Turtle Mountain Community College (TMCC) is a private tribal land-grant community college in Belcourt, North Dakota. It is located ten miles from the Canada–US border in Turtle Mountain, the north central portion of North Dakota. In 2012, TMCC's enrollment was 630 full- and part-time certificate and degree-seeking students.
The BioDIGS project is a nationwide initiative involving students, researchers and educators across more than 40 research and teaching institutions. Participants lead sample collection, computational analysis and results interpretation to understand the relationships between the soil microbiome, environment and health.
Data science skills are essential for advancing knowledge of health inequities in marginalized and medically underserved populations. As of 2021 there were no programs in data science in the Tribal Colleges and Universities in North Dakota and the neighboring states. The Virtual Indigenous Data Science Academy was established in 2021 by the University of North Dakota School of Medicine and Health Sciences in collaboration with three North Dakota Tribal Colleges to introduce Indigenous students to the core elements of data science. The Virtual Indigenous Data Science Academy teaching faculty designed, developed, delivered, and evaluated a culturally-informed online synchronous summer program consisting of ten learning modules. The Virtual Indigenous Data Science Academy was delivered synchronously over three consecutive summers (2022-2024) to a total of 48 Tribal college students with diverse backgrounds and academic interests. More than 70% of student participants met the skills-based learning objectives of the Academy. Results of the summer program student evaluations demonstrated higher than average ratings of satisfaction with the instruction. The VIDS Academy represents a promising model for the delivery of instruction in data science and Science, Technology, Engineering, and Mathematics-related fields that address the local workforce and healthcare needs of Tribal communities.
Gestational diabetes mellitus (GDM) is a state of hyperglycemia during pregnancy. GDM can increase risk of birth complications, and subsequent type 2 diabetes mellitus in the mother and offspring. Risk factors such as diet, BMI, and family history have demonstrated strong association with GDM but no clear pathophysiology has been ascertained. A diagnosis of GDM was abstracted from participant medical records. Analysis was conducted on 41 women with and 320 without GDM. Genotypes of 7 genetic variants were available. Additive and dominant genetic models were evaluated by chi-square and multivariate logistic regression methods. A genetic risk score comprised of total risk alleles among the 7 variants was also evaluated. Multivariate logistic regression showed significant, independent, positive associations between body-mass index (BMI), age, the posited genetic risk score and GDM. Genetic variant rs1421085 was associated with GDM in multivariate analysis (OR 2.12, 1.08-4.15, p = 0.029. The polygenic risk score showed association with GDM, odds ratio 1.27, 95% CI 1.09 - 1.48, p = 0.003. This relatively small cohort replicated a previously proposed polygenic risk score, as well as established risk factors for GDM (age and BMI). A previous meta-analysis in the literature reported a degree of heterogeneity between the 3 large cohorts analyzed, suggesting that the effect of these variants may differ according to genetic background. We validate a previously published polygenic risk score for GDM in an ethically unrelated population. (Supported by NIGMS P20GM103442)
Fine-grained emotion classification (FEC) aims to analyze speakers’ utterances and distinguish dozens of emotions with subtle differences, allowing for a more nuanced understanding of human emotional states. However, compared to traditional coarse-grained emotion classification, two difficulties arise as the granularity of emotions becomes finer, i.e., the presence of closely confusable emotions which are hard to distinguish, and the biased performance caused by long-tailed emotions. Although addressing both difficulties is vital to FEC, previous studies have predominantly focused on dealing with only one of them. In this paper, we propose TACO, a novel triple-view framework that treats FEC as an instance-label (i.e., utterance-emotion) joint embedding learning problem to tackle both difficulties concurrently by considering three complementary views. Specifically, we design a clustering-guided contrastive loss, which incorporates clustering techniques to guide the contrastive learning process and facilitate more discriminative instance embeddings. Additionally, we introduce the emotion label description as a helpful resource to refine label embeddings and mitigate the poor performance towards under-represented (i.e., long-tailed) emotions. Extensive experiments on two widely-used benchmark datasets demonstrate that our proposed TACO achieves substantial and consistent improvements compared to other competitive baseline methods.
Maternal variants including C-Reactive Protein, CRP rs1205 have previously been associated with risk of pre-eclampsia (PE). These findings were replicated in two non-American Indian populations. The rs1205 T allele is associated with reduced serum levels of CRP. Our objective was to determine if the fetal rs1205 genotype contributed to maternal risk of PE independent of maternal rs1205 genotype. Only offspring of both case and control mothers heterozygous for rs1205 were enrolled, thus controlling for maternal genetic influence at this locus. Offspring were then genotyped for rs1205 by TaqMan assay. Association was assessed by chi-square and multivariate logistic regression. Offspring of 10 of 45 normal pregnancies and 11 of 24 PE pregnancies exhibited the rs1205 C allele recessive genotype (Pearson chi square p = 0.042). Multivariate logistic regression analysis adjusted for maternal age, nulliparity and BMI demonstrates an odds ratio of 3.603, p = 0.043, 95% CI 1.042-12.457 for the fetal, C recessive genotype. Among 69 women, heterozygous for the rs1205 allele, both chi-square and multivariate adjusted logistic analysis shows significant association of PE among pregnancies with fetal rs1205 C recessive genotypes. This is consistent with previous findings of reduced risk associated with this maternal genotype, and with a pathophysiologic model wherein increased placental CRP expression increases risk of PE. (Supported by NIGMS P20GM103442)
