PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “Classification Algorithms”

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 1,387 records · Page 77Linked to original sources
PubMed2005

Neuropeptide and calcium-binding protein gene expression profiles predict neuronal anatomical type in the juvenile rat.

Neocortical neurones can be classified according to several independent criteria: morphological, physiological, and molecular expression (neuropeptides (NPs) and/or calcium-binding proteins (CaBPs)). While it has been suggested that particular NPs and CaBPs characterize certain anatomical subtypes of neurones, there is also considerable overlap in their expression, and little is known about simultaneous expression of multiple NPs and CaBPs in morphologically characterized neocortical neurones. Here we determined the gene expression profiles of calbindin (CB), parvalbumin (PV), calretinin (CR), neuropeptide Y (NPY), vasoactive intestinal peptide (VIP), somatostatin (SOM) and cholecystokinin (CCK) in 268 morphologically identified neurones located in layers 2-6 in the juvenile rat somatosensory neocortex. We used patch-clamp electrodes to label neurones with biocytin and harvest the cytoplasm to perform single-cell RT-multiplex PCR. Quality threshold clustering, an unsupervised algorithm that clustered neurones according to their entire profile of expressed genes, revealed seven distinct clusters. Surprisingly, each cluster preferentially contained one anatomical class. Artificial neural networks using softmax regression predicted anatomical types at nearly optimal statistical levels. Classification tree-splitting (CART), a simple binary neuropeptide decision tree algorithm, revealed the manner in which expression of the multiple mRNAs relates to different anatomical classes. Pruning the CART tree revealed the key predictors of anatomical class (in order of importance: SOM, PV, VIP, and NPY). We reveal here, for the first time, a strong relationship between specific combinations of NP and CaBP gene expressions and the anatomical class of neocortical neurones.

Animals↗
PubMed2005

Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.

OBJECTIVES: Implementation of the International Statistical Classification of Disease and Related Health Problems, 10th Revision (ICD-10) coding system presents challenges for using administrative data. Recognizing this, we conducted a multistep process to develop ICD-10 coding algorithms to define Charlson and Elixhauser comorbidities in administrative data and assess the performance of the resulting algorithms. METHODS: ICD-10 coding algorithms were developed by "translation" of the ICD-9-CM codes constituting Deyo's (for Charlson comorbidities) and Elixhauser's coding algorithms and by physicians' assessment of the face-validity of selected ICD-10 codes. The process of carefully developing ICD-10 algorithms also produced modified and enhanced ICD-9-CM coding algorithms for the Charlson and Elixhauser comorbidities. We then used data on in-patients aged 18 years and older in ICD-9-CM and ICD-10 administrative hospital discharge data from a Canadian health region to assess the comorbidity frequencies and mortality prediction achieved by the original ICD-9-CM algorithms, the enhanced ICD-9-CM algorithms, and the new ICD-10 coding algorithms. RESULTS: Among 56,585 patients in the ICD-9-CM data and 58,805 patients in the ICD-10 data, frequencies of the 17 Charlson comorbidities and the 30 Elixhauser comorbidities remained generally similar across algorithms. The new ICD-10 and enhanced ICD-9-CM coding algorithms either matched or outperformed the original Deyo and Elixhauser ICD-9-CM coding algorithms in predicting in-hospital mortality. The C-statistic was 0.842 for Deyo's ICD-9-CM coding algorithm, 0.860 for the ICD-10 coding algorithm, and 0.859 for the enhanced ICD-9-CM coding algorithm, 0.868 for the original Elixhauser ICD-9-CM coding algorithm, 0.870 for the ICD-10 coding algorithm and 0.878 for the enhanced ICD-9-CM coding algorithm. CONCLUSIONS: These newly developed ICD-10 and ICD-9-CM comorbidity coding algorithms produce similar estimates of comorbidity prevalence in administrative data, and may outperform existing ICD-9-CM coding algorithms.

Algorithms↗
PubMed2005

Using aerial video to train the supervised classification of Landsat TM imagery for coral reef habitats mapping.

Management of coral reef resources is a challenging task, in many cases, because of the scarcity or inexistence of accurate sources of information and maps. Remote sensing is a not intrusive, but powerful tool, which has been successfully used for the assessment and mapping of natural resources in coral reef areas. In this study we utilized GIS to combine Landsat TM imagery, aerial photography, aerial video and a digital bathymetric model, to assess and to map submerged habitats for Alacranes reef, Yucatán, México. Our main goal was testing the potential of aerial video as the source of data to produce training areas for the supervised classification of Landsat TM imagery. Submerged habitats were ecologically characterized by using a hierarchical classification of field data. Habitats were identified on an overlaid image, consisting of the three types of remote sensing products and the bathymetric model. Pixels representing those habitats were selected as training areas by using GIS tools. Training areas were used to classify the Landsat TM bands 1, 2 and 3 and the bathymetric model by using a maximum likelihood algorithm. The resulting thematic map was compared against field data classification to improve habitats definition. Contextual editing and reclassification were used to obtain the final thematic map with an overall accuracy of 77%. Analysis of aerial video by a specialist in coral reef ecology was found to be a suitable source of information to produce training areas for the supervised classification of Landsat TM imagery in coral reefs at a coarse scale.

Animals↗
PubMed2004

A Bayesian approach to joint feature selection and classifier design.

This paper adopts a Bayesian approach to simultaneously learn both an optimal nonlinear classifier and a subset of predictor variables (or features) that are most relevant to the classification task. The approach uses heavy-tailed priors to promote sparsity in the utilization of both basis functions and features; these priors act as regularizers for the likelihood function that rewards good classification on the training data. We derive an expectation-maximization (EM) algorithm to efficiently compute a maximum a posteriori (MAP) point estimate of the various parameters. The algorithm is an extension of recent state-of-the-art sparse Bayesian classifiers, which in turn can be seen as Bayesian counterparts of support vector machines. Experimental comparisons using kernel classifiers demonstrate both parsimonious feature selection and excellent classification accuracy on a range of synthetic and benchmark data sets.

Algorithms↗
PubMed2004

A decision tree--based method for the differential diagnosis of Aortic Stenosis from Mitral Regurgitation using heart sounds.

BACKGROUND: New technologies like echocardiography, color Doppler, CT, and MRI provide more direct and accurate evidence of heart disease than heart auscultation. However, these modalities are costly, large in size and operationally complex and therefore are not suitable for use in rural areas, in homecare and generally in primary healthcare set-ups. Furthermore the majority of internal medicine and cardiology training programs underestimate the value of cardiac auscultation and junior clinicians are not adequately trained in this field. Therefore efficient decision support systems would be very useful for supporting clinicians to make better heart sound diagnosis. In this study a rule-based method, based on decision trees, has been developed for differential diagnosis between "clear" Aortic Stenosis (AS) and "clear" Mitral Regurgitation (MR) using heart sounds. METHODS: For the purposes of our experiment we used a collection of 84 heart sound signals including 41 heart sound signals with "clear" AS systolic murmur and 43 with "clear" MR systolic murmur. Signals were initially preprocessed to detect 1st and 2nd heart sounds. Next a total of 100 features were determined for every heart sound signal and relevance to the differentiation between AS and MR was estimated. The performance of fully expanded decision tree classifiers and Pruned decision tree classifiers were studied based on various training and test datasets. Similarly, pruned decision tree classifiers were used to examine their differentiation capabilities. In order to build a generalized decision support system for heart sound diagnosis, we have divided the problem into sub problems, dealing with either one morphological characteristic of the heart-sound waveform or with difficult to distinguish cases. RESULTS: Relevance analysis on the different heart sound features demonstrated that the most relevant features are the frequency features and the morphological features that describe S1, S2 and the systolic murmur. The results are compatible with the physical understanding of the problem since AS and MR systolic murmurs have different frequency contents and different waveform shapes. On the contrary, in the diastolic phase there is no murmur in both diseases which results in the fact that the diastolic phase signals cannot contribute to the differentiation between AS and MR. We used a fully expanded decision tree classifier with a training set of 34 records and a test set of 50 records which resulted in a classification accuracy (total corrects/total tested) of 90% (45 correct/50 total records). Furthermore, the method proved to correctly classify both AS and MR cases since the partial AS and MR accuracies were 91.6% and 88.5% respectively. Similar accuracy was achieved using decision trees with a fraction of the 100 features (the most relevant). Pruned Differentiation decision trees did not significantly change the classification accuracy of the decision trees both in terms of partial classification and overall classification as well. DISCUSSION: Present work has indicated that decision tree algorithms decision tree algorithms can be successfully used as a basis for a decision support system to assist young and inexperienced clinicians to make better heart sound diagnosis. Furthermore, Relevance Analysis can be used to determine a small critical subset, from the initial set of features, which contains most of the information required for the differentiation. Decision tree structures, if properly trained can increase their classification accuracy in new test data sets. The classification accuracy and the generalization capabilities of the Fully Expanded decision tree structures and the Pruned decision tree structures have not significant difference for this examined sub-problem. However, the generalization capabilities of the decision tree based methods were found to be satisfactory. Decision tree structures were tested on various training and test data set and the classification accuracy was found to be consistently high.

Algorithms↗
PubMed1998

Detection of suspected malignant patterns in three-dimensional magnetic resonance breast images.

In this article, a Boolean Neural Network (BNN) is used for the detection of suspected malignant regions in 3D breast magnetic resonance (MR) images. The BNN is characterized by fast learning and classification, guaranteed convergence, and simple, integer weight calculations. The BNN learning algorithm is incremental, which allows the addition and deletion of training patterns without unlearning those already learned. The incremental learning algorithm automatically reduces the training set and trains the network only with those examples estimated to be useful. The architecture is suitable for parallel hardware implementation using available Very Large Scale Integration (VLSI) technology. The BNN was trained by using a set of malignant, benign, and false-positive patterns, extracted by experts, from selected MR studies, by using an incremental learning algorithm. After training, the network was tested by means of a consistency checking test, cross validation techniques, and patterns from actual MR breast images. During the consistency test, the BNN was tested by using the same patterns used for training. The BNN classification accuracy in this case was 99.75%, proving the ability of the BNN to select useful patterns from the training set. Then, a leave one out cross-validation (LOOCV) test was done by using patterns from the training set and the classification accuracy was 90%. Next, an extended training set was created by shifting the original patterns in different directions. A cross-validation test was then performed by dividing the set of patterns into a training and a test set. Classification accuracy was compared to the nearest neighbor classifier. Results showed that the BNN achieved an average of 77% classification accuracy while requiring only 34% of the original training set. On the other hand, the nearest neighbor classifier achieved an accuracy of 57.9% while retaining the whole training set. Another test using actual MR slices different from the training set was done and results compared favorably to a radiologist's findings. Test results show the BNN's capability to detect suspected malignant regions in 3D MR images of the breast. The proposed BNN architecture can save the radiologist a great deal of time browsing MR slices searching for suspected malignancies.

Algorithms↗
PubMed1990

Cluster analysis for automatic image segmentation in dynamic scintigraphies.

An original and entirely automatic algorithm is proposed to select regions of interest (ROIs) on dynamic scintigrams. This algorithm is based on factor analysis and on cluster analysis. It consists of first extracting the orthogonal factor images of the series using factor analysis of correspondence. These factor images are then automatically segmented in ROIs using a hierarchical ascendant classification procedure. The distance used for the classification is the 'minimum added intra-class variance' distance. This algorithm has been implemented on a fast computer dedicated to nuclear medicine (Nodecrest Micas V system). The time of calculation on 1000 pixels from 40 images is less than 5 min when three factor images are used. This algorithm is validated using a numerical phantom and is illustrated using renal (99Tcm DTPA) and cardiac (equilibrium gated angiography) dynamic scintigraphies. The results show that the algorithm is able to recognize the bladder, the renal cavities and the renal parenchyma on the renal series, and the ventricules and the atria on the cardiac series.

Algorithms↗
PubMed2004

[Retinal image analysis to detect lesions associated with diabetic retinopathy].

PURPOSE: Diabetic retinopathy is a leading cause of vision loss in developed countries. Regular diabetic retinal eye screenings are needed to detect early signs of retinopathy, so that appropriate treatments can be rendered to prevent blindness. Digital imaging is becoming available as a means of screening for diabetic retinopathy. However, with the large number of patients undergoing screenings, medical professionals require a tremendous amount of time and effort in order to analyse and diagnose the fundus photographs. Our aim is to develop an automatic algorithm using digital image analysis for detecting these early lesions from retinal images. METHODS: An automatic method to detect hard exudates, a lesion associated with diabetic retinopathy, is proposed. The algorithm is based on their colour, using a statistical classification, and their sharp edges, applying an edge detector, to localise them. RESULTS: A sensitivity of 79.62% with a mean number of 3 false positives per image is obtained in a database of 20 retinal images with variable colour, brightness and quality. It can also be seen that the number of the false negative cases increases when the hard exudates were very close to the vessel tree. CONCLUSION: The long term goal of the project is to automate the screening for diabetic retinopathy with retinal images. We have described the preliminary development of a tool to provide automatic analysis of digital fundus photographs to localise hard exudates. Future work will address the issue of improving the obtained results and also will focus on detecting other lesions.

Algorithms↗
PubMed2006

ICD-10-CA/CCI coding algorithms for defining clinical variables to assess outcome after aortic and mitral valve replacement surgery.

Implementation of the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Canada (ICD-10-CA) and the Canadian Classification of Interventions (CCI) coding system presents challenges for using Canadian administrative data. Thus, a multi-step process was conducted to develop ICD-10-CA/CCI coding algorithms to define nine comorbidities and three procedures. These clinical variables have been used in ICD-9-CM data for risk adjustment in assessment of outcomes after aortic and mitral valve replacement surgery. Among patients included in the ICD-9-CM data during 1999 and 2001 and in the ICD-10-CA/CCI data during 2002 and 2003 in a Canadian Health Region, frequencies of the nine comorbidities and the three procedures remained generally similar across databases. The newly developed ICD-10-CA/CCI and previous ICD-9-CM coding algorithms are comparable in detecting these clinical variables. However, performance of ICD-10-CA/CCI coding algorithms in risk adjustment should be evaluated in a larger database.

Algorithms↗
PubMed2006

Kernel methods for melanoma recognition.

Skin cancer is a spreading disease in the western world. Early detection and treatment are crucial for improving the patient survival rate. In this paper we present two algorithms for computer assisted diagnosis of melanomas. The first is the support vector machines algorithm, a state-of-the-art large margin classifier, which has shown remarkable performances on object recognition and categorization problems. The second method, spin glass-Markov random fields, combines results of statistical physics of spin glasses with Markov random fields. We compared the two approaches using color histograms as features. We benchmarked our methods with another algorithm presented in the literature, which uses a sophisticated segmentation technique and a set of features especially designed for melanoma recognition. To our knowledge, this algorithm represents the state of the art on skin lesions classification. We show with extensive experiments that the support vector machines approach outperforms the existing method and, on two classes out of three, it achieves performances comparable to those obtained by expert clinicians.

Algorithms↗
PubMed2005

An intelligent remote monitoring system for artificial heart.

A web-based database system for intelligent remote monitoring of an artificial heart has been developed. It is important for patients with an artificial heart implant to be discharged from the hospital after an appropriate stabilization period for better recovery and quality of life. Reliable continuous remote monitoring systems for these patients with life support devices are gaining practical meaning. The authors have developed a remote monitoring system for this purpose that consists of a portable/desktop monitoring terminal, a database for continuous recording of patient and device status, a web-based data access system with which clinicians can access real-time patient and device status data and past history data, and an intelligent diagnosis algorithm module that noninvasively estimates blood pump output and makes automatic classification of the device status. The system has been tested with data generation emulators installed on remote sites for simulation study, and in two cases of animal experiments conducted at remote facilities. The system showed acceptable functionality and reliability. The intelligence algorithm also showed acceptable practicality in an application to animal experiment data.

Decision Support Systems, Clinical↗
PubMed2002

Supervised clustering of genes.

BACKGROUND: We focus on microarray data where experiments monitor gene expression in different tissues and where each experiment is equipped with an additional response variable such as a cancer type. Although the number of measured genes is in the thousands, it is assumed that only a few marker components of gene subsets determine the type of a tissue. Here we present a new method for finding such groups of genes by directly incorporating the response variables into the grouping process, yielding a supervised clustering algorithm for genes. RESULTS: An empirical study on eight publicly available microarray datasets shows that our algorithm identifies gene clusters with excellent predictive potential, often superior to classification with state-of-the-art methods based on single genes. Permutation tests and bootstrapping provide evidence that the output is reasonably stable and more than a noise artifact. CONCLUSIONS: In contrast to other methods such as hierarchical clustering, our algorithm identifies several gene clusters whose expression levels clearly distinguish the different tissue types. The identification of such gene clusters is potentially useful for medical diagnostics and may at the same time reveal insights into functional genomics.

Algorithms↗
PubMedDate not supplied

[Classical and non-classical taxonomy: where does the boundary pass?].

Rise of non-classical science during XX century had certain influence upon development of biological taxonomy. Scientific pluralism (especially normative naturalism of Laudan), contrary to positivist and early post-positivist treatments, made taxonomy acknowledged scientific discipline of its own right. The present state of some schools of taxonomy makes it possible to consider them as a part of non-classical science and constituting the non-classical taxonomy. The latter is characterized by the following most important features. Ontological substantiation of both classificatory approaches and particular classifications is requested which invalidates such formal approaches as nominalistic and phenetic (numerical) schools. This substantiation takes a form of content-wise background preferably causal models which include certain axioms and presumptions about taxonomic diversity being studied, together with its causes, and thus define initial conditions of classificatory procedures. From this viewoint, phylogenetic classificatory approach is the most developed part of non-classical taxonomy. The entire taxonomic diversity is structured into several aspects of different levels of generality, each being outlined by a particular consideration aspect. The latter makes personal knowledge constituting an irremovable part of any scientific statement about taxonomic diversity, thus opposition of "objectively" and "subjectively" elaborated classifications becomes vague. Interrelation of various species concepts corresponding to its different consideration aspects is described by uncertainty relation principle. Classificatory algorithms are to be compatible with the conditions of a background model to ensure particular classifications obtained by their means are interpretable within the same model: this is provided by the correspondence principle. Classification is considered as a taxonomic hypothesis, i.e. a conjectural judgement about structure of particular fragment of taxonomic diversity considered within given consideration aspect; wich is to be forwarded and tested according to certain rules. Recognition of different aspects of taxonomic diversity makes it "legal" to elaborate several classifications of equal status, each reflecting a particular aspect of a fixed fragment of that diversity. This viewpoint makes classical ideas of the "ultimate" Natural (whatever might be its definition) or the best reference systems futile. In general, any pretension of an approach to be "the best" in reflecting taxonomic divesrity is contr-productive. Instead, elaboration of particular spectra of complementary classifications becomes the main task of non-classical taxonomy which describes in sum the entire taxonomic diversity. So, not opposition but correct mutual interpretation of such classifications and uniting them into the comprehensive picture of taxonomic diversity become focal points of non-classical taxonomy.

Biodiversity↗
PubMed2005

A novel ensemble machine learning for robust microarray data classification.

Microarray data analysis and classification has demonstrated convincingly that it provides an effective methodology for the effective diagnosis of diseases and cancers. Although much research has been performed on applying machine learning techniques for microarray data classification during the past years, it has been shown that conventional machine learning techniques have intrinsic drawbacks in achieving accurate and robust classifications. This paper presents a novel ensemble machine learning approach for the development of robust microarray data classification. Different from the conventional ensemble learning techniques, the approach presented begins with generating a pool of candidate base classifiers based on the gene sub-sampling and then the selection of a sub-set of appropriate base classifiers to construct the classification committee based on classifier clustering. Experimental results have demonstrated that the classifiers constructed by the proposed method outperforms not only the classifiers generated by the conventional machine learning but also the classifiers generated by two widely used conventional ensemble learning methods (bagging and boosting).

Algorithms↗
PubMed2005

Development of a support vector machine-based image analysis system for assessing the thyroid nodule malignancy risk on ultrasound.

An SVM-based image analysis system was developed for assessing the malignancy risk of thyroid nodules. Ultrasound images of 120 cytology confirmed thyroid nodules (78 low-risk and 42 high-risk of malignancy) were manually segmented by a physician using a custom developed software in C++. From each nodule, 40 textural features were automatically calculated and were used with the SVM algorithm in the design of the image analysis system. Highest classification accuracy was 96.7%, misdiagnosing two high-risk and two low-risk thyroid nodules. The proposed system may be of value to physicians as a second opinion tool for avoiding unnecessary invasive procedures.

Algorithms↗
PubMed2006

New actigraph for long-term tremor recording.

A new method of movement analysis is validated, allowing an actigraph to discriminate tremor from other movements and store duration and intensity measures of both movement types. For algorithm optimization, wrist acceleration was recorded in nine controls and nine Parkinson's disease patients, while simultaneously rating their observed tremor minute by minute on item 20 of the Unified Parkinson's Disease Rating Scale. An optimization procedure to minimize false positives in controls while maximizing tremor detection in patients resulted in false positive tremor classification in 2.4% +/- 2.5% of the movement time of control subjects (range, 0%-7%), while providing tremor classification in 82.1% +/- 15.4% of the movement time in patients (range, 55%-100%), correlating r = 0.93 with their averaged observed tremor score. A second, generalizability study showed that application of the optimized algorithm resulted in accurate classification of 71% +/- 14% of the observed tremor time (range, 46%-90%) in another 9 patients and in a false positive classification in only 0.5% +/- 0.8% of the time in another 10 controls (range, 0%-2.4%). The commercial availability of this actigraph now for the first time makes it possible to investigate tremor fluctuations over several weeks. An example is given of how long-term monitoring can be of use in evaluation of symptom management.

Aged↗
PubMed1994

Pattern recognition and classification of images of biological macromolecules using artificial neural networks.

The goal of this work was to analyze an image data set and to detect the structural variability within this set. Two algorithms for pattern recognition based on neural networks are presented, one that performs an unsupervised classification (the self-organizing map) and the other a supervised classification (the learning vector quantization). The approach has a direct impact in current strategies for structural determination from electron microscopic images of biological macromolecules. In this work we performed a classification of both aligned but heterogeneous image data sets as well as basically homogeneous but otherwise rotationally misaligned image populations, in the latter case completely avoiding the typical reference dependency of correlation-based alignment methods. A number of examples on chaperonins are presented. The approach is computationally fast and robust with respect to noise. Programs are available through ftp.

Algorithms↗