PubMed Health⌕ Search
Home/Archive/PubMed

PubMed · 2130498

[Classification of multidimensional observations using linear discriminant functions].

H Skalská·P Petýrek·O Neruda
View original source ↗Export metadata ↓

Abstract

In the present paper the role of linear discriminant analysis is delimitated all along with a range of problems which could be solved using this method in appropriate conditions. The analysis is given of possibilities on how to involve the decrease in number of variables in a model, and how to predict probability of incorrect classification. As an example of application, the prediction of dog survival is calculated in radiobiological experiment on the basis of biochemical and hematological values which were obtained by the 2nd day after the irradiation as well as an error probability was determined.

Explore related subjects

Discriminant AnalysisEnglish AbstractJournal Article

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

H Skalská, P Petýrek, O Neruda. 1990. [Classification of multidimensional observations using linear discriminant functions].. https://pubmed.ncbi.nlm.nih.gov/2130498/

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed1999

Microstructure of longitudinal 24 hour electroencephalograms in healthy preterm infants.

BACKGROUND: The aims of the present study were to find electroencephalographic parameters that appropriately represent the microstructure of electroencephalograms (EEG) in different sleep states and to find quantitative criteria for an automatic system of sleep-state classification in preterm infants. METHODS AND RESULTS: Continuous 24 h EEG was performed in 14 normal preterm infants for whom 26 EEG records were obtained. Based on respiratory activity, body movements and rapid eye movements, the different sleep states were determined visually in 30 s epochs. Seven EEG parameters, Minimum Akaike Information Criterion (Min-AIC), total power (TP), component powers (delta, theta, alpha and beta), and the discontinuity were calculated by means of autoregressive and component analyses in 30 s epochs. The student's t test was performed independently for each parameter. Four of the seven parameters (Min-AIC, TP, delta component power and the discontinuity) showed significant differences in different sleep states. The results of multivariate discriminant analysis revealed that the combination of Min-AIC, TP, delta component power and the discontinuity of EEG defined the EEG sleep states well. CONCLUSION: The combination of Min-AIC. TP, delta component power and the discontinuity of EEG defined the EEG sleep states well and might be used to predict sleep state changes in preterm infants of conceptional ages of more than 30 weeks.

Discriminant Analysis↗
PubMed1998

Sexual dimorphism in the crania and mandibles of South African whites.

Numerous studies have clearly demonstrated that skeletal characteristics vary by population. To date, there are no metric cranial criteria for South African whites. Therefore, the purpose of this study is to establish population specific standards for sex determination from the skull. A total of 12 standard cranial and five mandibular measurements were taken from 44 male and 47 female skeletons of known sex and race from the Pretoria and Dart collections. These were subjected to SPSS discriminant function analysis. Bizygomatic breadth was the most dimorphic dimension. Five functions were developed from the complete cranium, vault, face, mandible and bizygomatic breadth. Dimensions from the complete cranium provided the best accuracy. In the mandible, bigonial breadth was the most dimorphic of the measurements taken. Average accuracies ranged from 80% (bizygomatic breadth alone) to 86% (cranium). These accuracies are similar to those obtained by researchers on other groups (e.g., 84% in Japanese crania and about 86% in American whites and blacks). Diagnostic accuracy, however, is lower than that obtained from the South African femur and tibia.

Discriminant Analysis↗
PubMed1998

Relative value of clinical and transesophageal echocardiographic variables for risk stratification in patients with infective endocarditis.

BACKGROUND: Infective endocarditis remains a life-threatening disease, and its optimal management is of paramount importance. Transesophageal echocardiography (TEE) is useful for the diagnosis of endocarditis-induced lesions, but the prognostic significance of the method remains controversial. HYPOTHESIS: The purpose of this study was to relate clinical and TEE characteristics to the occurrence of mortality and/or systemic embolization in a consecutive series of 45 patients with a diagnosis of infective endocarditis. METHODS: All patients underwent at least one monoplane TEE. Clinical data, episodes of embolization, and echocardiographic characteristics were prospectively recorded. Stepwise logistic discriminant analysis was performed to identify the independent variables that best predicted three binary outcomes: systemic embolization, death, and systemic embolization and/or death. RESULTS: Twelve of the 45 patients (27%) died from the endocarditis. Significant univariate predictors of death were the presence of paravalvular abscess (p = 0.025), number of vegetations (p = 0.021), Staphylococcus aureus isolated in blood cultures (p = 0.002), medical treatment alone (p < 0.002), and systemic embolism (p < 0.001). In multivariate analysis, systemic embolism (chi 2 = 29.3; p < 0.01), echocardiographic evidence of paravalvular abscess (chi 2 = 5.6; p = 0.018), Staphylococcus aureus endocarditis (chi 2 = 5.5; p = 0.016), and medical treatment alone (chi 2 = 5.11; p = 0.024) emerged as optimal predictors of death. Systemic embolization occurred in 12 patients. Independent variables predicting systemic embolization were a total length of vegetations > 14 mm (p = 0.01), greater age (p = 0.02), and medical treatment alone (p = 0.03). When two or more vegetations were observed, the total length is the sum of the individual sizes. Independent risk factors for the development of systemic emboli and/or death as a combined end point were total length of vegetations on TEE (chi 2 = 6.4; p = 0.003) and medical treatment alone (chi 2 = 4.1; p = 0.047). CONCLUSIONS: High-risk patients may be identified by the combination of clinical variables and TEE characteristics.

Discriminant Analysis↗