Application of logistic regression in epidemiology
Primary Data, Stratification and Moving Average
Bibliographic Data
| ID | 5968635 |
|---|---|
| Authors | Anatoly N Varaksin (0000-0003-2689-3006, Institute of Industrial Ecology), Yulia V Shalaumova (0000-0002-0173-6293, Institute of Plant and Animal Ecology), Tatyana A Maslakova (0000-0001-6642-9027, Institute of Industrial Ecology) |
| Year | 2025 |
| Volume | 31 |
| Issue | 9 |
| Pages | 678-691 |
| Publication date | 2025-04-08 |
| Peer Reviewed | Yes |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Ekologiya Cheloveka (Human Ecology (JOURNAL) |
| Journal identifiers | ISSN: 1728-0869 • E-ISSN: 2949-1444 |
| Publisher | ECO-Vector LLC (PUBLISHER • RU) |
| DOI | 10.17816/humeco642576 |
| OpenAlex | W4410441619 |
| Language | EN |
| References cited | 10 |
Background: Logistic regression is the most commonly used method for establishing statistical relationships between quantitative predictors X and a dichotomous response Y (Y=0 or Y=1). Therefore, it is relevant to develop new approaches to the analysis of relationships between X and Y of this type.Aim: To demonstrate the specific characteristics of the application of stratification, moving average and cumulative probability function methods in the construction and analysis of logistic regression models in the context of health risk assessment.Materials and methods: The analysis of logistic regression models employs a range of statistical methods, including the stratification, moving average, cumulative probability function, goodness-of-fit tests, and proportion comparison tests.Results: It is shown that the standard stratification methods are not sufficient for exploring the nature of the relationships between dichotomous Y and quantitative X. Additional methods, including moving average and cumulative likelihood function, facilitate the identification of features characterizing these relationships. The utility of graphical representations of logistic regression results in elucidating the statistical relationships between variables X and Y is demonstrated. The efficacy of the stratification, moving average and cumulative probability function methods is illustrated by examples from the field of epidemiology.Conclusion: The combination of moving average and cumulative probability function methods with stratification enables the reliable identification of the nature of the relationship between dichotomous Y and quantitative X, as well as the potential for deviations from the conditions of applicability of logistic regression models
Biology · Geography · Logistic regression · Physics · Regression · Statistics · Complementary and Alternative Medicine Studies · Mathematics · Medicine · Epidemiology · Internal Medicine
| Citation velocity | historical |
|---|---|
| Highly cited | No |