Belatacept with time-limited tacrolimus coimmunosuppression modifies the 3-year risk of eplet mismatch in kidney transplantation.

Solid organ transplant donor-recipient eplet mismatch has been correlated with donor-specific antibody (DSA) formation, antibody-mediated rejection, and overall rejection rates. However, studies have been predominantly in patients on tacrolimus-based immunosuppression regimens and have not fully explored differences in ethnically and racially diverse populations. Evidence indicates that patients on belatacept have lower rates of DSA formation, suggesting mediation of the immunogenicity of mismatched human leukocyte antigen polymorphisms. We performed a retrospective, single-center analysis of class II eplet disparity in a cohort of kidney transplant recipients treated using belatacept with tacrolimus induction (Bela/TacTL) or tacrolimus regimens between 2016 and 2019. Bela/TacTL (n = 294) and tacrolimus (n = 294) cohorts were propensity score-matched with standardized difference <0.15. Single-molecule eplet risk level was associated with immune event rates for both groups. In Cox regression analysis stratified by eplet risk level, Bela/TacTL immunosuppression was associated with a decreased rate of DSA (hazard ratio [HR] = 0.4), antibody-mediated rejection (HR = 0.2), and rejection (HR = 0.45). In the low-risk group, cumulative graft failure was lower for patients on Bela/TacTL (P < .02). Analysis of eplet mismatch burden may be a useful adjunct in identifying high-risk populations with increased immunosuppression requirements and should encourage the design of allocation rules to incentivize lower-risk pairings without negatively impacting equity in access.

Black-white differences in chronic stress exposures to predict preterm birth: interpretable, race/ethnicity-specific machine learning model.

BACKGROUND: Differential exposure to chronic stressors by race/ethnicity may help explain Black-White inequalities in rates of preterm birth. However, researchers have not investigated the cumulative, interactive, and population-specific nature of chronic stressor exposures and their possible nonlinear associations with preterm birth. Models capable of computing such high-dimensional associations that could differ by race/ethnicity are needed. We developed machine learning models of chronic stressors to both predict preterm birth more accurately and identify chronic stressors and other risk factors driving preterm birth risk among non-Hispanic Black and non-Hispanic White pregnant women. METHODS: Multivariate Adaptive Regression Splines (MARS) models were developed for preterm birth prediction for non-Hispanic Black, non-Hispanic White, and combined study samples derived from the CDC's Pregnancy Risk Assessment Monitoring System data (2012-2017). For each sample population, MARS models were trained and tested using 5-fold cross-validation. For each population, the Area Under the ROC Curve (AUC) was used to evaluate model performance, and variable importance for preterm birth prediction was computed. RESULTS: Among 81,892 non-Hispanic Black and 277,963 non-Hispanic White live births (weighted sample), the best-performing MARS models showed high accuracy (AUC: 0.754-0.765) and similar-or-better performance for race/ethnicity-specific models compared to the combined model. The number of prenatal care visits, premature rupture of membrane, and medical conditions were more important than other variables in predicting preterm birth across the populations. Chronic stressors (e.g., low maternal education and intimate partner violence) and their correlates predicted preterm birth only for non-Hispanic Black women. CONCLUSIONS: Our study findings reinforce that such mid or upstream determinants of health as chronic stressors should be targeted to reduce excess preterm birth risk among non-Hispanic Black women and ultimately narrow the persistent Black-White gap in preterm birth in the U.S.

Evaluating Natural Language Processing Packages for Predicting Hospital-Acquired Pressure Injuries From Clinical Notes.

Incidence of hospital-acquired pressure injury, a key indicator of nursing quality, is directly proportional to adverse outcomes, increased hospital stays, and economic burdens on patients, caregivers, and society. Thus, predicting hospital-acquired pressure injury is important. Prediction models use structured data more often than unstructured notes, although the latter often contain useful patient information. We hypothesize that unstructured notes, such as nursing notes, can predict hospital-acquired pressure injury. We evaluate the impact of using various natural language processing packages to identify salient patient information from unstructured text. We use named entity recognition to identify keywords, which comprise the feature space of our classifier for hospital-acquired pressure injury prediction. We compare scispaCy and Stanza, two different named entity recognition models, using unstructured notes in Medical Information Mart for Intensive Care III, a publicly available ICU data set. To assess the impact of vocabulary size reduction, we compare the use of all clinical notes with only nursing notes. Our results suggest that named entity recognition extraction using nursing notes can yield accurate models. Moreover, the extracted keywords play a significant role in the prediction of hospital-acquired pressure injury.

Sphinganine is associated with 24-h MAP in the non-sleepy with OSA.

INTRODUCTION: Excessive daytime sleepiness is a debilitating symptom of obstructive sleep apnea (OSA) linked to cardiovascular disease, and metabolomic mechanisms underlying this relationship remain unknown. We examine whether metabolites from inflammatory and oxidative stress-related pathways that were identified in our prior work could be involved in connecting the two phenomena. METHODS: This study included 57 sleepy (Epworth Sleepiness Scale (ESS) ≥ 10) and 37 non-sleepy (ESS < 10) participants newly diagnosed and untreated for OSA that completed an overnight in-lab or at home sleep study who were recruited from the Emory Mechanisms of Sleepiness Symptoms Study (EMOSS). Differences in fasting blood samples of metabolites were explored in participants with sleepiness versus those without and multiple linear regression models were utilized to examine the association between metabolites and mean arterial pressure (MAP). RESULTS: The 24-h MAP was higher in sleepy 92.8 mmHg (8.4) as compared to non-sleepy 88.8 mmHg (8.1) individuals (P = 0.03). Although targeted metabolites were not significantly associated with MAP, when we stratified by sleepiness group, we found that sphinganine is significantly associated with MAP (Estimate = 8.7, SE = 3.7, P = 0.045) in non-sleepy patients when controlling for age, BMI, smoking status, and apnea-hypopnea index (AHI). CONCLUSION: This is the first study to evaluate the relationship of inflammation and oxidative stress related metabolites in sleepy versus non-sleepy participants with newly diagnosed OSA and their association with 24-h MAP. Our study suggests that Sphinganine is associated with 24 hour MAP in the non-sleepy participants with OSA.

A modified physiological strain index for workplace-based assessment of heat strain experienced by agricultural workers.

BACKGROUND: As global temperatures rise, increasing numbers of individuals will work in hot environments. Interventions to protect their health are critical, as are reliable methods to measure the physiological strain experienced from heat exposure. The physiological strain index (PSI) is a measure of heat strain that relies on heart rate and core temperature but is challenging to calculate in a real-world occupational setting. METHODS: We modified the PSI for use in field settings where resting temperature and heart rate are not available and used the modified physiological strain index (mPSI) to describe risk factors for high heat strain (mPSI ≥ 7) experienced by agricultural workers in Florida during the summers of 2015 through 2017. mPSI was calculated for 221 workers, yielding 465 days of data. RESULTS: A higher heat index (ß = 0.185; 95% CI: 0.064, 0.307) and higher levels of physical activity at work (0.033; 95% CI: 0.017, 0.050) were associated with a higher maximum mPSI. More years worked in US agriculture (-0.041; 95% CI: -0.061, -0.020) were protective against a higher maximum mPSI. Out of 23 workdays that a participant experienced a maximum mPSI ≥ 7, 22 were also classified as strained by at least one other measure of high heat strain (core temperature [Tc] >38.5°C, sustained heart rate >(180 - age), and mean heart rate > 115 bpm). CONCLUSIONS: This study provides critical information on risk factors for elevated heat strain for agricultural workers and suggests a practical approach for using PSI in field-based settings.

Behaviors, movements, and transmission of droplet-mediated respiratory diseases during transcontinental airline flights.

With over 3 billion airline passengers annually, the inflight transmission of infectious diseases is an important global health concern. Over a dozen cases of inflight transmission of serious infections have been documented, and air travel can serve as a conduit for the rapid spread of newly emerging infections and pandemics. Despite sensational media stories and anecdotes, the risks of transmission of respiratory viruses in an airplane cabin are unknown. Movements of passengers and crew may facilitate disease transmission. On 10 transcontinental US flights, we chronicled behaviors and movements of individuals in the economy cabin on single-aisle aircraft. We simulated transmission during flight based on these data. Our results indicate there is low probability of direct transmission to passengers not seated in close proximity to an infectious passenger. This data-driven, dynamic network transmission model of droplet-mediated respiratory disease is unique. To measure the true pathogen burden, our team collected 229 environmental samples during the flights. Although eight flights were during Influenza season, all qPCR assays for 18 common respiratory viruses were negative.

Using Twitter for Nursing Research: A Tweet Analysis on Heat Illness and Health.

AIM: To provide an example of a tweet analysis for nurse researchers using Twitter in their research. DESIGN: A content analysis using tweets about "heat illness + health." METHODS: Tweets were pulled from Twitter's application programming interface with premium access using Postman and the key words "heat illness + health." All data cleaning and analysis was performed in R Version 3.5.2, and the tweet set was analyzed for term frequency, sentiment, and topic modeling. Principal R packages included LDAvis, tidytext, tm, and zyuzhet. RESULTS: 6,317 tweets were analyzed with a date range of April 6, 2009, to December 30, 2019. The most common terms in the tweets were heat (n = 4,532), illness (n = 4,085), and health (n = 2,257). Sentiment analysis showed that the majority of tweets (55%) had a negative sentiment. Topic modeling showed that there were three topics within the tweet set: increasing impact, prevention and safety, and symptoms. CONCLUSIONS: Twitter can be a useful tool for nursing researchers, serving as a viable adjunct to current research methodologies. This practical example has facilitated a deeper understanding of the social media representation of heat illness and health that can be applied to other research. CLINICAL RELEVANCE: Twitter serves as a tool for collecting health information for multiple groups, ranging from clinicians and researchers to patients. By utilizing the plethora of data that comes from the platform, we can work towards developing theories and interventions related to numerous health conditions and phenomena.

Toolkit for Best Practice Use of Electronic Health Record Data in Quality Improvement.

This project piloted an educational intervention focused on use and management of EHR data by Doctor of Nursing Practice students in quality improvement initiatives. Recommendations from academic and clinical nursing promote the integration of EHR data findings into practice. Nursing's general lack of understanding about how to use and manage data is a barrier to using EHR data to guide quality improvement initiatives. Doctor of Nursing Practice students at a hospital-affiliated university participated in a pre-test, training, and post-test through an online learning management system. Training content and assessments focused on data and planning for its use in quality improvement initiatives. Sixteen students experienced a median of 17.6% increase in scores after completing the post-test. There was a statistically significant increase in scores between the pre-test and post-test (P = .0006). These results suggest educational content included in the Doctor of Nursing Practice Quality Improvement Toolkit increases knowledge about use and management of EHR data. Future considerations include use for educating a variety of students and healthcare staff.

Analyzing the predictors of health care utilization in the agricultural worker population using decision tree analysis: Does language matter?

OBJECTIVES: To analyze the predictors of health care utilization among respondents to the National Agricultural Worker Survey. Specifically, we hypothesized that English proficiency would predict utilization of health care services within the last 2 years. METHODS: Using the 2015-2016 National Agricultural Worker Survey, we performed a secondary data analysis to analyze the predictors of health care utilization within the last 2 years in the United States' agricultural worker population. Data were cleaned and analyzed using decision tree analysis, which produced a classification tree model that was trained on 90% of the data and validated on 10%. RESULTS: Exposure to English was not a predictor of health care utilization in our classification tree. The first major partition that predicted utilization was insurance status. Additional partitions were on age, gender, hypertension diagnosis, and public aid. CONCLUSIONS: By partitioning on insurance status and use of public aid, the decision tree provided evidence that systemic factors are key determinants of health care utilization in the agricultural worker community. This highlights the importance of agencies that connect agricultural workers with resources that provide insurance and improve access to health care. This is especially important given that agricultural workers are one of the highest risk groups for occupational injury or death in the United States.

Cooling intervention studies among outdoor occupational groups: A review of the literature.

BACKGROUND: The purpose of this systematic review is to examine cooling intervention research in outdoor occupations, evaluate the effectiveness of such interventions, and offer recommendations for future studies. This review focuses on outdoor occupational studies conducted at worksites or simulated occupational tasks in climatic chambers. METHODS: This systematic review was performed in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. PubMed, Embase, and Web of Science were searched to identify original research on intervention studies published in peer-reviewed journals that aimed at reducing heat stress or heat-related illness from January 2000 to August 2020. RESULTS: A systematic search yielded a total of 1042 articles, of which 21 met the inclusion criteria. Occupations with cooling intervention studies included agriculture (n = 5), construction (n = 5), industrial workers (n = 4), and firefighters (n = 7). The studies focused on multiple types of cooling interventions cooling gear (vest, bandanas, cooling shirts, or head-cooling gel pack), enhanced heat dissipation clothing, forearm or lower body immersion in cold water, water dousing, ingestion of a crushed ice slush drink, electrolyte liquid hydration, and modified Occupational Safety and Health Administration recommendations of drinking water and resting in the shade. CONCLUSION: Current evidence indicates that using multiple cooling gears along with rest cycles may be the most effective method to reduce heat-related illness. Occupational heat-related illnesses and death may be mitigated by targeted cooling intervention and workplace controls among workers of vulnerable occupational groups and industries.

Use of Palliative Care Consultation Services for Infants With Life-Threatening Conditions in a Metropolitan Hospital.

BACKGROUND: Palliative care is becoming an important component for infants with life-limiting or life-threatening conditions and their families. Yet palliative care practices appear to be inconsistent and sporadically used for infants. PURPOSE: The purpose of this study was to describe the use of an established pediatric palliative care team for seriously ill infants in a metropolitan hospital. METHODS: This was a retrospective medical record review. FINDINGS: The population included 64 infants who were admitted to a level IV neonatal intensive care unit (NICU) and then died during hospitalization between January 2015 and December 2016. Most infants died in an ICU (n = 63, 95%), and only 20 infants (31%) received palliative care consultation. Most common reasons for consultation were care coordination, defining goals of care and end-of-life planning, and symptom management. IMPLICATIONS FOR PRACTICE: Palliative care consultation at this institution did not change the course of end-of-life care. Interventions provided by the ICU team to infants surrounding end of life were similar to those in infants receiving palliative care services from the specialists. Our findings may be useful for developing guidelines regarding how to best utilize palliative care services for infants with life-threatening conditions who are admitted to an ICU. IMPLICATIONS FOR RESEARCH: These finding support continued research in neonatal palliative care, more specifically the impact of palliative care guidelines and algorithms.

The Airplane Cabin Microbiome.

Serving over three billion passengers annually, air travel serves as a conduit for infectious disease spread, including emerging infections and pandemics. Over two dozen cases of in-flight transmissions have been documented. To understand these risks, a characterization of the airplane cabin microbiome is necessary. Our study team collected 229 environmental samples on ten transcontinental US flights with subsequent 16S rRNA sequencing. We found that bacterial communities were largely derived from human skin and oral commensals, as well as environmental generalist bacteria. We identified clear signatures for air versus touch surface microbiome, but not for individual types of touch surfaces. We also found large flight-to-flight beta diversity variations with no distinguishing signatures of individual flights, rather a high between-flight diversity for all touch surfaces and particularly for air samples. There was no systematic pattern of microbial community change from pre- to post-flight. Our findings are similar to those of other recent studies of the microbiome of built environments. In summary, the airplane cabin microbiome has immense airplane to airplane variability. The vast majority of airplane-associated microbes are human commensals or non-pathogenic, and the results provide a baseline for non-crisis-level airplane microbiome conditions.

Physical activity and work activities in Florida agricultural workers.

BACKGROUND: Laboring in hot and humid conditions is a risk factor for heat-related illnesses. Little is known about the amount of physical activity performed in the field setting by agricultural workers, a population that is among those at highest risk for heat-related mortality in the United States. METHODS: We measured accelerometer-based physical activity and work activities performed in 244 Florida agricultural workers, 18 to 54 years of age, employed in the fernery, nursery, and crop operations during the summer work seasons of 2015-2017. Environmental temperature data during the participant's workdays were collected from the Florida Automated Weather Network and used to calculate wet bulb globe temperature (WBGT). Generalized linear mixed model regression was used to examine the association between WBGT on physical activity, stratified by the agricultural sector. RESULTS: Fernery workers had the highest overall volume of physical activity, spending nearly 4 hours in moderate to vigorous activity per workday. Activity over the course of the workday also differed by the agricultural sector. A reduction on average physical activity with increasing environmental temperature was observed only among crop workers. CONCLUSIONS: The quantity and patterns of physical activity varied by the agricultural sector, sex, and age, indicating that interventions that aim to reduce heat-related morbidity and mortality should be tailored to specific subpopulations. Some workers did not reduce overall physical activity under dangerously hot environmental conditions, which has implications for policies protecting worker health. Future research is needed to determine how physical activity and climatic conditions impact the development of heat-related disorders in this population.

Stool Immune Profiles Evince Gastrointestinal Inflammation in Parkinson's Disease.

BACKGROUND: Gastrointestinal symptoms are common in Parkinson's disease and frequently precede the development of motor impairments. Intestinal inflammation has been proposed as a driver of disease pathology, and evaluation of inflammatory mediators in stool could possibly identify valuable early-stage biomarkers. We measured immune- and angiogenesis-related proteins in human stool to examine inflammatory profiles associated with Parkinson's disease. METHODS: Stool samples and subjects' self-reported metadata were obtained from 156 individuals with Parkinson's disease and 110 without, including spouse and nonhousehold controls. Metadata were probed for disease-associated differences, and levels of 37 immune and angiogenesis factors in stool homogenates were measured by multiplexed immunoassay and compared across experimental groups. RESULTS: Parkinson's disease patients reported greater incidence of intestinal disease and digestive problems than controls. Direct comparison of levels of stool analytes in patients and controls revealed elevated vascular endothelial growth factor receptor 1, interleukin-1α, and CXCL8 in patients' stool. Paired comparison of patients and spouses suggested higher levels of multiple factors in patients, but this was complicated by sex differences. Sex, body mass index, a history of smoking, and use of probiotics were found to strongly influence levels of stool analytes. Multivariate analysis accounting for these and other potential confounders confirmed elevated levels of interleukin-1α and CXCL8 and also revealed increased interleukin-1ß and C-reactive protein in stool in Parkinson's disease. These differences were not dependent on subject age or disease duration. CONCLUSIONS: Levels of stool immune factors indicate that intestinal inflammation is present in patients with Parkinson's disease. © 2018 International Parkinson and Movement Disorder Society.

Very early administration of progesterone for acute traumatic brain injury.

BACKGROUND: Traumatic brain injury (TBI) is a major cause of death and disability worldwide. Progesterone has been shown to improve neurologic outcome in multiple experimental models and two early-phase trials involving patients with TBI. METHODS: We conducted a double-blind, multicenter clinical trial in which patients with severe, moderate-to-severe, or moderate acute TBI (Glasgow Coma Scale score of 4 to 12, on a scale from 3 to 15, with lower scores indicating a lower level of consciousness) were randomly assigned to intravenous progesterone or placebo, with the study treatment initiated within 4 hours after injury and administered for a total of 96 hours. Efficacy was defined as an increase of 10 percentage points in the proportion of patients with a favorable outcome, as determined with the use of the stratified dichotomy of the Extended Glasgow Outcome Scale score at 6 months after injury. Secondary outcomes included mortality and the Disability Rating Scale score. RESULTS: A total of 882 of the planned sample of 1140 patients underwent randomization before the trial was stopped for futility with respect to the primary outcome. The study groups were similar with regard to baseline characteristics; the median age of the patients was 35 years, 73.7% were men, 15.2% were black, and the mean Injury Severity Score was 24.4 (on a scale from 0 to 75, with higher scores indicating greater severity). The most frequent mechanism of injury was a motor vehicle accident. There was no significant difference between the progesterone group and the placebo group in the proportion of patients with a favorable outcome (relative benefit of progesterone, 0.95; 95% confidence interval [CI], 0.85 to 1.06; P=0.35). Phlebitis or thrombophlebitis was more frequent in the progesterone group than in the placebo group (relative risk, 3.03; CI, 1.96 to 4.66). There were no significant differences in the other prespecified safety outcomes. CONCLUSIONS: This clinical trial did not show a benefit of progesterone over placebo in the improvement of outcomes in patients with acute TBI. (Funded by the National Institute of Neurological Disorders and Stroke and others; PROTECT III ClinicalTrials.gov number, NCT00822900.).

Candidate inflammatory biomarkers display unique relationships with alpha-synuclein and correlate with measures of disease severity in subjects with Parkinson's disease.

BACKGROUND: Efforts to identify fluid biomarkers of Parkinson's disease (PD) have intensified in the last decade. As the role of inflammation in PD pathophysiology becomes increasingly recognized, investigators aim to define inflammatory signatures to help elucidate underlying mechanisms of disease pathogenesis and aid in identification of patients with inflammatory endophenotypes that could benefit from immunomodulatory interventions. However, discordant results in the literature and a lack of information regarding the stability of inflammatory factors over a 24-h period have hampered progress. METHODS: Here, we measured inflammatory proteins in serum and CSF of a small cohort of PD (n = 12) and age-matched healthy control (HC) subjects (n = 6) at 11 time points across 24 h to (1) identify potential diurnal variation, (2) reveal differences in PD vs HC, and (3) to correlate with CSF levels of amyloid ß (Aß) and α-synuclein in an effort to generate data-driven hypotheses regarding candidate biomarkers of PD. RESULTS: Despite significant variability in other factors, a repeated measures two-way analysis of variance by time and disease state for each analyte revealed that serum IFNγ, TNF, and neutrophil gelatinase-associated lipocalin (NGAL) were stable across 24 h and different between HC and PD. Regression analysis revealed that C-reactive protein (CRP) was the only factor with a strong linear relationship between CSF and serum. PD and HC subjects showed significantly different relationships between CSF Aß proteins and α-synuclein and specific inflammatory factors, and CSF IFNγ and serum IL-8 positively correlated with clinical measures of PD. Finally, linear discriminant analysis revealed that serum TNF and CSF α-synuclein discriminated between PD and HC with a minimum of 82% sensitivity and 83% specificity. CONCLUSIONS: Our findings identify a panel of inflammatory factors in serum and CSF that can be reliably measured, distinguish between PD and HC, and monitor inflammation as disease progresses or in response to interventional therapies. This panel may aid in generating hypotheses and feasible experimental designs towards identifying biomarkers of neurodegenerative disease by focusing on analytes that remain stable regardless of time of sample collection.

Contact networks in the emergency department: Effects of time, environment, patient characteristics, and staff role.

Emergency departments play a critical role in the public health system, particularly in times of pandemic. Infectious patients presenting to emergency departments bring a risk of cross-infection to other patients and staff through close proximity interactions or contacts. To understand factors associated with cross-infection risk, we measured close proximity interactions of emergency department staff and patients by radiofrequency identification in a working emergency department. The number of contacts (degree) is not related to patient demographic characteristics. However, the amount of time in close proximity (weighted degree) of patients with ED personnel did differ, with black patients having approximately 15 min more contact with staff than non-white patients. Patients arriving by EMS had fewer contacts with other patients than patients arriving by other means. There are differences in the number of contacts based on staff role and arrival mode. When crowding is low, providers have the most contact time with patients, while administrative staff have the least. However, when crowding is high, this differential is reversed. The effect of arrival mode is modified by the extent of crowding. When crowding is low, patients arriving by EMS had longer contact with administrative staff, compared to patients arriving by other means. However, when crowding is high, patients arriving by EMS had less contact with administrative staff compared to patients arriving by other means. Our findings should help designers of emergency care focus on higher risk situations for transmission of dangerous pathogens in an emergency department. For instance, the effects of arrival and crowding should be considered as targets for engineering or architectural interventions that could artificially increase social distances.

An AdaBoost-based algorithm to detect hospital-acquired pressure injury in the presence of conflicting annotations.

Hospital-acquired pressure injury is one of the most harmful events in clinical settings. Patients who do not receive early prevention and treatment can experience a significant financial burden and physical trauma. Several hospital-acquired pressure injury prediction algorithms have been developed to tackle this problem, but these models assume a consensus, gold-standard label (i.e., presence of pressure injury or not) is present for all training data. Existing definitions for identifying hospital-acquired pressure injuries are inconsistent due to the lack of high-quality documentation surrounding pressure injuries. To address this issue, we propose in this paper an ensemble-based algorithm that leverages truth inference methods to resolve label inconsistencies between various case definitions and the level of disagreements in annotations. Application of our method to MIMIC-III, a publicly available intensive care unit dataset, gives empirical results that illustrate the promise of learning a prediction model using truth inference-based labels and observed conflict among annotators.

Black-White Differences in Chronic Stress Exposures to Predict Preterm Birth: Interpretable, Race/Ethnicity-Specific Machine Learning Models.

We developed Multivariate Adaptive Regression Splines (MARS) machine learning models of chronic stressors using the Pregnancy Risk Assessment Monitoring System data (2012-2017) to predict preterm birth (PTB) more accurately and identify chronic stressors driving PTB among non-Hispanic (N-H) Black and N-H White pregnant women in the U.S. We trained the MARS models using 5-fold cross-validation, whose performance was evaluated with AUC. We computed variable importance for PTB prediction. Our models showed high accuracy (AUC: 0.754-0.765). The number of prenatal care visits, premature rupture of membrane, and medical conditions were the most important variables in predicting PTB across the populations. Chronic stressors (e.g., low maternal education and violence) and their correlates were pivotal for PTB prediction only for N-H Black women. Interpretable, race/ethnicity-specific MARS models can predict PTB accurately and explain the most impactful life stressors and their magnitude of effect on PTB risk among N-H Black and N-H White women.