Klees, Nora (2026) Demographic Moderation of Lifestyle-Diversity Associations in the American Gut Project. Bachelor thesis, Data Science and Society (DSS).
|
PDF
BA5694302NEKlees.pdf Download (1MB) | Preview |
Abstract
The gut microbiome's alpha diversity is shaped by both lifestyle and demographic factors, yet most research treats these as parallel predictors rather than asking whether demographic context changes how lifestyle relates to diversity. This thesis examines that question in the American Gut Project cohort. A random forest regression modeled Shannon diversity from nineteen lifestyle and anthropometric features, fit once on the full cohort and separately within nine demographic subgroups defined by age, BMI, sex, and country, with permutation importance used to rank predictors. Lifestyle carried only a weak predictive signal overall, with a global R² of 0.044., consistent with survey-based microbiome studies. A permutation test supports the model exceeding chance. Additionally, that predictability itself differed by subgroup. The model recovered signal in most groups, including the United States, the 19-to-60 group, and normal-weight participants, but not in the United Kingdom, the obese group, or adults over 61. Among the models that predicted, probiotic frequency and antibiotic history carried most of the behavioral signal in the United States, while BMI remained among the strongest contributors throughout. The number of plant types, a prominent univariate finding, attenuated to near zero once correlated variables were included. These differences are observed across subgroups rather than formally tested, but they show that demographic framing reveals patterns a single combined model conceals.
| Item Type: | Thesis (Bachelor) |
|---|---|
| Name supervisor: | Haleem, N. |
| Date Deposited: | 10 Jun 2026 11:47 |
| Last Modified: | 10 Jun 2026 11:47 |
| URI: | https://campus-fryslan.studenttheses.ub.rug.nl/id/eprint/819 |
Actions (login required)
 |
View Item |