High-Risk Populations: The Pimas of Arizona and Mexico

Limesand

Journal of Endocrinology

2019

184

162

Section: Dohad and Evidence From Human Studiesmentioning

confidence: 82%

Epigenetic responses and the developmental origins of health and disease

Limesand

Journal of Endocrinology

2019

184

162

“…Pima Indians in Arizona represent a classic and well-studied example with a very high prevalence of T2D. Clinically, their diabetes appears to be a subtype of T2D, characterized by obesity, insulin resistance, and a relative insulin deficiency [4,5]. The increase in prevalence among Pima Indians has occurred along with a transition from a traditional lifestyle with limited food supply and high physical activity to a modern, sedentary lifestyle with a consistent food supply.…”

Section: Prevalence Of Type 2 Diabetesmentioning

confidence: 99%

Diabetes in Population Isolates: Lessons from Greenland

Grarup¹,

Moltke²,

Albrechtsen³

et al. 2015

Rev Diabet Stud

■ AbstractType 2 diabetes (T2D) is an increasing health problem worldwide with particularly high occurrence in specific subpopulations and ancestry groups. The high prevalence of T2D is caused both by changes in lifestyle and genetic predisposition. A large number of studies have sought to identify the genetic determinants of T2D in large, open populations such as Europeans and Asians. However, studies of T2D in population isolates are gaining attention as they provide several advantages over open populations in genetic disease studies, including increased linkage disequilibrium, homogeneous environmental exposure, and increased allele frequency. We recently performed a study in the small, historically isolated Greenlandic population, in which the prevalence of T2D has increased to more than 10%. In this study, we identified a common nonsense variant in TBC1D4, which has a population-wide impact on glucose-stimulated plasma glucose, serum insulin levels, and T2D. The variant defines a specific subtype of non-autoimmune diabetes characterized by decreased post-prandial glucose uptake and muscular insulin resistance. These and other recent findings in population isolates illustrate the value of performing medical genetic studies in genetically isolated populations. In this review, we describe some of the advantages of performing genetic studies of T2D and related cardio-metabolic traits in a population isolate like the Greenlandic, and we discuss potentials and perspectives for future research into T2D in this population.

“…The Pima Indians have been willing participants in a myriad of epidemiological, clinical and genetic studies that have ultimately contributed to a better understanding of the pathogenesis and diagnosis of type II diabetes and obesity. Here, we study a dataset acquired from the National Institute of Diabetes and Digestive and Kidney Diseases and fit logistic regression and random forest models to the data to predict diabetes as a binary outcome as defined by the World Health Organization [30,31].…”

Section: Case Study 2: Random Forest and Logistic Regression Featuresmentioning

confidence: 99%

“…In our next example, we will study a case where adding interaction terms did not improve model performance. Diseases and fit logistic regression and random forest models to the data to predict diabetes as a binary outcome as defined by the World Health Organization [30,31]. The logistic regression and random forest model both exhibit similar performance (0.83 Cstatistic; Table 1) while generally agreeing on which variables are the most important for the model's decisions.…”

Section: Selected Biomedical Case Studiesmentioning

confidence: 99%

Don’t Dismiss Logistic Regression: The Case for Sensible Extraction of Interactions in the Era of Machine Learning

Levy

O’Malley

2019

Preprint

Background:Machine learning approaches have become increasingly popular modeling techniques, relying on data-driven heuristics to arrive at its solutions. Recent comparisons between these algorithms and traditional statistical modeling techniques have largely ignored the superiority gained by the former approaches due to involvement of model-building search algorithms. This has led to alignment of statistical and machine learning approaches with different types of problems and the under-development of procedures that combine their attributes. In this context, we hoped to understand the domains of applicability for each approach and to identify areas where a marriage between the two approaches is warranted. We then sought to develop a hybrid statistical-machine learning procedure with the best attributes of each. Methods:We present three simple examples to illustrate when to use each modeling approach and posit a general framework for combining them into an enhanced logistic regression model building procedure that aids interpretation. We study 556 benchmark machine learning datasets to uncover when machine learning techniques outperformed rudimentary logistic regression models and so are potentially well-equipped to enhance them. We illustrate a software package, InteractionTransformer, which embeds logistic regression with advanced model building capacity by using machine learning algorithms to extract candidate interaction features from a random forest model for inclusion in the model. Finally, we apply our enhanced logistic regression analysis to two real-word biomedical examples, one where predictors vary linearly with the outcome and another with extensive second-order interactions. Results:Preliminary statistical analysis demonstrated that across 556 benchmark datasets, the random forest approach significantly outperformed the logistic regression approach. We found a statistically significant increase in predictive performance when using hybrid procedures and greater clarity in the association with the outcome of terms acquired compared to directly interpreting the random forest output. Conclusions:When a random forest model is closer to the true model, hybrid statistical-machine learning procedures can substantially enhance the performance of statistical procedures in an automated manner while preserving easy interpretation of the results. Such hybrid methods may help facilitate widespread adoption of machine learning techniques in the biomedical setting.