PUBBLICAZIONE/PUBLICATION

Permanent URI for this communityhttps://gendermore.unimore.it/handle/123456789/1575

Browse

Settings

Start date: 2022Subject: search.filters.subject.Machine Learning

Search Results

Now showing 1 - 4 of 4
  • Item type:Publication,
    Authorship and citation trends in dementia research: A path to equitable career development for scientists Open PDF 
    (Wiley, 2026-06)
    Viswanathan, Jayalakshmi
    ;
    Karamacoska, Diana
    ;
    Thakur, Lokendra S.
    ;
    Sawan, Mouna
    ;
    Gomez‐Arboledas, Angela
    Abstract Introduction: Women are increasingly entering the dementia research workforce, but they frequently fail to attain senior leadership positions in academia. Discrepancies in academic research outputs were investigated to guide equity-focused policy recommendations. Methods: Bibliometrics was conducted on 400,482 original research articles (2003-2022) followed by surveying the AD/ADRD research community, to quantitatively and qualitatively evaluate discrepancies in publication and authorship trends. In addition, an unsupervised learning algorithm was developed to analyze underlying relationships between author groups. Results: Male author gender was consistently predictive of better publication and citation metrics across all bibliometrics outcomes, with senior male authors outperforming all other subgroups and fewer senior female authors in most research areas. Female survey respondents reported more barriers, but comparable productivity and citation strategies as male authors. Discussion: Multi-pronged approaches by stakeholders across institutions, funders, and journals are necessary to provide career support for early-career scientists working to transition to senior research roles.
  • Item type:Publication,
    Bias in Student Evaluations of Surgical Attendings: Role of Gender, Race, Age and Experience Open PDF 
    (Elsevier BV, 2026-06)
    Theis, Claudia
    ;
    Jacob, Anusha S.
    ;
    Kapadia, Muneera Rehana
    ;
    Pascarella, Luigi
    Introduction Student evaluations influence faculty promotion but may reflect implicit bias. We assessed whether surgeon gender, race, age, and experience were associated with differences in student evaluations at a large academic medical center. Materials and methods This retrospective cohort study included 149 surgical attendings evaluated by medical students at the University of North Carolina between 2016 and 2020. Quantitative evaluation items were rated on 5-point Likert scales and analyzed using surgeon-level summary comparisons (Wilcoxon rank-sum and Kruskal–Wallis tests), evaluation-level multivariable logistic regression, and generalized estimating equations (GEEs) to account for clustering of multiple evaluations per attending. Qualitative free-text comments were analyzed using a validated natural language processing framework and summarized on a 5-point sentiment scale. Results A total of 149 surgical attendings were evaluated, including 39 women (26.2%) and 110 men (73.8%). The racial and ethnic distribution was 102 White (68.5%), 28 Asian (18.8%), 13 Black (8.7%), and 6 Latino (4.0%). In total, 2475 quantitative evaluations were analyzed, of which 1542 (62.3%) included narrative comments. The median time in practice was 14 y (Q1-Q3: 9-22), and the median time at the University of North Carolina was 11 y (Q1-Q3: 7-19.5). Median composite quantitative evaluation scores were lower for women than for men (4.42 [4.32-4.61] versus 4.61 [4.40-4.82] on a 5-point scale; P = 0.002). In GEE models accounting for clustering of evaluations within attendings and adjusting for age, race/ethnicity, and years in practice, women remained independently associated with lower evaluation scores (β = −0.206; 95% confidence interval: −0.294 to −0.118; P < 0.001). Composite evaluation scores did not differ significantly across racial or ethnic groups (Kruskal–Wallis P = 0.40), and race was not a significant predictor in adjusted models (GEE P = 0.13). Attendings aged ≥50 y had lower unadjusted evaluation scores than younger colleagues (P = 0.03); however, age was not independently associated with evaluation scores after clustering adjustment (GEE P = 0.14). Qualitative sentiment scores were uniformly high (median 4.0 [Q1-Q3: 4.0-4.5]) and did not differ by gender, race, or age. Conclusions Women surgeons received lower quantitative scores than their male colleagues, despite similarly positive qualitative feedback. Age differences were observed in unadjusted analyses but were attenuated after accounting for clustering, whereas no significant differences were observed by race or years in practice. Although absolute score differences were modest, these findings suggest that commonly used student evaluation metrics may reflect gender-related bias and should be interpreted cautiously in academic surgical assessment.
  • Item type:Publication,
    Predicting the future impact of Computer Science researchers: Is there a gender bias? Open PDF 
    (2022)
    Kuppler, M.
    The advent of large-scale bibliographic databases and powerful prediction algorithms led to calls for data-driven approaches for targeting scarce funds at researchers with high predicted future scientific impact. The potential side-effects and fairness implications of such approaches are unknown, however. Using a large-scale bibliographic data set of N = 111,156 Computer Science researchers active from 1993 to 2016, I build and evaluate a realistic scientific impact prediction model. Given the persistent under-representation of women in Computer Science, the model is audited for disparate impact based on gender. Random forests and Gradient Boosting Machines are used to predict researchers’ h-index in 2010 from their bibliographic profiles in 2005. Based on model predictions, it is determined whether the researcher will become a high-performer with an h-index in the top-25% of the discipline-specific h-index distribution. The models predict the future h-index with an accuracy of R2= 0.875 and correctly classify 91.0% of researchers as high-performers and low-performers. Overall accuracy does not vary strongly across researcher gender. Nevertheless, there is indication of disparate impact against women. The models under-estimate the true h-index of female researchers more strongly than the h-index of male researchers. Further, women are 8.6% less likely to be predicted to become high-performers than men. In practice, hiring, tenure, and funding decisions that are based on model predictions risk to perpetuate the under-representation of women in Computer Science. © 2022, The Author(s).
  • Item type:Publication,
    A machine learning approach to quantify gender bias in collaboration practices of mathematicians Open PDF 
    (2023)
    Steinfeldt, C
    ;
    Mihaljevic, H
    Collaboration practices have been shown to be crucial determinants of scientific careers. We examine the effect of gender on coauthorship-based collaboration in mathematics, a discipline in which women continue to be underrepresented, especially in higher academic positions. We focus on two key aspects of scientific collaboration-the number of different coauthors and the number of single authorships. A higher number of coauthors has a positive effect on, e.g., the number of citations and productivity, while single authorships, for example, serve as evidence of scientific maturity and help to send a clear signal of one's proficiency to the community. Using machine learning-based methods, we show that collaboration networks of female mathematicians are slightly larger than those of their male colleagues when potential confounders such as seniority or total number of publications are controlled, while they author significantly fewer papers on their own. This confirms previous descriptive explorations and provides more precise models for the role of gender in collaboration in mathematics.