71446 research outputs found
Sort by
Institutional Books 1.0: A 242B Token Dataset from Harvard Library's Collections, Refined for Accuracy and Usability
No full textLarge language models (LLMs) use data to learn about the world in order to produce meaningful correlations and predictions. As such, the nature, scale, quality, and diversity of the datasets used to train these models, or to support their work at inference time, have a direct impact on their quality. The rapid development and adoption of LLMs of varying quality has brought into focus the scarcity of publicly available, high-quality training data and revealed an urgent need to ground the stewardship of these datasets in sustainable practices with clear provenance chains. To that end, this technical report introduces Institutional Books 1.0, a large collection of public domain books originally digitized through Harvard Library's participation in the Google Books project, beginning in 2006. Working with Harvard Library, we extracted, analyzed, and processed these volumes into an extensively-documented dataset of historic texts. This analysis covers the entirety of Harvard Library's collection scanned as part of that project, originally spanning 1,075,899 volumes written in over 250 different languages for a total of approximately 250 billion tokens. As part of this initial release, the OCR-extracted text (original and post-processed) as well as the metadata (bibliographic, source, and generated) of the 983,004 volumes, or 242B tokens, identified as being in the public domain have been made available. This report describes this project's goals and methods as well as the results of the analyses we performed, all in service of making this historical collection more accessible and easier for humans and machines alike to filter, read and use.Author's Origina
Metamorphoses of the Past: Post-Soviet Memory in the Twenty-First Century
No full textThis dissertation examines how memories of the Soviet past are addressed and reimagined across three distinct cultural spheres: religion, literature, and film. It engages with the problem of fragmented and incomplete historical memory—a condition that many scholars identify as central to memories of the Soviet past. My dissertation explores how these three cultural domains attempt to respond to this challenge. I argue that disrupted memory demands more creative engagement in order to be shaped into a coherent narrative and to be more closely linked to the identity of its beholder. I also suggest that the creative strategies used to approach this memory must remain flexible and open-ended in order to accommodate its fragmented nature. In the discourse of the Russian Orthodox Church Outside of Russia (ROCOR), these memories take shape through the theological concept of martyrdom, adopting a highly antagonistic mode of memory. In contrast, the literary works of Maria Stepanova and Sergei Lebedev are treated as examples of cosmopolitan memory. These texts reject moral binaries and instead embrace the ethical complexity of the past. Finally, I analyze Il’ia Khrzhanovskii’s Dau project, an experimental cinematic work in which people improvised their lives within a simulated Soviet society. Dau is an example of an agonistic memory that presents the Soviet past as a multifaceted, ethically ambiguous landscape. These three examples of engagement with Soviet memory challenge simplified narratives of the past and reveal the creative potential of memory work.Slavic Languages and Literature
Differentiating on Diversity: How Disclosing Workforce Diversity Influences Consumer Choice
No full textCompanies are facing increased pressure to “walk the talk” on diversity, equity, and inclusion (DEI) in their operations. One specific call-to-action from stakeholders is the public disclosure of EEO-1s. Companies with 100+ employees are federally mandated to annually report the intersectional diversity data of their workforce in the EEO-1. We examine how consumers perceive the strategic decision companies make regarding whether to disclose workforce diversity information. We find no evidence that a company’s disclosure of its workforce diversity data negatively affects attitudes or perceived company commitment to diversity, even when it reveals racial disparities across job categories. Instead, we find that consumers perceive firms that disclose their workforce data more positively and to be more committed to DEI initiatives, relative to firms that choose not to disclose, particularly when these disclosures reveal diversity within the workforce.Author's Origina
Fruits, Vegetables, Plant-based Diets, and Breast Cancer
No full textBreast cancer is the most common type of malignancy among women, and incidence rates have been rising in recent years and are expected to continue increasing. Diet is thought to play a role in breast tumorigenesis; healthy plant foods, including fruits and vegetables, and healthy plant-based diets have been associated with reduced risk of breast cancer. These results appear to vary by hormone receptor status, and there is inconclusive evidence on whether these results differ by molecular subtype. Recent advances in ‘omics technology have allowed for the development of methods to measure levels of metabolites present in food or reflective of biological response to dietary intake. Few studies have explored circulating metabolites, especially unidentified metabolomic features, related to fruit and vegetable consumption and risk of breast cancer.
In Chapter 1, we used plasma metabolite values measured in nested case control studies from the Nurses’ Health Study (NHS) and NHSII to develop a metabolomic signature of fruit and vegetable intake. We examined associations between the metabolomic signatures and dietary fruit and vegetable intake with breast cancer risk. In this study, we identified metabolomic signatures reflecting fruit and vegetable consumption and underlying biological processes that may link diet with breast cancer.
In Chapter 2, we pooled data from 28 cohorts in the Pooling Project of Prospective Studies of Diet and Cancer (DCPP). Associations between fruit and vegetable intakes and breast cancer risk were assessed. Higher fruit intake was inversely associated with all subtypes of breast cancer, with the strongest associations observed for ER-negative and triple-negative breast cancer. Higher vegetable intake was associated with lower rates of ER-negative and triple-negative tumors. Results were generally consistent across studies and strata defined by sociodemographic, medical, lifestyle, and study characteristics. These results provided additional evidence supporting consumption of fruits and vegetables for cancer prevention.
In Chapter 3, we investigated relationships between the healthy plant-based dietary index (hPDI) and unhealthy PDI (uPDI) and breast cancer risk overall and by subtypes among 27 cohorts in the DCPP. The hPDI was significantly associated with lower incidence of overall, ER-positve, ER-negative, HER2-enriched, and triple-negative breast cancer. The uPDI was significantly associated with higher incidence of overall and ER-negative tumors. Weaker and nonsignificant results were observed for the remaining subtypes. These findings suggest that healthful plant-based diets were associated with a slightly lower incidence of breast cancer, particularly ER-negative tumors, and unhealthful plant-based diets were associated with higher incidence of the same subtypes.Population Health Science
Postpartum Metabolic Health: Social Determinants of Health, Weight Retention, and Predictors of LDL-C in Women with Recent Gestational Diabetes Mellitus (GDM)
No full textPaper 1:
Title:
Association of Social Determinants of Health Measured by Social Vulnerability Index (SVI) With Postpartum Weight Retention among Women with Recent Gestational Diabetes Mellitus
Authors: Moomtahina Fatima, MBBS 1,4; Jacinda Nicklas, MD, MPH 2; Geetha Iyer, MBBS, MD, MPH, PHD 3; Ellen W Seely, MD 1,4
Affiliation: 1Harvard Medical School ; 4Division of Endocrinology, Brigham and Women's Hospital, Boston, USA ; 3University of Colorado Medical Center, Aurora, CO, USA ; 4Harvard Pilgrim Health Care Institute
Abstract
Background and Objective:
In women with a history of gestational diabetes mellitus (GDM), postpartum weight retention (PPWR) increases the risk of type 2 diabetes mellitus (T2DM). The association between social determinants of health (SDoH) and PPWR in women with recent GDM remains unclear. We examined the relationship between the Social Vulnerability Index (SVI), a composite measure of community-level disadvantage, and PPWR among women with recent GDM. Additionally, we explored the association between individual components of SVI and PPWR to assess the relative contributions of specific domains of social vulnerability.
Methods:
This secondary analysis of the Balance After Baby Intervention (BABI) trial (2016–2019) used county-level SVI scores from the CDC’s 2020 database, categorized into tertiles (low, moderate, high vulnerability). PPWR at 12 months postpartum was calculated as the difference between pre-pregnancy and postpartum weight. Multivariable linear regression models evaluated the association between SVI tertiles and PPWR, adjusting for key confounders.
Results:
Among 142 participants, women in the second tertile (moderate SVI) had significantly higher PPWR at 12 months postpartum than those in the lowest tertile (adjusted mean difference: 6.71 lbs, 95% CI: 2.07–11.35, p = 0.007), while PPWR was not significantly different for the third tertile. The socioeconomic and racial/ethnic minority status domains of SVI were positively associated with PPWR, while the housing and transportation domain was negatively associated.
Conclusions:
Moderate social vulnerability is associated with increased PPWR in women with prior GDM. These findings highlight the need for targeted interventions in moderately vulnerable populations to reduce PPWR, thereby mitigating risk of T2DM.
Keywords: postpartum weight retention, gestational diabetes mellitus, social vulnerability index, social determinants of health.
Paper 2:
Title: Six-Week Postpartum Predictors of LDL-C Levels at One Year Postpartum in Women with Recent Gestational Diabetes Mellitus
Authors: Moomtahina Fatima, MBBS 1,5; Emily C Tan, BS 2; Christine E Horn, BA 3; Jacinda Nicklas, MD, MPH 4 ; Ellen W Seely, MD 1,5
Affiliation: 1Harvard Medical School ; 5Division of Endocrinology, Brigham and Women's Hospital, Boston, USA ; 2Memorial Sloan Kettering Cancer Center; 3Larner College of Medicine, University of Vermont; 4University of Colorado Medical Center, Aurora, CO, USA.
Abstract
Background: Women with prior gestational diabetes mellitus (GDM) are at an increased risk of developing type 2 diabetes (T2DM). During pregnancy, they exhibit higher levels of LDL, triglycerides (TG), and total cholesterol (TC), and lower HDL levels than those without GDM. However, the postpartum (PP) trajectory of LDL-C remains unclear. Identifying early postpartum predictors of LDL-C at 12 months could help target high-risk women for timely follow-up and intervention to reduce long-term cardiometabolic risk.
Objective: This study aims to identify predictors of LDL-C levels at 12 months PP in women with recent GDM based on clinical measures at six weeks PP, using the best-fitting backward stepwise multivariable linear regression analysis. A secondary objective was to assess whether change in PP weight, continued systemic hormonal contraceptive use, and breastfeeding from six weeks to 12 months PP improved model fit.
Methods: This was a secondary analysis of the Balance After Baby Intervention (BABI) trial, a web-based lifestyle intervention conducted between 2016 and 2019 across three U.S. clinical sites among women with GDM in their most recent pregnancy.
Results: The final analytic sample comprised 122 women. The mean age at screening was 33 ± 5 years. The cohort was racially and ethnically diverse: 33% Hispanic, 49% White, 19% Black/African American, 19% Asian, and 13% other races/ethnicities. Regarding socioeconomic status, 55% were college graduates, and 81% had an annual household income ≥ $35,000. Stepwise backward multivariable linear regression analysis revealed LDL-C at six weeks postpartum (β = 0.66, 95% CI: 0.57–0.76, p 0.001) and breastfeeding at six weeks postpartum (β = -9.13, 95% CI: -17.34, -0.93, p = 0.029) were significant predictors of LDL-C levels at 12 months postpartum. In an extended model incorporating weight change from six weeks to 12 months postpartum, continued systemic hormonal contraceptive use (β = 14.67, 95% CI: 3.26–26.08, p = 0.012) and maternal age (β = 0.82, 95% CI: 0.05–1.60, p = 0.036) were also significantly associated with LDL-C levels at 12 months. The final model (Model 4) demonstrated improved fit (Adjusted R² = 0.633, AIC = 884.318).
Conclusion: LDL-C levels at six weeks postpartum is a significantly predictor of LDL-C levels at 12 months in women with recent GDM. Early postpartum breastfeeding appears protective against elevated LDL-C, whereas continued systemic hormonal contraceptive use is associated with higher LDL-C levels. Additionally, older maternal age was linked to higher LDL-C at 12 months postpartum. Given that the ADA and ACOG recommend an OGTT at 4–12 weeks postpartum for women with GDM, this routine visit presents a potential opportunity to concurrently perform lipid screening. Identifying women at risk of elevated LDL-C at 12 months postpartum through early postpartum lipid screening enables the implementation of targeted interventions aimed at reducing long-term cardiovascular risk.
Keywords: Gestational diabetes mellitus, postpartum dyslipidemia, LDL cholesterol, breastfeeding, hormonal contraception, cardiovascular riskGraduate Educatio
LEG-LENGTH DISCREPANCY IN LOWER-EXTREMITY VASCULAR MALFORMATIONS: PREVALENCE, RISK FACTORS, AND HEMATOLOGIC COMPLICATIONS FOLLOWING EPIPHYSIODESIS
No full textVascular malformations (VaMs) are congenital anomalies of the vascular system that arise from errors in vascular development. They can be categorized by flow characteristics into low-flow (capillary, venous, or lymphatic) or high-flow (arterio-venous) lesions. A significant concern in lower-extremity VaMs is the development of major leg-length discrepancy (LLD), defined as a ≥2 cm difference between limbs, which can impair gait and joint health. Epiphysiodesis, which arrests growth in the longer limb, is used to address major LLD in children. However, dysplastic vessels and coagulopathies often complicate surgical outcomes in these cases.
A. Manuscript 1 Context
“Prevalence and Risk Factor Analysis of Leg-Length Discrepancy in Patients with a Capillary Malformation on the Lower Extremities”
Capillary malformations (CMs), a subset of low-flow VaMs, frequently lead to progressive LLD. This manuscript quantifies the prevalence of major LLD (≥2 cm) among pediatric patients with lower-extremity CM and identifies risk factors associated with severe discrepancies. This study supports early detection and optimal timing for orthopedic intervention by pinpointing which children are at greatest risk.
B. Manuscript 2 Context
“Hematologic Complications After Epiphysiodesis for Vascular Malformation–Related Leg-Length Discrepancy”
Broadening the scope to all VaM types, this manuscript examines the incidence of venous thromboembolism (VTE) and bleeding events following epiphysiodesis. By analyzing potential perioperative risk factors, it provides insights into how best to safeguard these patients, given their susceptibility to coagulopathic complications and surgical risks.Medical Scienc
Between Brothers and Beyond: The Cultural Evolution of the Naxi in the Wake of Polyandry’s Decline
No full textWhy has polyandry, where multiple men, typically brothers, share a single wife,
persisted for centuries among the Naxi of Eya, a remote community in southern Sichuan,
and why is it now declining? This thesis examines Naxi polyandry as a culturally
transmitted adaptive strategy, shaped by ecological scarcity, kinship-based economic
logics, and shifting pathways of cultural transmission. While polyandry has been
documented across various Himalayan societies, the Naxi case differs significantly from
the better-known models of their neighbors, the Tibetans and the Mosuo. Drawing on
cultural evolutionary theory, household-centered analysis, and gender dynamics, this
study investigates how this marriage system emerged, stabilized, and is now transforming
in response to both external pressures and internal negotiations.
Based on original ethnographic fieldwork in twelve households, including semi-
structured interviews, spatial and ritual observations, and social media analysis, this
research documents how polyandrous families achieve significant advantages over
monogamous ones. They control larger landholdings, coordinate labor more effectively,
and preserve estate continuity. These arrangements are not only materially efficient but
also reproductively advantageous at the household level: polyandrous families average
more children than monogamous ones, even as individual men often experience reduced
reproductive outcomes. Women, by contrast, frequently gain modest advantages in
reproductive success and household influence, benefiting from redistributed paternal
investment and cooperative household structures. These findings position polyandry as a
stable strategy that privileges collective welfare over individual maximization.
Yet material advantages alone do not explain its persistence. Polyandry's viability
has historically relied on integration into a broader cultural ecology that emerged through
success-biased and prestige-biased transmission mechanisms, including permissive
attitudes toward sexual flexibility, influence from neighboring Tibetan fraternal systems,
and selective absorption of Confucian ideals that stress lineage continuity without
mandating monogamy. These culturally transmitted features enabled the acceptance and
normalization of polyandrous arrangements. Naxi households developed institutional
mechanisms to manage internal tensions, such as strategic ambiguity around paternity,
conjugal rotation embedded in architectural design, and Dongba ritual legitimation of
marital forms.
Today, these cultural foundations are weakening. As younger Naxi pursue
education, digital entrepreneurship, and urban employment, they increasingly turn to Han
Chinese models of autonomy and monogamy. While many still acknowledge polyandry's
benefits, fewer choose it. Emerging hybrid arrangements, such as sequential polyandry,
where brothers join the marriage over time, and distributed households, reflect active
adaptation rather than cultural rupture.
This thesis argues that Naxi polyandry is neither a cultural oddity nor an
inexplicable vestige, but a contextually rational and culturally legitimated strategy. Its
transformation exemplifies how marriage systems evolve through adaptive negotiation,
balancing inherited logics with shifting aspirations.Extension Studie
“Give me the Scotsman Every Time:” Exploring Scottish National Identity Through World War I Prisoner of War Experiences
No full textThis research examines the Scottish experience in World War I, emphasizing the ways in which the Great War experience magnified questions of identity, including how Scottish soldiers viewed themselves and how they were viewed by others, particularly their German adversaries. Among the research questions explored are: did captured Scottish soldiers identify as primarily Scottish or British or both? Were they treated differently than their British counterparts as a result of their Scottishness? If so, how? Did the war impact feelings of either Scottish differentiation or kinship, or both, as part of their greater British identity? If so, how?
This thesis further highlights the under-researched experience of Scottish prisoners of war (POWs) who were held in German prison camps and explores how their time in captivity affected questions of national identity. Ultimately, the Scottish experience in WWI was largely comparable to that of other Britons; however, patterns of Scottish POWs being singled out for humiliation, harsher treatment, and, paradoxically, curiosity and recruitment did exist and suggest a reinforcement of Scottish, rather than British, identity.Extension Studie
Political Economy of Primary Health Care: A Comparison of Health System Reforms
No full textThis doctoral thesis examines the political economy of primary health care (PHC) reforms across nine countries—the Democratic Republic of Congo, Dominica, Egypt, Kazakhstan, Kenya, New Zealand, Thailand, Tunisia, and Uruguay—which represent diverse political, economic, and health system contexts. Addressing a gap in the literature where the political economy of PHC remains underexplored, the study investigates: (1) How have political economy factors driven PHC reforms? and (2) What are the distinct political economy factors of PHC reforms, and how do they differ across different types of PHC reforms—those focusing on financing versus organization?
Employing historical institutionalism alongside the Control Knob Framework, this thesis analyzes how political economy factors interact to shape reform trajectories. The research draws on primary qualitative data from 324 respondents via interviews, focus groups, and expert consultations across the sampled countries, complemented by an extensive review of literature and policy documents. Using the Framework Method, both deductive analysis for theory application and inductive approaches for theory building were applied. Process tracing provides an in-depth analyses of each reform, while the Comparative Sequential Method examines reform trajectories across health systems.
The findings reveal that political economy dynamics operate at multiple interrelated levels. Temporality is critical—historical legacies and path dependencies can constrain and enable reforms, while political upheavals or economic crises create windows for transformative change. Institutional contexts—political systems, electoral rules, constitutional provisions, and bureaucracies—significantly influence reform outcomes. Interest groups such as physicians, civil society, donors, and citizens actively shape reforms, while the politics of ideas affect public support and sustainability.
This research identifies six distinct domains of political economy tensions in PHC reforms: (1) public versus private sector, (2) hospital versus primary care, (3) physicians versus non-physicians, (4) centralized versus decentralized administration, (5) expanding versus limiting citizens’ choice, and (6) selective versus comprehensive changes. Moreover, the chosen reform entry point—whether financing or organization—shapes how these tensions unfold. Understanding these dynamics can help policy actors anticipate resistance and devise strategies to navigate reforms. To build resilient health systems, countries must strengthen local capacities for political economy analysis and engage diverse stakeholders.Public Healt
Care Delivery in Nursing Homes: The Role of Quality, Insurance, and Specialty Clinicians
No full textAbstract
This dissertation describes three papers evaluating factors which contribute to the care and wellbeing of nursing home residents. Taken together, the chapters will provide important evidence on how the quality of care experienced in U.S. nursing homes is influenced by a multitude of factors – including the prevalence of antipsychotic use, availability of insurance plans with targeted benefits, and role of the admitting clinician in a nursing home. This work informs how policy research can improve the care experience and outcomes among post-acute care short-stay and long-stay nursing home residents.
Chapter 1 (with David Grabowski) quantifies quality of care following an admission to a nursing home with low or high antipsychotic drug use. We use an instrumental variable approach to estimate the effect of facility-level antipsychotic use on patient outcomes for both post-acute care short-stay and long-stay residents. Our instrument, differential distance, measures how much farther an individual would have to travel to avoid a high-use facility. Receiving care from a low-use facility reduces the diagnosis of schizophrenia, use of restraints, and hospitalizations among both short-stay and long-stay residents. We also observe larger reductions among residents with dementia. Given concerns about the misuse of antipsychotics in U.S. nursing homes, these findings highlight the need for continued monitoring and intervention by policymakers.
Chapter 2 (with David Grabowski, Mary Beth Landrum, and J. Michael McWilliams) studies the growth of provider-led Institutional Special Needs Plans (I-SNPs), nursing homes which engage in financial partnerships with an insurer to operate a plan for their own residents. We use a stacked difference-in-differences design to estimate the effect of offering a provider-led I-SNP on enrollment and quality of care. Offering a provider-led I-SNP led to a 17.0 percentage point increase in I-SNP enrollment among facility residents within four years relative to control nursing homes. We also estimate that offering provider-led I-SNPs significantly decreased hospitalizations and increased the use of hypnotic drugs and reporting of pressure ulcers. These results suggest that risk bearing by nursing homes in the form of provider-led I-SNPs may successfully reduce utilization, but with unclear implications for quality and competition.
Chapter 3 (with J. Michael McWilliams) assesses how patients are distributed within skilled nursing facilities (SNFs) to clinicians who specialize in nursing facility care (SNFists) to understand the influence of clinician factors on patient outcomes. Using multi-level modeling, we find that the variation in patient characteristics across admitting clinicians is substantially less among patients admitted to SNFists vs. non-SNFists, consistent with a less selective distribution of patients to clinicians within a facility when the admitting clinician is a SNFist. We find minimal variation in the rates of mortality, hospitalization, and antipsychotics across SNFists and outcomes did not differ between advanced practice provider and physician SNFists. Our results suggest that the admitting clinician appears to have little influence on key outcomes in the post-acute setting, in contrast with similar research conducted in other care settings.Health Polic