SEARCH · PubMed Health

Results for “Multiple Classification Analysis”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 163 records · Page 9Linked to original sources

PubMed2002

Medical burden, cerebrovascular disease, and cognitive impairment in geriatric depression: modeling the relationships with the CART analysis.

Herein, the background information reflecting roles of medical burden, cerebrovascular disease and risk factors, and cognitive impairment in geriatric depression are reviewed. The authors then propose a nonparametric statistical approach to the data analysis of multiple putative causal variables for late-life depression, the Classification and Regression Tree Analysis. This analysis presents a useful approach to modeling nonlinear relationships and interactions among variables measuring physical and mental health, as well as magnetic resonance imaging and cognitive measures in depressed elderly. This method uncovers the existing interactions among multiple predictor variables, and provide thresholds for each variable, at which its predictive power becomes statistically significant. It presents a "hierarchy" of the predictors in a form of a decision tree by finding the best combination of predictors of an outcome. The authors present two models based on demographic variables, measures of vascular and nonvascular medical burden, neuroimaging indices, the Mini-Mental State Examination score, and neuropsychological test scores of 81 elderly depressed subjects. Cognitive tests of verbal fluency and executive function are identified as the best predictors of depression, followed by the frontal lobe volume and Mini-Mental State Examination. The authors observed that an interaction between frontal lobe volume, total lesion volume, and medical burden was predictive of depression.

Journal Article↗

PubMed2005

Tools for statistical analysis with missing data: application to a large medical database.

Missing data is a common feature of large data sets in general and medical data sets in particular. Depending on the goal of statistical analysis, various techniques can be used to tackle this problem. Imputation methods consist in substituting the missing values with plausible or predicted values so that the completed data can then be analysed with any chosen data mining procedure. In this work, we study imputation in the context of multivariate data and we evaluate a number of methods which can be used by today's standard statistical software packages. Imputation using multivariate classification, multiple imputation and imputation by factorial analysis are compared using simulated data and a large medical database (from the diabetes field) with numerous missing values. Our main result is to provide a control chart for assessing data quality after the imputation process. To this end, we developed an algorithm for which the input is a set of parameters describing the underlying data (e.g., covariance matrix, distribution) and the output is a chart which plots the change in the prediction error with respect to the proportion of missing values. The chart is built by means of an iterative algorithm involving four steps: (1) a sample of simulated data is drawn by using the input parameters; (2) missing values are randomly generated; (3) an imputation method is used to fill in the missing data and (4) the prediction error is computed. Steps 1 to 4 are repeated in order to estimate the distribution of the prediction error. The control chart was established for the 3 imputation methods studied here, assuming a multivariate normal distribution of data. The use of this tool on a large medical database was then investigated. We show how the control chart can be used to assess the quality of the imputation process in the pre-processing step upstream of data mining procedures.

Algorithms↗

PubMed1999

Proposed classification criteria of psoriatic arthritis. A preliminary study in 260 patients.

UNLABELLED: Psoriatic arthritis probably owes to its radioclinical presentation its position as the most controversial and poorly understood of all major chronic inflammatory joint diseases. Differentiating psoriatic arthritis from ankylosing spondylitis and rheumatoid arthritis remains difficult. OBJECTIVE: To conduct a statistical analysis aimed at identifying clinical, radiological, and laboratory criteria for classifying psoriatic arthritis. PATIENTS AND METHODS: 260 patients were studied retrospectively, including 100 cases with psoriatic arthritis and 160 controls with ankylosing spondylitis meeting Amor's criteria (n = 80) or with rheumatoid arthritis meeting American College of Rheumatology criteria (n = 80). Mean disease duration was five years. Thirty-nine variables were recorded for each patient. Multiple logistic regression and discriminant analysis were used to select the classification criteria. RESULTS: Each of the two statistical methods selected the same nine criteria. After assigning a weighting coefficient to each of these criteria, sensitivity and specificity were better with the multiple logistic regression model (95% and 98%, respectively) than with the discriminant analysis model. CONCLUSION: Our classification criteria require further evaluation in multicenter prospective studies.

Adult↗

PubMed2005

Cross-platform analysis of cancer microarray data improves gene expression based classification of phenotypes.

BACKGROUND: The extensive use of DNA microarray technology in the characterization of the cell transcriptome is leading to an ever increasing amount of microarray data from cancer studies. Although similar questions for the same type of cancer are addressed in these different studies, a comparative analysis of their results is hampered by the use of heterogeneous microarray platforms and analysis methods. RESULTS: In contrast to a meta-analysis approach where results of different studies are combined on an interpretative level, we investigate here how to directly integrate raw microarray data from different studies for the purpose of supervised classification analysis. We use median rank scores and quantile discretization to derive numerically comparable measures of gene expression from different platforms. These transformed data are then used for training of classifiers based on support vector machines. We apply this approach to six publicly available cancer microarray gene expression data sets, which consist of three pairs of studies, each examining the same type of cancer, i.e. breast cancer, prostate cancer or acute myeloid leukemia. For each pair, one study was performed by means of cDNA microarrays and the other by means of oligonucleotide microarrays. In each pair, high classification accuracies (> 85%) were achieved with training and testing on data instances randomly chosen from both data sets in a cross-validation analysis. To exemplify the potential of this cross-platform classification analysis, we use two leukemia microarray data sets to show that important genes with regard to the biology of leukemia are selected in an integrated analysis, which are missed in either single-set analysis. CONCLUSION: Cross-platform classification of multiple cancer microarray data sets yields discriminative gene expression signatures that are found and validated on a large number of microarray samples, generated by different laboratories and microarray technologies. Predictive models generated by this approach are better validated than those generated on a single data set, while showing high predictive power and improved generalization performance.

Animals↗

PubMed2004

Customizing chemotherapy for colon cancer: the potential of gene expression profiling.

The value of gene expression profiling, or microarray analysis, for the classification and prognosis of multiple forms of cancer is now clearly established. For colon cancer, expression profiling can readily discriminate between normal and tumor tissue, and to some extent between tumors of different histopathological stage and prognosis. While a definitive in vivo study demonstrating the potential of this methodology for predicting response to chemotherapy is presently lacking, the ability of microarrays to distinguish other subtleties of colon cancer phenotype, as well as recent in vitro proof-of-principle experiments utilizing colon cancer cell lines, illustrate the potential of this methodology for predicting the probability of response to specific chemotherapeutic agents. This review discusses some of the recent advances in the use of microarray analysis for understanding and distinguishing colon cancer subtypes, and attempts to identify challenges that need to be overcome in order to achieve the goal of using gene expression profiling for customizing chemotherapy in colon cancer.

Colonic Neoplasms↗

PubMed2003

Biochemical markers of fibrosis in patients with chronic hepatitis C: a comparison with prothrombin time, platelet count, and age-platelet index.

As an alternative to liver biopsy, an index of five biochemical markers (alpha2-macroglobulin, apolipoprotein A1, haptoglobin, total bilirubin, gamma-glutamyl-transpeptidase) has been shown to predict the severity of hepatitis C-related fibrosis. The objective of this study was to compare this index with other markers frequently used for this purpose (prothrombin time, platelets, age-platelet index). In 323 hepatitis C-infected patients, the discriminative values of these markers for F2-F4 fibrosis (by the METAVIR classification) were compared. By multiple logistic regression analysis, only the five-marker index (P < 0.0001) and prothrombin time (P = 0.02) were independently predictive of F2-F4 fibrosis. For this outcome, the area under the receiver operating characteristic curve was significantly higher for the five-marker index (0.836 +/- 0.024) than the age-platelet index (P = 0.002), and the platelet count and prothrombin time (P < 0.001), indicating greater diagnostic value. The addition of the latter markers to the five-marker index proved unhelpful for increasing its accuracy. In conclusion, an index of five biochemical markers accurately predicts significant hepatitis C-related fibrosis and is superior to traditional markers.

Adult↗

PubMed2003

Pregnancy complications and the risk of asthma among Norwegians born between 1967 and 1993.

BACKGROUND: Fetal life events may affect the development of the immune and/or respiratory system and increase the risk of asthma and allergic diseases. The objective of this study was to test the hypothesis that pregnancy complications are associated with the risk of developing asthma in the off-spring. METHODS: The study population comprised Norwegian live births 1967-1993 (n = 1,548,429) linking the Medical Birth Registry of Norway (MBRN) (exposure variables) and the National Insurance Administration Register (NIAR) (outcome variables), which covers all Norwegians. The MBRN variables included pregnancy complications, pregnancy outcomes and diseases of the mother. The NIAR provided data on all Norwegians who had received cash benefit for treatment of asthma from 1967 to 1996 (n = 5938, 3.9/1000 persons). RESULTS: In multiple logistic regression analysis, pregnancy complications (International Classification of Diseases (ICD)-8-codes: 630-634) were associated with the risk of asthma (odds ratio 1.82, 95% confidence interval: 1.67-1.98). This was also the case if analyses were performed in different strata according to year of birth, plurality, maternal atopy, geographical district of birth, and maternal education. CONCLUSIONS: Pregnancy complications may represent risk factors for the development of asthma in the offspring or express early signs of increased risk for developing the disease.

Adolescent↗

PubMed1990

Impact of initial triage decisions on nursing intensity for patients with acute chest pain.

The results of a prospective evaluation of the patient-specific correlates of nursing intensity for 183 consecutive emergency room patients admitted for evaluation of acute chest pain, including 33 (18%) with acute myocardial infarction (AMI), are reported. These correlates were measured with a previously-validated, commercially-available patient classification tool (Medicus). In multiple linear regression analysis that adjusted for the effects of 31 clinical variables from the initial presentation and subsequent course, initial triage to the coronary care unit had a significant independent correlation with initial Medicus score (P less than 0.0001) and mean Medicus score from the first three days of hospitalization (P less than 0.0001). In a subset of 74 uncomplicated non-AMI patients, coronary care unit admissions were more likely to have vital signs taken every 2 hours, to receive oxygen therapy and assistance with feeding, and to be transferred to another unit within three days. Findings suggest that, after adjusting for severity of illness, initial triage of patients with acute chest pain to the coronary care unit is associated with increased nursing utilization because of 1) the routine application of standard coronary care unit protocols that were developed for high-risk patients, and 2) the nursing efforts required by early transfer of patients out of the coronary care unit.

Adult↗

PubMed1985

A contribution to the validity of Leonhard's classification of endogenous psychoses.

The author made a clinical-genetical study of the schizophrenias and cycloid (schizoaffective) psychoses following the Leonhardian classification, using the traditional pedigree analysis and multiple threshold method. According to the latter method--on the basis of phenotypical correlations--the systematic schizophrenias could be clearly distinguished from the cycloid psychoses, while the non-systematic category presumably occupies a genetical position between the two former categories.

Adolescent↗

PubMed2004

Neuroanatomical localisation and clinical correlates of white matter lesions in the elderly.

BACKGROUND: White matter lesions (WML) in elderly people co-occur with hypertension, depression, and cognitive impairment. Little is known about the density and distribution of WML in normal elderly people, whether they occur randomly in the aging brain or tend to cluster in certain areas, or whether patterns of WML aggregation are linked to clinical symptoms. OBJECTIVES: To describe patterns of WML distribution in a large representative population of elderly people using non-inferential cluster analysis; and to determine the extent to which such patterns are associated with clinical symptomatology. METHOD: A population sample of 1077 elderly people was recruited. Multiple analysis of correspondence followed by automatic classification methods was used to explore overall patterns of WML distribution. Correspondence was then sought between these patterns and a range of cerebrovascular, psychiatric, and neurological symptoms. RESULTS: Three distinct patterns of spatial localisation within the brain were observed, corresponding to distinct clusters of clinical symptoms. In particular WML aggregation in temporal and occipital areas was associated with greater age, hypertension, late onset depressive disorder, poor global cognitive function, and overall WML frequency. CONCLUSIONS: WML localisation is not random in the aging brain, and their distribution is associated with age and the presence of clinical symptoms. Age differences suggest there may be patterns of progression across time; however, this requires confirmation from longitudinal imaging studies.

Aged↗

PubMed2005

What characteristics of applicants to emergency medicine residency programs predict future success as an emergency medicine resident?

OBJECTIVES: Program directors of emergency medicine (EM) residencies attempt to select candidates who will subsequently perform well as residents. This study was undertaken to identify characteristics available at the time of application to an EM residency that predict future success in residency. METHODS: The EM faculty at the University of California San Diego (UCSD) completed a one-time confidential assessment of EM residents on performance in residency at the time of graduation. The faculty member compared the graduate with all residents (both EM and non-EM) with whom the faculty member had previously worked using the five-point scale: > or =90th percentile, 70th-89th percentile, 50th-69th percentile, 30th-49th percentile, or < 30th percentile. Descriptive statistics, ordinal logistic regression (OLR), classification and regression tree (CART) analysis, and multiple additive regression tree (MART) analysis were used to find predictors for each of the outcome variables. RESULTS: Fifty-four graduates were evaluated. The medical school attended (MSA) was the strongest predictor of overall performance in residency in all regression models. OLR showed that MSA and "distinctive factors" (being a championship athlete, medical school officer, etc.) were significant predictors and may deserve greater weighting in the selection process. The most robust MART model demonstrated that MSA, dean's letter of recommendation, and distinctive factors had the most impact on overall performance in an EM residency. CONCLUSIONS: Using regression modeling, it may be possible to predict future resident performance from characteristics contained in residency applications. Applicants from top-tier medical schools and those with distinctive talents were more successful in the UCSD EM residency.

Adult↗

PubMed1981

Predictors of the outcome of transsphenoidal surgery for prolactin-secreting pituitary adenomas.

Sixty-seven women who underwent transsphenoidal resection of PRL-secreting pituitary adenomas between 1976 and 1979 were separated into categories according to the manner of their clinical presentation. Forty-one had oral contraceptive-related onset of amenorrhea-galactorrhea, 5 had pregnancy-related onset of amenorrhea-galactorrhea, 5 had primary amenorrhea, and 16 had the spontaneous onset of amenorrhea-galactorrhea unrelated to estrogen use. Surgical success, defined as the resumption of regular menses and normalization of serum PRL concentration, was achieved in 54% of those with estrogen-related onset of amenorrhea-galactorrhea compared with 19% in the other group. The analysis of multiple preoperative features, including clinical classification, age, preoperative PRL concentration, and duration of amenorrhea-galactorrhea by logistic regression, demonstrated that the clinical classification was the most important factor in predicting the outcome of transsphenoidal surgery. It should be a prime consideration in the selection of therapy for PRL-secreting adenomas.

Adenoma↗

PubMed1997

[Development and evaluation of a spectroscopy system for classification of laser-induced arterial fluorescence spectra].

The present study evaluated the potential of fluorescence guidance of laser angioplasty without using a second laser for fluorescence excitation. A prototype spectroscopy system with a grating spectrograph, microchannel plate, CCD array and digital image processor on a personal computer was developed and coupled to a clinical XeCl excimer laser. Using multifibre catheters, specimens of human aorta were ablated in physiological saline and blood. The spectra thus generated were recorded and validated histologically. Five types of spectra could be differentiated. Based on a training set, classification algorithms were developed using multiple linear regression and linear discriminant analysis with intensity ratios as predictor variables. Discriminant analysis yielded prospective classification of the remaining validation spectra with high sensitivity and specificity for each type. These data demonstrate that fluorescence spectroscopy during excimer laser ablation at 308 nm does not require a diagnostic laser. Principal types of atherosclerotic lesions and the media can be differentiated spectroscopically in physiological saline and blood.

Adult↗

PubMed1989

The oral health of a group of 15-17 year old British school children of different ethnic origin.

The influence of ethnic origin on the oral health of a group of 472 British schoolchildren aged 15-17 years attending two adjacent schools was investigated. Ethnic origin was assessed as European (E; n = 197), Afro-Caribbean (AC; n = 97) and Asian (As; n = 140) leaving 38 subjects of other or mixed origin. Plaque, bleeding on probing and subgingival retentive factors, together with probing depths and attachment loss were recorded at the four approximal sites of teeth 16, 11, 26, 36, 31, 46. DMFT scores were collected for the whole mouth. Socio-economic status was assessed using a Classification of Residential Neighbourhoods (ACORN). Multiple regression analysis indicated that ethnic origin had a significant influence on plaque, subgingival calculus, gingivitis, pockets and DMFT scores. The overall prevalence of periodontitis was 11.2 per cent; European, Afro-Caribbean and Asian groups had prevalences of 3, 21 and 14 per cent respectively.

Adolescent↗

PubMed1994

[Long-term results of surgical therapy of colon cancer. Results of the Colorectal Cancer Study Group].

In a prospective multicentre observation study with free choice of treatment 1157 patients with solitary colonic carcinoma were entered free of selection between August 1984 and November 1986. The 5-year-survival rate for all patients was observed 45.7 +/- 3.0%, relative (age-corrected) 60.2 +/- 4.0%. Multivariate analysis using multiple logistic regression analysis identified R-classification, stage, institution and timing of surgery as independent prognostic factors regarding survival. In R0 resected patients occurrence of locoregional recurrence was the strongest prognostic factor. Study data could demonstrate a substantial variability in frequency of locoregional recurrence and survival among the participating institutions. The data support the importance of application of principles of radical surgery and gives evidence for the outstanding importance of surgery for the prognosis of colonic carcinoma.

Adenocarcinoma↗

PubMed1980

Predicting speech discrimination from the audiometric thresholds.

To develop a method for predicting a speech discrimination score (SDS) from audiometric thresholds (SRT and pure-tones) three prediction systems were investigated: a stepwise multiple regression procedure, smear-and-sweep analysis and a clinical classification of the audiometric configuration. Test results of 529 ears with sensorineural hearing loss were taken from copies of audiograms obtained as part of a normal audiology clinic caseload. The three prediction systems had similar predictive ability and yielded slightly higher correlations with the SDS than those in previously reported studies. Squared correlations in this study ranged from 0.58 to 0.60. Smear-and-sweep analysis yielded the best results; however, its complexity makes clinical application difficult at this time. The stepwise multiple regression models or the clinical classification system provided more clinically useful methods for predicting the SDS. An over-riding influence of increasing variability in the SDS with increased hearing loss was observed and significantly limited the accuracy of prediction for the moderate-to-severe hearing loss groups. Small changes in the slope of the audiometric configuration were noted to affect the SDS only when the degree of hearing loss was slight.

Audiometry↗

PubMed2000

Computer aided gnathosonic analysis: distinguishing between single and multiple tooth impact sounds.

Watt developed a classification of tooth contact sounds that distinguished between the short sharp, reproducible sounds heard when the teeth meet simultaneously and the dull prolonged, poorly reproducible sounds heard when tooth contacts are sequential. However, when a large occlusal prematurity, for instance a high restoration, is introduced, tooth contact sounds are also short sharp and highly reproducible. In this study, a method of distinguishing single from multiple tooth contact sounds is described, based on an analysis of the phase and amplitude of sounds detected by headphones placed over the ears.

Adult↗

PubMed2001

[Role of HDL cholesterol in the prediction of exercise stress test abnormalities in asymptomatic high-risk patients].

The diagnosis of coronary artery disease in asymptomatic patients is useful in order to target therapeutic intervention in the patients at highest risk. Systematic testing of all asymptomatic adults with coronary risk factors is not feasible. The aim of this study, carried out in 950 healthy subjects, was to assess the predictive value of classical risk factors for positive exercise stress tests (EE). All subjects underwent stress testing using the Bruce protocol. Statistical analysis was performed by multiple logistic regression on half the samples, then by CART (Classification and Regression Trees) analysis on all subjects. Age, HDL-cholesterol and interaction between lipid lowering treatment and LDL-cholesterol were significantly correlated (p < 0.05) to a positive exercise stress test. In both groups, treated or untreated by lipid lowering drugs. CART identified HDL-cholesterol (< 0.40 g/l) as a predictive factor for positive stress testing. Subgroups of elderly patients (> or = 60 years) with probabilities of 20 to 28% for a positive stress test were identified. The authors conclude that the diagnosis of coronary artery disease by systematic exercise stress testing is potentially valuable in elderly patients with low HDL-cholesterol values.

Adult↗