcMin/+ mice were randomly assigned to sedentary control (SC-Apc n = 10), aerobic exercise (A-Apc n = 10), resistance exercise (R-Apc n = 10), and combination of aerobic and resistance exercise (AR-Apc n = 10) for an 8-wk intervention period. Additionally, ten wild-type C57BL mice (WT) served as a control group. The A-Apc group performed treadmill running at a speed of 18 m/min for 55 min at a 5% grade. The R-Apc group performed 3 sets of 4 repetition ladder climbs with weight attached to the base of their tail. The AR-Apc group performed the combined aerobic and strength program. Gastrocnemius wet weight and mRNA levels of inflammatory mediators [Tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6)] and E3 ubiquitin ligases (atrogin-1) were assessed at post-intervention. Statistical analyses were conducted using one-way ANOVA. RESULTS: Following the 8-wk intervention, the gastrocnemius wet weight of A-APC was 17.1% higher than SC-APC, which approached statistical significance (p = 0.076). TNF-α mRNA expression in SC-APC was significantly greater than WT (+145 ± 59%, p = 0.024) and was non-significantly higher than AR-APC (+123 ± 54%, p = 0.085). SC-APC presented greater IL-6 mRNA levels compared to WT (+191 ± 71%, p = 0.024), with a non-significant higher level than AR-APC (+116 ± 45%, p = 0.092). Atrogin-1 mRNA expression in SC-APC was greatly higher compared to WT (+634 ± 106%, p < 0.001) and AR-APC (+137 ± 89%, p = 0.021). CONCLUSION: Our findings indicate that aerobic exercise may help mitigate muscle wasting, while combined exercise appears to attenuate proteolysis mediators in tumor-bearing mice, suggesting EC as a promising therapeutic strategy to counteract catabolic syndrome in cancer patients. Supported by: Supported by U54 Partnership for The Advancement of Cancer Research (PACR) Grant.

  PublisherDOI

• Quality of Care in Patients With Newly Diagnosed Glaucoma

  JAMA Ophthalmology · 2025-09-18 · 4 citations

  articleOpen access

  Importance: Various communities continue to experience relatively high rates of glaucoma-related visual impairment and blindness. Identifying potential nonmedical influences on glaucoma outcomes may lead to strategies to improve glaucoma care. Objective: To assess possible associations between nonmedical variables and quality of glaucoma care among patients with newly diagnosed primary open-angle glaucoma (POAG). Design, Setting, and Participants: This retrospective cohort study included 1466 patients with newly diagnosed POAG receiving care at health systems in the Sight Outcomes Research Collaborative (SOURCE) Consortium from January 2010 to December 2022. Data analysis was completed from March 2024 to June 2025. Exposures: Various nonmedical variables, including self-reported race and ethnicity, urbanicity of residence, affluence of patients' residential community, and presence of children in the household. Main Outcomes and Measures: The primary outcomes were odds of 15% or greater intraocular pressure (IOP) reduction at 12 to 18 months following initial POAG diagnosis and odds of loss to follow-up (LTFU). Results: Mean (SD) age of patients was 70 (12) years; among 1466 patients, 793 (54%) were female. By self-reported race and ethnicity, 39 patients (3%) were Asian American, 469 patients (32%) were Black, 95 (7%) were Latinx, and 831 (57%) were White. Among 1030 patients (70%) with 1 or more follow-up evaluations within 12 to 18 months following initial POAG diagnosis, 783 (76%) achieved 15% or higher IOP reduction in 1 or more eyes. Patients in the lowest wealth quartile had 5- to 9-fold lower odds of achieving 15% or greater IOP decrease compared with patients in higher quartiles; the odds of LTFU were 61% lower in the wealthiest patient quartile than in the least-wealthy group (odds ratio [OR], 0.39; 95% CI, 0.18-0.84; P = .02). Patients in rural communities (OR, 5.54; 95% CI, 1.13-27.08) were more likely than urban residents to experience LTFU. Patients with children in the household experienced, on average, a 4-mm Hg (95% CI, 0.99-7.13) greater IOP reduction compared with those without children in the household (P = .01). Conclusions and Relevance: In this cohort study, patients with newly diagnosed POAG in the lowest wealth quartile were substantially less likely to achieve the US National Quality Forum's recommended IOP percentage reduction and considerably more likely to experience LTFU than those with higher wealth levels. These findings support the premise that clinicians should understand financial circumstances of patients when making management decisions and reinforce the need for clinicians and payors to find ways to ensure that patients can access IOP-lowering interventions and receive follow-up care in accordance with established guidelines.

  PublisherOA PDFDOI

Recent grants

• NIH Grant R34EY022399

  NIH · $233k · 2014

• Personalized Forecasting of Disease Trajectory for Patients with Open Angle Glaucoma

  NIH · $2.9M · 2016–2022

• NIH Grant K23EY019511

  NIH · $1.0M · 2014

Frequent coauthors

• Gary C. Brown

  Center for Value Based Medicine (United States)

  103 shared

• Melissa M. Brown

  Center for Value Based Medicine (United States)

  99 shared

• Nidhi Talwar

  University of Michigan–Ann Arbor

  90 shared

• Paul P. Lee

  Wayne State University

  74 shared

• Chris Andrews

  University of Michigan–Ann Arbor

  72 shared

• Bin Nan

  Shanghai Institute of Technology

  61 shared

• David C. Musch

  University of Michigan–Ann Arbor

  59 shared

• Janey L. Wiggs

  Harvard University

  59 shared

Awards & honors

• Michael J. Rabins Leadership Award, ASME Dynamic Systems and…
• Dedicated Service Award, American Society of Mechanical Engi…
• Fellow, American Society of Mechanical Engineers, 2005
• ME Teacher Incentive Program Outstanding Teacher, Department…
• Distinguished Publication Award, Journal of Dynamic Systems…