AMERICAN JOURNAL OF EPIDEMIOLOGY 是历史最悠久，也是主要的流行病学期刊之一，致力于发表流行病学研究领域的实证研究结果、观点文章和方法论发展。使用流行病学数据的人，包括公共卫生工作者和临床医生。

The American Journal of Epidemiology is one of the oldest and most prominent journals in the field of epidemiology, dedicated to publishing empirical research findings, opinion articles, and methodological developments in epidemiology. Its readership includes public health professionals and clinicians who utilize epidemiological data.

《American Journal of Epidemiology》是历史最悠久且最重要的流行病学期刊之一，专注于发表流行病学领域的实证研究结果、观点文章和方法论的发展。该期刊的读者包括公共卫生工作者和临床医生，他们利用流行病学数据进行研究和实践。

American Journal of Epidemiology

Abstract Machine learning is a branch of computer science that has the potential to transform epidemiologic sciences. Amid a growing focus on “Big Data,” it offers epidemiologists new tools to tackle problems for which classical methods are not well-suited. In order to critically evaluate the value of integrating machine learning algorithms and existing methods, however, it is essential to address language and technical barriers between the two fields that can make it difficult for epidemiologists to read and assess machine learning studies. Here, we provide an overview of the concepts and terminology used in machine learning literature, which encompasses a diverse set of tools with goals ranging from prediction to classification to clustering. We provide a brief introduction to 5 common machine learning algorithms and 4 ensemble-based approaches. We then summarize epidemiologic applications of machine learning techniques in the published literature. We recommend approaches to incorporate machine learning in epidemiologic research and discuss opportunities and challenges for integrating machine learning and existing epidemiologic research methods.

Qifang Bi

Katherine E Goodman

Joshua Kaminsky

Justin Lessler

What is Machine Learning? A Primer for the Epidemiologist

什么是机器学习？流行病学家的入门指南

The authors examined the effects of cognitive function, as assessed by the Mini-Mental State Examination, and drug use on the incidence of hip fracture in a community-based Swedish population of 1,608 subjects who were aged ≥75 years on October 1,1987, and who had not had a hip fracture. During the 7,123.8 person-year follow-up, 134 first hip fractures were identified. The Cox proportional hazards model was used to estimate the relative risk of developing hip fracture, taking into account several potential confounders. Compared with those without cognitive impairment, subjects with mild impairment (Mini-Mental State Examination scores 18–23) had a relative risk of 2.04 (95% confidence interval (Cl) 1.29–3.24), and subjects with moderate-severe impairment (Mini-Mental State Examination score <18) had a relative risk of 2.09 (95% Cl 1.17–3.72). Subjects using opioid analgesics (97% took propoxyphene) had a relative risk of 2.01 (95% Cl 1.19–3.40). Taking potassium supplements (99% took potassium chloride) was related to a reduced risk of hip fracture (relative risk = 0.55, 95% Cl 0.31–0.98), while diuretics did not have an independent impact. In summary, the results show that cognitive impairment and use of propoxyphene are associated with increased risk of hip fracture. The observed protection of potassium chloride merits further attention. The limitation of the study was that the assessment of drug use was made only at baseline. Am J Epidemiol 1998; 148:887–92.

Z. Guo

P. Wills

M. Viitanen

J. Fastbom

B. Winblad

Cognitive Impairment, Drug Use, and the Risk of Hip Fracture in Persons over 75 Years Old: A Community-based Prospective Study

认知障碍、药物使用与75岁以上人群髋部骨折风险：一项基于社区的前瞻性研究

This study examined prepregnancy cardiometabolic risk factors and gestational diabetes mellitus (GDM) in subsequent pregnancies. The authors selected 1,164 women without diabetes before pregnancy who delivered 1,809 livebirths between 5 consecutive examinations from 1985 to 2006 in the Coronary Artery Risk Development in Young Adults (CARDIA) Study. The authors measured prepregnancy cardiometabolic risk factors and performed multivariate repeated-measures logistic regression to compute the odds of GDM adjusted for race, age, parity, birth order, and other covariates. Impaired fasting glucose (100-125 vs. <90 mg/dL), elevated fasting insulin (>15-20 and >20 vs. <10 μU/mL), and low levels of high-density lipoprotein cholesterol (<40 vs. >50 mg/dL) before pregnancy were directly associated with GDM: The odds ratios = 4.74 (95% confidence interval (CI): 2.14, 10.51) for fasting glucose, 2.19 (95% CI: 1.15, 4.17) for middle insulin levels and 2.36 (95% CI: 1.20, 4.63) for highest insulin levels, and 3.07 (95% CI: 1.62, 5.84) for low levels of high-density lipoprotein cholesterol among women with a negative family history of diabetes; all P < 0.01. Among overweight women, 26.7% with 1 or more cardiometabolic risk factors developed GDM versus 7.4% with none. Metabolic impairment exists before GDM pregnancy in nondiabetic women. Interconceptual metabolic screening could be included in routine health assessments to identify high-risk women for GDM in a subsequent pregnancy and to potentially minimize fetal exposure to metabolic abnormalities that program future disease.

E. P. Gunderson

C. P. Quesenberry

D. R. Jacobs

J. Feng

C. E. Lewis

S. Sidney

Longitudinal Study of Prepregnancy Cardiometabolic Risk Factors and Subsequent Risk of Gestational Diabetes Mellitus: The CARDIA Study

孕前心脏代谢危险因素与随后妊娠糖尿病风险的纵向研究：CARDIA研究

The method of inverse probability weighting (henceforth, weighting) can be used to adjust for measured confounding and selection bias under the four assumptions of consistency, exchangeability, positivity, and no misspecification of the model used to estimate weights. In recent years, several published estimates of the effect of time-varying exposures have been based on weighted estimation of the parameters of marginal structural models because, unlike standard statistical methods, weighting can appropriately adjust for measured time-varying confounders affected by prior exposure. As an example, the authors describe the last three assumptions using the change in viral load due to initiation of antiretroviral therapy among 918 human immunodeficiency virus-infected US men and women followed for a median of 5.8 years between 1996 and 2005. The authors describe possible tradeoffs that an epidemiologist may encounter when attempting to make inferences. For instance, a tradeoff between bias and precision is illustrated as a function of the extent to which confounding is controlled. Weight truncation is presented as an informal and easily implemented method to deal with these tradeoffs. Inverse probability weighting provides a powerful methodological tool that may uncover causal effects of exposures that are otherwise obscured. However, as with all methods, diagnostics and sensitivity analyses are essential for proper use.

S. R. Cole

M. A. Hernan

Constructing Inverse Probability Weights for Marginal Structural Models

构建边际结构模型的逆概率权重

The pathogenicity of norovirus is definitively established. However, norovirus is frequently detected in the stool of healthy individuals. To gain understanding of the apparent high prevalence of asymptomatic infection, we analyzed a dynamic transmission model of norovirus infection, disease, and immunity. We simulated norovirus epidemiology in low- and high-transmission settings by varying the basic reproduction number (R0). We predicted annual disease incidence values in children aged 0-4 years of 25% with a low R0 and 29% with a high R0. However, the point prevalence of asymptomatic infection rose sharply from 3% to 48% from the low to high R0 settings. Among older children and adults, the models projected that incidence of disease would rise from 6% to 16% from the low to high R0 settings, whereas asymptomatic infection prevalence was lower in this age group. Asymptomatic prevalence of norovirus can change dramatically with small changes in R0. The ratio of prevalence in cases to controls could be high in a developed country and close to or even less than 1 in a high-exposure setting, despite similar disease incidence. These findings highlight an important limitation of case-control studies for pathogens for which there is suboptimal diagnostic specificity.

B. Lopman

K. Simmons

M. Gambhir

J. Vinje

U. Parashar

Epidemiologic Implications of Asymptomatic Reinfection: A Mathematical Modeling Study of Norovirus

无症状再感染的流行病学意义：诺如病毒的数学建模研究

Multimorbidity data is typically analysed by tallying disease counts, which overlooks nuanced relationships among conditions. We identified clusters of multimorbidity and subpopulations with varying risks and examined their association with all-cause mortality using a data-driven approach. We analysed 8-year follow-up data of people ≥35 years who were part of the CRONICAS Cohort Study, a multisite cohort from Peru. First, we used Partitioning Around Medoids and multidimensional scaling to identify multimorbidity clusters. We then estimated the association between multimorbidity clusters and all-cause mortality. Second, we identified subpopulations using finite mixture modelling. Our analysis revealed three clusters of chronic conditions: respiratory (cluster 1: bronchitis, COPD and asthma), lifestyle, hypertension, depression and diabetes (cluster 2), and circulatory (cluster 3: heart disease, stroke and peripheral artery disease). While only the cluster comprising circulatory diseases showed a significant association with all-cause mortality in the overall population, we identified two latent subpopulations (named I and II) exhibiting differential mortality risks associated with specific multimorbidity clusters. These findings underscore the importance of considering multimorbidity clusters and sociodemographic characteristics in understanding mortality risks. They also highlight the need for tailored interventions to address the unique needs of different subpopulations living with multimorbidity to reduce mortality risks effectively.

Juan Carlos Bazo-Alvarez

Darwin Del Castillo

Luis Piza

Antonio Bernabé-Ortiz

Rodrigo M Carrillo-Larco

Liam Smeeth

Robert H Gilman

William Checkley

J Jaime Miranda

Multimorbidity patterns, sociodemographic characteristics, and mortality: Data science insights from low-resource settings

多病共存模式、社会人口学特征与死亡率：来自低资源环境的数据科学洞察

The Green Heart Project is a controlled, community-based clinical trial to evaluate the effects of increasing greenery on community health. The study was initiated in 2018 in a low-to-middle-income residential area of nearly 30,000 racially diverse residents in Louisville, KY. Community engagement was maintained throughout the project, with feedback integrated into its design and implementation. Based on land use, demographics, and greenness in the area, we designated 8 paired clusters of demographically- and environmentally matched "target" (T) and adjacent "control" (C) areas (total of 16 clusters). Levels of ultrafine particles, ozone, and nitrogen oxides were measured in each cluster. In-person exams were conducted for 735 participants in Wave 1 (2018-2019) and 545 participants in Wave 2 (2021). Blood, urine, nail, and hair samples were collected to evaluate cardiovascular risk, inflammation, stress, and pollutant exposure. Demographic and psychosocial data were collected as well. Cardiovascular function was assessed by measuring arterial stiffness and flow-mediated dilation. After Wave 2, more than 8,000 mature, mostly evergreen, trees and shrubs were planted in the T clusters. Post planting data were collected during Wave 3 (2022) from 561 participants. We plan to follow changes in area characteristics and participant health to evaluate the long-term impact of the greening intervention.

Aruni Bhatnagar

Rachel Keith

Ray Yeager

Daniel Riggs

Clara Sears

Brent Bucknum

Ted Smith

Daniel Fleischer

Chris Chandler

Kandi L Walker

Joy L Hart

Sanjay Srivastava

Jay Turner

Shesh Rai

The Green Heart Project: Objectives, Design, and Methods

绿色心脏项目：目标、设计与方法

Persistent racial and ethnic disparities exist in severe maternal and neonatal morbidity, which may be due in part to differences in labor and delivery unit practices across hospitals. We used data collected from 184 hospitals in California (2015-2018) to assess whether nulliparous individuals with low-risk pregnancies differ by race and ethnicity in giving birth at hospitals that tend to use lower-interventional labor and delivery unit practices, and whether such differences contribute to disparities in severe maternal and neonatal morbidity. We classified labor and delivery units as higher- or lower-interventional based on a latent class analysis of survey responses about the frequency of using lower-interventional practices. We used a modified doubly robust g-estimator to estimate counterfactual disparity measures, setting all hospitals to be lower-interventional. Among 348,990 low-risk livebirths, the proportion occurring at lower-interventional hospitals was lowest in Black and Latino individuals (17% and 16%, respectively) and highest in American Indian and Alaska Native (AI/AN) and White individuals (29% in both). Severe maternal and neonatal morbidity occurred most frequently among AI/AN individuals. Counterfactual disparity measures suggested that if all births occurred at lower-interventional hospitals, racial and ethnic disparities in the outcomes would modestly increase, except for severe neonatal morbidity among AI/AN individuals.

Stephanie A Leonard

Xiao Xu

Shantay Davies-Balch

Elliott K Main

Brian T Bateman

David H Rehkopf

Henry C Lee

Jessica Illuzzi

Irogue Igbinosa

Ijeoma Iwekaogwu

Deirdre J Lyell

Labor and delivery unit practices and racial and ethnic disparities in severe maternal and neonatal morbidity among nulliparous individuals with low-risk pregnancies

低风险初产妇中分娩单元实践与严重孕产妇和新生儿发病率的种族和民族差异

Although the public health field has increasingly studied the collateral consequences of incarceration, we know little about the health consequences of other forms of criminal legal contact, including probation and parole. Understanding spatial and racial-ethnic variation in probation/parole across US states provides new insights into how community supervision impacts population health disparities. However, state-level probation/parole prevalence has not been adequately described. Using data from the Bureau of Justice Statistics and the Census for the years 2001 to 2018, we provide the first state-level estimates of probation and parole populations by race over time in the US. We find large variation in disparities across states and time that is masked by national-level estimates. The US probation population decreased, and its racial composition remained steady between 2001 and 2018. However, in all but five states, the Black-White gap in probation rates declined. The Black-White gap in parole rates declined in all but seven states. The extent to which these race-specific changes in probation or parole over time reflect adjudication processes favoring White people, and/or affect population health, warrant further investigation.

David Rigby

Daichi Hibi

Ruth Wygle

Hedwig Lee

Joan Casey

Alison Gemmill

Tim Bruckner

State-level changes in racial disparities in probation and parole rates in the United States, 2001–2018

美国州级缓刑和假释率中种族差异的变化，2001-2018年

From 2001-2021, the age- and sex-adjusted Veteran suicide rate increased 76.3%. Surveillance of suicidal ideation (SI) and non-fatal suicidal self-directed violence (NF-SSDV) is a critical component of public health-oriented suicide prevention efforts. To facilitate national NF-SSDV surveillance, a biennial, population-based survey was initiated: Assessing Social and Community Environments with National Data (ASCEND) for Veteran Suicide Prevention. 17,396 Veterans participated in the first large-scale ASCEND survey (2022). This manuscript reports on SI and NF-SSDV prevalence among Veterans residing in the continental U.S., Hawaii, and Puerto Rico. Lifetime SI was reported by 31.98% (95%CI=30.97-32.99), post-military SI by 25.88% (95%CI=24.91-26.85) and past-year SI by 12.69% (95%CI=11.90-13.47). The most commonly considered SI method among those with past-year SI was gunshot. Additionally, 34.42% (95%CI=33.07-35.78) of Veterans with lifetime SI reported lifetime preparatory behaviors. Moreover, 9.13% (95%CI=8.43-9.82) of Veterans reported lifetime interrupted attempts. Lifetime suicide attempts (SA) were reported by 6.99% (95%CI=6.41-7.56) of Veterans, with 4.88% (95%CI=4.39-5.36) reporting post-military SA. The most common method in prior attempts was medication overdose. ASCEND provides a novel opportunity to elucidate the prevalence of SI and different types of NF-SSDV in the Veteran population. Recurring administration will elucidate changes in SI and NF-SSDV prevalence in the Veteran population over time.

Claire A Hoffmire

Sean M Barnes

Ryan Holliday

Julie A Kittel

Alexandra L Schneider

Lisa A Brenner

Jamie L Tock

Lindsey L Monteith

Non-Fatal Suicidal Self-Directed Violence among U.S. Veterans (2022): The Assessing Social and Community Environments with National Data (ASCEND) for Veteran Suicide Prevention Study

美国退伍军人非致命性自杀自伤行为（2022年）：利用国家数据评估社会与社区环境以预防退伍军人自杀研究（ASCEND）

Abbreviation	Am. J. Epidemiol.
Journal Impact	4.85
Quartiles(Global)	PUBLIC, ENVIRONMENTAL & OCCUPATIONAL HEALTH(Q1)

ISSN	0002-9262, 1476-6256
h-index	288

Publication Information	Publisher: Oxford University Press，Publishing cycle: Semimonthly，Journal Type: journal，Open Access Journals: No
Basic data	Year of publication: 1921，Proportion of original research papers: 97.10%，Self Citation Rate:2.00%， Gold OA Rate: 30.46%
Average review cycle	网友分享经验：一般,3-6周
Average recruitment ratio	网友分享经验：很难

American Journal of Epidemiology

Journal Citation Format

A journal article with 1 author

A journal article with 2 authors

A journal article with 3 authors

A journal article with 5 or more authors

Books Citation Format

Thesis Citation Format

Web sites Citation Format

Patent Citation Format

Staying up late manually editing references? ivySCI automatically matches journals and helps you generate references with a single click.

Share Submission Experience