Discovering Causal Models with Optimization: Confounders, Cycles, and Feature Selection

Based on benchmarking various methods and analyzing multiple real scRNA-seq datasets, a computational platform/workflow and a set of tips for best practices are developed for analyzing causal interactions or relationships in single cells.

Discovering and understanding the causal relationships underlying natural phenomena is important for many scientific disciplines, such as economics, computer science, education, medicine and biology. Meanwhile, new knowledge is revealed by discovering causal relationships from data. The causal discovery approach can be characterized as causal structure learning, where variables and their conditional dependencies are represented by a directed acyclic graph. Hence, causal structure discovery methods are necessary for discovering causal relationships from data. In this survey, we review the background knowledge and the causal discovery methods comprehensively. These methods are isolated into four categories, including constraint-based methods, score-based methods, functional causal models based methods and continuous optimization based methods. We mainly focus on the advanced methods which leverage continuous optimization. In addition, we introduce commonly utilized benchmark datasets and open source codes for researchers to evaluate and apply causal discovery methods.KeywordsCausal discoveryCausal structure learningDirected acyclic graphsContinuous optimization

We present an algorithmic framework for learning local causal structure around target variables of interest in the form of direct causes/effects and Markov blankets applicable to very large data sets with relatively small samples. The selected feature sets can be used for causal discovery and classification. The framework (Generalized Local Learning, or GLL) can be instantiated in numerous ways, giving rise to both existing state-of-the-art as well as novel algorithms. The resulting algorithms are sound under well-defined sufficient conditions. In a first set of experiments we evaluate several algorithms derived from this framework in terms of predictivity and feature set parsimony and compare to other local causal discovery methods and to state-of-the-art non-causal feature selection methods using real data. A second set of experimental evaluations compares the algorithms in terms of ability to induce local causal neighborhoods using simulated and resimulated data and examines the relation of predictivity with causal induction performance. Our experiments demonstrate, consistently with causal feature selection theory, that local causal feature selection methods (under broad assumptions encompassing appropriate family of distributions, types of classifiers, and loss functions) exhibit strong feature set parsimony, high predictivity and local causal interpretability. Although non-causal feature selection methods are often used in practice to shed light on causal relationships, we find that they cannot be interpreted causally even when they achieve excellent predictivity. Therefore we conclude that only local causal techniques should be used when insight into causal structure is sought. In a companion paper we examine in depth the behavior of GLL algorithms, provide extensions, and show how local techniques can be used for scalable and accurate global causal graph learning.

Diversity is an essential element of an effective health care system. A key to developing a diverse workforce is establishing a diverse student population in health professions programs. To examine the diversity of students in Doctor of Medicine (MD), Doctor of Osteopathic Medicine (DO), Doctor of Dental Surgery (DDS), Doctor of Dental Medicine (DMD), and Doctor of Pharmacy (PharmD) programs with emphasis on the trends of underrepresented minoritized groups (American Indian or Alaska Native, Black or African American, Hispanic or Latino, and Native Hawaiian or Other Pacific Islander) and sex relative to the overall age-adjusted US population. This cross-sectional study used deidentified, self-reported data from 2003 to 2019 from the Association of American Medical Colleges, American Association of Colleges of Osteopathic Medicine, American Dental Education Association, American Dental Association, and American Association of Colleges of Pharmacy. Data analysis was performed from 2003 to 2004 and from 2018 to 2019. Data on the race, ethnicity, and sex of applicants, matriculants, and degrees conferred by health professions programs were collected and compared with the age-adjusted population in the US Census (aged 20-34 years) over time. The main outcomes were trends in the proportions of underrepresented minoritized groups and sex diversity among applicants, matriculants, and degrees conferred relative to the overall age-adjusted US population. Trends were measured using the representation quotient, which is defined as the ratio of the proportion of each subgroup to the total population of applicants, matriculants, or graduates relative to the proportion for that subgroup within the US Census population of similar age. Regression analysis was used to evaluate the trend over time. A total of 594 352 applicants were analyzed across the examined programs. From 2003 to 2019, the proportions of individuals from underrepresented groups increased for DDS and DMD (applicants, from 1003 of 8176 to 1962 of 11 298 [5.1%]; matriculants, from 510 of 4528 to 966 of 6163 [4.2%]; degrees awarded, from 484 of 4350 to 878 of 6340 [2.7%]), PharmD (applicants, from 9045 of 71 966 to 11 653 of 50 482 [9.0%]; matriculants, from 5979 of 42 627 to 10 129 to 62 504 [6.3%]; degrees awarded, from 922 of 7770 to 2190 of 14 800 [3.0%]), and DO (applicants, from 740 of 6814 to 3478 of 21 090 [5.4%]; degrees awarded, 199 of 2713 to 582 of 6703 [1.4%]) programs, but decreased for MD programs (applicants, from 6066 of 34 791 to 7889 of 52 777 [-2.3%]; matriculants, 2506 of 16 541 to 2952 of 21 622 [-2.4%]; degrees awarded, from 2167 of 15 829 to 2349 of 19 937 [-0.1%]). Compared with age-adjusted US Census data, all programs had more Asian students and fewer male, American Indian or Alaska Native, Black or African American, Hispanic or Latino, and Native Hawaiian or Other Pacific Islander students (representation quotient <1). In this cross-sectional study, most of the health professions in the study saw increases in underrepresented minority applicants, matriculants, and degrees conferred from 2003 to 2019; however, all programs were below the age-adjusted US Census data. The increased racial, ethnic, and sex diversity in the programs illustrates progress, but additional strategies are needed to achieve a more representative health care workforce.

Improved clinical trial race/ethnicity reporting and updated inclusion profile, 2017–2022: A New Jersey snapshot

Public health professionals have long played a vital-albeit underappreciated-role in shaping, not simply using, US Census data, so as to provide the factual evidence required for good governance and health equity. Since its advent in 1790, the US Census has constituted a key political instrument, given the novel mandate of the US Constitution to allocate political representation via a national decennial census. US Census approaches to categorizing and enumerating people and places have profound implications for every branch and level of government and the resources and representation accorded across and within US states. Using a health equity lens to consider how public health has featured in each generation's political battles waged over and with census data, this essay considers three illustrations of public health's engagement with the enduring ramifications of three foundational elements of the US Census: its treatment of slavery, Indigenous populations, and the politics of place. This history underscores how public health has major stakes in the values and vision for governance that produces and uses census data.

Variation in Incidence of ESKD Among Individuals of Native Hawaiian/Pacific Islander Race Based on Data From the US Renal Data System

Introduction: Lipoprotein (a) [Lp(a)] is an emerging target for atherosclerotic cardiovascular disease (ASCVD) with racial/ethnic differences that may link with ASCVD. Compared to non-Hispanic White (NHW) individuals, Asian and Black individuals have high Lp(a) levels. We evaluated diverse racial/ethnic reporting and representation in high impact studies of Lp(a) and ASCVD that inform major Lp(a) scientific statements. Methods: We reviewed the 2022 American Heart Association (AHA) Scientific Statement on Lipoprotein (a) and the 2019 Scientific Statement from the National Lipid Association (NLA) on the Use of Lipoprotein (a) in Clinical Practice for all cited observational studies or trials of Lp(a) and ASCVD. For studies of the same cohort, only the one with the largest sample size was included. Meta-analyses were excluded to avoid duplication. Manuscripts and supplements were systematically reviewed for participant race/ethnicity data. Participation (%) was pooled across studies for overall representation. In a sub-analysis of US studies, pooled representation was compared with 2020 US Census data. Results: A total of 58 studies met inclusion criteria. Only 33 (57%) reported any race/ethnicity. Reporting studies (participant n=701859) had 92% NHW, 2.6% Black, 0.6% Hispanic, 3.2% Asian, and &lt;0.1% American Indian/Alaskan Native or Native Hawaiian/Pacific Islander participants. No Lp(a) studies disaggregated Hispanic subgroups and 5 disaggregated Asian subgroups. In US studies, Hispanic and Asian representation was lower than 2020 US Census data (Fig). Conclusions: High impact studies of Lp(a) and ASCVD underreport and rarely disaggregate participant race/ethnicity and have low overall Black, Hispanic, and Asian representation. Among US studies, Hispanic and Asian groups are underrepresented versus 2020 Census data. Efforts are needed to improve diverse study race/ethnicity reporting and representation for this emerging therapeutic target.

Ethnic Minority Representation in Female Sexual Dysfunction Questionnaire Validation: A Scoping Review

Abortion travel within the United States: An observational study of cross-state movement to obtain abortion care in 2017

Causal discovery is essential across various scientific fields to uncover causal structures within data. Traditional methods relying on observational data have limitations due to confounding variables. This paper presents an optimization-based approach using integer programming (IP) to design minimal intervention sets that ensure causal structure identifiability. Our method provides exact and modular solutions, adaptable to different experimental settings and constraints. We demonstrate its effectiveness through comparative analysis across different settings demonstrating its applicability and robustness.

Ethnoracial and Geographic Representation in United States Glaucoma Clinical Trials (2004-2023).

Racial Disparities in Mortality Trends from Acute Myeloid Leukemia: A Population-Based Study in the United States between 2000 and 2019

Asian Americans and disproportionate exposure to carcinogenic hazardous air pollutants: A national study

The emergence of drug-resistant Streptococcus pneumoniae poses new clinical challenges and may also reflect a change in the epidemiology of S. pneumoniae infections. A variety of studies have shown that drug-resistant S. pneumoniae infections are linked to antimicrobial use. It has been hypothesized that persons of high socioeconomic status are at increased risk for a drug-resistant infection because of greater access to antimicrobial drugs. To assess whether median household income is associated with increased risk of penicillin-nonsusceptible S. pneumoniae infections, the authors geocoded and linked data from population-based surveillance for invasive pneumococcal disease with data from the 1990 US Census. Among invasive pneumococcal isolates from Atlanta, Georgia, in 1994, increasing proportions of penicillin-nonsusceptible isolates were associated with higher median household incomes (chi2 for trend, 15.17; p=0.002). Despite higher rates of invasive pneumococcal disease among blacks and persons who resided within lower median household income areas, white patients in areas with higher median household income had a higher risk of being infected with strains that were not susceptible to penicillin (Wilcoxon rank sum, Z=2.66, p=0.008). These findings demonstrated the utility of geocoding and US Census data in describing the epidemiology of drug-resistant S. pneumoniae and also provided more evidence that socioeconomic factors may influence the development of drug resistance.