Lung spirometry parameters and diffusion capacity are decreased in patients with Type 2 diabetes.

AIMS: In cohort studies, Type 2 diabetes mellitus has been associated with decreased forced 1 s expiratory volume and forced vital capacity. We examined if forced vital capacity, forced 1 s expiratory volume and diffusion lung capacity correlate with diabetes mellitus across different races in a clinical setting. METHODS: We examined the medical records of 19,882 adults 18-97 years of age in our centre from 1 January 2000 to 1 May 2009. After excluding patients with diseases causing abnormal lung function, 4164 subjects were available for analysis. We used multiple linear regressions to examine cross-sectional differences in forced vital capacity, forced 1 s expiratory volume and carbon monoxide diffusing capacity between patients with and without diabetes mellitus, after adjustment for age, sex, race, height, smoking, BMI and heart failure. RESULTS: Patients with diabetes (n = 560) were older (62 ± 12 vs. 55 ± 16 years), more likely to be men (56 vs. 43%), overweight (BMI 31.7 ± 8.5 vs. 27.3 ± 6.7 kg/m2 ), have heart failure (33 vs. 14%) and less likely to be Caucasians (65 vs. 76%) and never smokers (66 vs. 72%) compared with patients without diabetes (n = 3604). The mean unadjusted values in patients with diabetes vs. those without were: forced vital capacity 2.78 ± 0.91 vs. 3.19 ± 1.03 l; forced 1 s expiratory volume 2.17 ± 0.74 vs. 2.49 ± 0.0.83; and carbon monoxide diffusing capacity 16.67 ± 5.53 vs. 19.18 ± 6.72 ml(-1) min(-1) mmHg, all P < 0.0001. These differences remained significant after adjustment for covariates. After race stratification, only Caucasians with diabetes had a significant decrease in all lung function measures. CONCLUSIONS: Patients with diabetes have decreased lung function compared with those without diabetes. Caucasians with diabetes have more global lung function impairment compared with African-Americans and Hispanics.

Mixtures of probabilistic principal component analyzers.

Principal component analysis (PCA) is one of the most popular techniques for processing, compressing, and visualizing data, although its effectiveness is limited by its global linearity. While nonlinear variants of PCA have been proposed, an alternative paradigm is to capture data complexity by a combination of local linear PCA projections. However, conventional PCA does not correspond to a probability density, and so there is no unique way to combine PCA models. Therefore, previous attempts to formulate mixture models for PCA have been ad hoc to some extent. In this article, PCA is formulated within a maximum likelihood framework, based on a specific form of gaussian latent variable model. This leads to a well-defined mixture model for probabilistic principal component analyzers, whose parameters can be determined using an expectation-maximization algorithm. We discuss the advantages of this model in the context of clustering, density modeling, and local dimensionality reduction, and we demonstrate its application to image compression and handwritten digit recognition.