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 1,711 records · Page 95Linked to original sources

PubMed2000

A new approach to applying feedforward neural networks to the prediction of musculoskeletal disorder risk.

A new and improved method to feedforward neural network (FNN) development for application to data classification problems, such as the prediction of levels of low-back disorder (LBD) risk associated with industrial jobs, is presented. Background on FNN development for data classification is provided along with discussions of previous research and neighborhood (local) solution search methods for hard combinatorial problems. An analytical study is presented which compared prediction accuracy of a FNN based on an error-back propagation (EBP) algorithm with the accuracy of a FNN developed by considering results of local solution search (simulated annealing) for classifying industrial jobs as posing low or high risk for LBDs. The comparison demonstrated superior performance of the FNN generated using the new method. The architecture of this FNN included fewer input (predictor) variables and hidden neurons than the FNN developed based on the EBP algorithm. Independent variable selection methods and the phenomenon of 'overfitting' in FNN (and statistical model) generation for data classification are discussed. The results are supportive of the use of the new approach to FNN development for applications to musculoskeletal disorders and risk forecasting in other domains.

Back Injuries↗

PubMed2004

Low dimensional adaptive texture feature vectors from class distance and class difference matrices.

In many popular texture analysis methods, second or higher order statistics on the relation between pixel gray level values are stored in matrices. A high dimensional vector of predefined, nonadaptive features is then extracted from these matrices. Identifying a few consistently valuable features is important, as it improves classification reliability and enhances our understanding of the phenomena that we are modeling. Whatever sophisticated selection algorithm we use, there is a risk of selecting purely coincidental "good" feature sets, especially if we have a large number of features to choose from and the available data set is limited. In a unified approach to statistical texture feature extraction, we have used class distance and class difference matrices to obtain low dimensional adaptive feature vectors for texture classification. We have applied this approach to four relevant texture analysis methods. The new adaptive features outperformed the classical features when applied to the most difficult set of 45 Brodatz texture pairs. Class distance and difference matrices also clearly illustrated the difference in texture between cell nucleus images from two different prognostic classes of early ovarian cancer. For each of the texture analysis methods, one adaptive feature contained most of the discriminatory power of the method.

Algorithms↗

PubMed1998

Ventriculoatrial conduction metrics for classification of ventricular tachycardia with 1:1 retrograde conduction in dual-chamber sensing implantable cardioverter defibrillators.

The introduction of dual-chamber sensing in implantable cardioverter defibrillators (ICDs) has greatly reduced the incidence of false detection due to supraventricular tachycardias. The remaining arrhythmias which serve to confound classification are supraventricular tachycardias (SVTs) with 1:1 anterograde conduction and ventricular tachycardias (VT) with 1:1 retrograde conduction. An algorithm has been designed and tested (28 patients) which employs ventriculoatrial (VA) conduction measurements to separate 1:1 VTs from 1:1 SVTs. A study was conducted to assess realistic VA interval boundaries for classification of arrhythmias with 1:1 retrograde atrial conduction. Intracardiac atrial and ventricular recordings of 7 passages of VT with retrograde conduction, 12 passages of atrioventricular nodal reentrant tachycardia (AVNRT), 3 passages of atrial tachycardia (AT), 8 passages of sinus tachycardia (ST), and 2 passages of orthodromic reentrant tachycardia (ORT) were analyzed. Automated real-time atrial and ventricular waveform recognition was performed on each passage and VA intervals were measured. Separation of VT with retrograde conduction from other 1:1 supraventricular tachycardias was effected by imposing discrete VA interval boundaries. VA boundaries of 80 ms to 234 ms classified 1:1 VT with 100% sensitivity (SENS) and 80% specificity (SPEC). In addition, the lower boundary completely classified AVNRT with 100% SENS and 100% SPEC, and all passages of ST were contained above the upper boundary. These findings could be of importance in algorithms for next-generation implantable cardioverter defibrillators which include two-chamber (atrial and ventricular) sensing and two-chamber interval measurements.

Algorithms↗

PubMed2004

Segmentation of 4D cardiac MR images using a probabilistic atlas and the EM algorithm.

In this paper an automatic atlas-based segmentation algorithm for 4D cardiac MR images is proposed. The algorithm is based on the 4D extension of the expectation maximisation (EM) algorithm. The EM algorithm uses a 4D probabilistic cardiac atlas to estimate the initial model parameters and to integrate a priori information into the classification process. The probabilistic cardiac atlas has been constructed from the manual segmentations of 3D cardiac image sequences of 14 healthy volunteers. It provides space and time-varying probability maps for the left and right ventricles, the myocardium, and background structures such as the liver, stomach, lungs and skin. In addition to using the probabilistic cardiac atlas as a priori information, the segmentation algorithm incorporates spatial and temporal contextual information by using 4D Markov Random Fields. After the classification, the largest connected component of each structure is extracted using a global connectivity filter which improves the results significantly, especially for the myocardium. Validation against manual segmentations and computation of the correlation between manual and automatic segmentation on 249 3D volumes were calculated. We used the 'leave one out' test where the image set to be segmented was not used in the construction of its corresponding atlas. Results show that the procedure can successfully segment the left ventricle (LV) (r = 0.96), myocardium (r = 0.92) and right ventricle (r = 0.92). In addition, 4D images from 10 patients with hypertrophic cardiomyopathy were also manually and automatically segmented yielding a good correlation in the volumes of the LV (r = 0.93) and myocardium (0.94) when the atlas constructed with volunteers is blurred.

Algorithms↗

PubMed2002

Emergency department asthma: compliance with an evidence-based management algorithm.

BACKGROUND: Recent surveys in emergency medicine departments show inadequacies in many aspects of acute asthma management. OBJECTIVE: The objective of this study was to evaluate the adherence to management algorithm for acute asthma in the emergency department which focused on evidence-based treatment steps rather than documentation and patient classification. METHODS: A retrospective audit of consecutive adult patients with acute exacerbations of asthma in the emergency medicine department of a university hospital managed in the context of a clinical algorithm. RESULTS: We collated information from 344 episodes of acute asthma (94% of total) over a 4-month period. The first-line treatment was nebulised bronchodilators in 97%, combination of salbutamol and ipratropium bromide in 93% and the combination in recommended dosages in 87%. Systemic corticosteroid treatment was administered to 82% of patients. A further course of systemic corticosteroid was prescribed at discharge for 94% of patients. Overall, 93% of patients received some form of systemic corticosteroid treatment. The admission rate was 35.2%, and was significantly higher in women and the elderly. Of those who were admitted, 46.2% received > or = 3 nebulised treatments and 69% received intravenous hydrocortisone. CONCLUSIONS: In the management of acute asthma, we found excellent compliance with specific treatment steps based upon clinical evidence. However, adherence to second-line treatment was less satisfactory. Nevertheless, whenever second-line treatment was complied with, reasonable outcomes were achieved. It may be more appropriate to emphasise evidence-based treatment rather than extensive documentation.

Acute Disease↗

PubMed2002

A method for quantifying atrial fibrillation organization based on wave-morphology similarity.

A new method for quantifying the organization of single bipolar electrograms recorded in the human atria during atrial fibrillation (AF) is presented. The algorithm relies on the comparison between pairs of local activation waves (LAWs) to estimate their morphological similarity, and returns a regularity index (rho) which measures the extent of repetitiveness over time of the detected activations. The database consisted of endocardial data from a multipolar basket catheter during AF and intraatrial recordings during atrial flutter. The index showed maximum regularity (rho = 1) for all atrial flutter episodes and decreased significantly when increasing AF complexity as defined by Wells (type I: rho = 0.75 +/- 0.23; type II: rho = 0.35 +/- 0.11; type III: rho = 0.15 +/- 0.08; P < 0.01). The ability to distinguish different AF episodes was assessed by designing a classification scheme based on a minimum distance analysis, obtaining an accuracy of 85.5%. The algorithm was able to discriminate among AF types even in presence of few depolarizations as no significant rho changes were observed by reducing the signal length down to include five LAWs. Finally, the capability to detect transient instances of AF complexity and to map the local regularity over the atrial surface was addressed by the dynamic and multisite evaluation of rho, suggesting that our algorithm could improve the understanding of AF mechanisms and become useful for its clinical treatment.

Algorithms↗

PubMed2005

Functional magnetic resonance imaging activation detection: fuzzy cluster analysis in wavelet and multiwavelet domains.

PURPOSE: To present novel feature spaces, based on multiscale decompositions obtained by scalar wavelet and multiwavelet transforms, to remedy problems associated with high dimension of functional magnetic resonance imaging (fMRI) time series (when they are used directly in clustering algorithms) and their poor signal-to-noise ratio (SNR) that limits accurate classification of fMRI time series according to their activation contents. MATERIALS AND METHODS: Using randomization, the proposed method finds wavelet/multiwavelet coefficients that represent the activation content of fMRI time series and combines them to define new feature spaces. Using simulated and experimental fMRI data sets, the proposed feature spaces are compared to the cross-correlation (CC) feature space and their performances are evaluated. In these studies, the false positive detection rate is controlled using randomization. To compare different methods, several points of the receiver operating characteristics (ROC) curves, using simulated data, are estimated and compared. RESULTS: The proposed features suppress the effects of confounding signals and improve activation detection sensitivity. Experimental results show improved sensitivity and robustness of the proposed method compared to the conventional CC analysis. CONCLUSION: More accurate and sensitive activation detection can be achieved using the proposed feature spaces compared to CC feature space. Multiwavelet features show superior detection sensitivity compared to the scalar wavelet features.

Algorithms↗

PubMed2006

A study of structural and parametric learning in XCS.

The performance of a learning classifier system is due to its two main components. First, it evolves new structures by generating new rules in a genetic process; second, it adjusts parameters of existing rules, for example rule prediction and accuracy, in an evaluation step, which is not only important for applying the rules, but also for the genetic process. The two components interleave and in the case of XCS drive the population toward a minimal, fit, non-overlapping population. In this work we attempt to gain new insights as to the relative contributions of the two components. We find that the genetic component has an additional role when using the train/test approach which is not present in online learning. We compare XCS to a system in which the rule set is restricted to the initial random population (XCS-NGA, that is, XCS No Genetic Algorithm). For small Boolean functions we can give XCS-NGA all possible rules of a particular condition length. In online learning, XCS-NGA can, given sufficiently many rules, achieve a surprisingly high classification accuracy, comparable to that of XCS. In a train/test approach, however, XCS generalises better than XCS-NGA and there seem to be limitations of XCS-NGA which cannot be overcome simply by increasing the population size. This illustrates that the requirements of a function approximator tend to differ between reinforcement learning (which is typically online) and concept learning (which is typically train/test).

Algorithms↗

PubMed1994

A method for the quantification of synchrony and oscillatory properties of neuronal activity.

Interactions between neurones can be analysed by simultaneously recording from several cells and computing correlation functions between the respective activities. Recent studies have revealed that neuronal responses are often synchronous and exhibit an oscillatory temporal structure. These two properties are commonly assessed together from correlation functions. In order to evaluate these variables independently a method was devised for the quantification of a generalized Gabor function that was fitted to the correlograms. The performance of the method was tested on a large data set from cat area 17 and its stability was examined with respect to its dependence on the number of free parameters. The results demonstrate that the proposed fitting algorithm is robust in that it is rather independent of starting conditions and converges to optimal fits even with different settings of free variables. The fitted correlation function allow for an automatic and independent classification of synchrony on the one hand and oscillatory firing patterns on the other.

Algorithms↗

PubMed2006

Source analysis of epileptic discharges using multiple signal classification analysis.

Multiple signal classification is an alternative to the traditional dipole fitting source analysis methods. Our aim was to assess the clinical usefulness of this algorithm and to compare the localization of the epileptiform electroencephalography discharges with the regions of altered cerebral blood flow in 10 patients with complex partial seizures undergoing preoperative investigation. We performed multiple signal classification analysis of ictal and interictal discharges, and registered single-photon emission computed tomography. Localization of the ictal, but not the interictal discharges, as determined by multiple signal classification analysis was consistent with the regions showing perfusion changes on the single-photon emission computed tomography. Multiple signal classification analysis is a promising tool in localizing foci in patients with complex partial seizures and may contribute to the preoperative evaluation.

Adult↗

PubMed1991

New approaches in genome analysis by pulsed-field gel electrophoresis: application to the analysis of Pseudomonas species.

A general method for the evaluation of macrorestriction fragment patterns is presented and its applicability to the taxonomy of bacteria is demonstrated for 32 Pseudomonas species. Strains were differentiated at the species and subspecies level by genome size and macrorestriction fragment fingerprints of the chromosome that had been separated on pulsed-field gels. The relatedness of bacteria was ascertained from the similarity of AsnI, DraI, SpeI, SspI or XbaI fragment patterns. In general, the dendrograms calculated from the genome fingerprints corresponded with the phylogenetic classification obtained from phenotypic marker or nucleic acid hybridization analysis, but several exceptions were noted. The techniques and algorithms presented herein are generally applicable to the genome analysis of bacteria, lower eukaryotes, and DNA fragments cloned in yeast artificial chromosomes.

Bacteria↗

PubMed2004

Evaluation of an algorithm for integrated management of childhood illness in an area of Vietnam with dengue transmission.

OBJECTIVES: To determine whether nurses, using the WHO/UNICEF algorithm for integrated management of childhood illness (IMCI), modified to include dengue infection, satisfactorily classified children in an area endemic for dengue haemorrhagic fever (DHF). METHODS: Nurses assessed and classified, using the modified IMCI algorithm, a systematic sample of 1250 children aged 2 months to 10 years (n = 1250) presenting to a paediatric hospital in Dong Nai Province, Vietnam. Their classification was compared with that of a paediatrician, blind to the result of the nurses' assessment, which could be modified in the light of simple investigations, e.g. dengue serology. RESULTS: In children aged 2-59 months (n = 859), the nurses were able to classify, using the modified chart, the presenting illness in >99% of children and found more than one classification in 70%. For the children with pneumonia, diarrhoea, dengue shock syndrome, severe DHF and severe disease requiring urgent admission, the nurse's classification was >60% sensitive and >85% specific compared with that of the paediatrician. For the nurse's classification of DHF the specificity was 50-55% for the children <5 years and in children with definitive dengue serology. Alterations in the DHF algorithm improved specificity at the expense of sensitivity. CONCLUSION: Using the IMCI chart, nurses classified appropriately many of the major clinical problems in sick children <5 years in southern Vietnam. However, further modifications will be required in the fever section, particularly for dengue. The impact of using the IMCI chart in peripheral health stations remains to be evaluated.

Algorithms↗

PubMed2006

Coded Chief Complaints--automated analysis of free-text complaints.

OBJECTIVES: To describe a new chief-complaint categorization schema, the development of a computer text-parsing algorithm to automatically classify free-text chief complaints into this schema, and use of these coded chief complaints to describe the case mix of a community emergency department (ED). METHODS: Coded Chief Complaints for Emergency Department Systems (CCC-EDS) is a new and untested schema of 228 chief complaints, grouped within dimensions of type and system. A computerized text-parsing algorithm for automatically reading and classifying free-text chief complaints into 1 of these 228 coded chief complaints was developed by using a consecutive derivation sample of 46,602 patients who presented to a community teaching-hospital ED in 2004. Descriptive statistics included frequency of patients presenting with the 228 coded chief complaints; percentage of free-text complaints not categorizable by the CCC-EDS; and admission rate, age, and gender differences by chief complaint. RESULTS: In the derivation sample, the text-parsing algorithm classified 87.5% of 45,329 ED visits with non-null free-text chief complaints into 1 of 194 coded chief complaints. The text-parsing algorithm successfully classified 87.3% of the free-text chief complaints in a validation sample. The five most common coded chief complaints were Abdominal Pain (3,734 visits), Fever (2,234), Chest Pain (2,183), Breathing Difficulty (2,030), and Cuts-Lacerations (2,028). CONCLUSIONS: The CCC-EDS is a new comprehensive, granular, and useful classification schema for categorizing chief complaints in an ED. A CCC-EDS text-parsing algorithm successfully classified the majority of free-text chief complaints from an ED computer log. These coded chief complaints were used to describe the case mix of a community teaching-hospital ED.

Algorithms↗

PubMed1997

The integrated management of childhood illness in western Uganda.

Bringing together various disease-specific guidelines for sick children, WHO and UNICEF have developed an Integrated Management of Childhood Illness (IMCI) algorithm, one component of which (assess and classify) was tested in the outpatient department of a rural district hospital in western Uganda. Children aged 2-59 months were seen first by a Ugandan medical assistant trained in IMCI, and then evaluated by a medical officer. Sensitivity, specificity and positive predictive values were determined by comparing the IMCI classifications with a reference standard based on the medical officers' diagnoses and laboratory tests. Of the 1226 children seen, 69% were classified into more than one symptom category, 7% were not classified in any symptom category, 8% had a danger sign, and 16% were classified into a severe category, for which the IMCI approach recommended urgent hospital referral. Specificity for most classifications was good, though sensitivity and positive predictive values were variable. We conclude that the IMCI algorithm is an important advance in the primary care of sick children in developing countries.

Acute Disease↗

PubMed2002

Hair-MAP: a prototype automated system for forensic hair comparison and analysis.

This paper demonstrates the feasibility of the automation of forensic hair analysis and comparison task using neural network explanation systems (NNESs). Our system takes as input microscopic images of two hairs and produces a classification decision as to whether or not the hairs came from the same person. Hair images were captured using a NEXTDimension video board in a NEXTDimension color turbo computer, connected to a video camera. Image processing was done on an SGI indigo workstation. Each image is segmented into a number of pieces appropriate for classification of different features. A variety of image processing techniques are used to enhance this information. Use of wavelet analysis and the Haralick texture algorithm to pre-process data has allowed us to compress large amounts of data into smaller, yet representative data. Neural networks are then used for feature classification. Finally, statistical tests determine the degree of match between the resulting collection of hair feature vectors. An important issue in automation of any task in criminal investigations is the reliability and understandability of the resulting system. To address this concern, we have developed methods to facilitate explanation of neural network's behavior using a decision tree. The system was able to achieve a performance of 83% hair match accuracy, using 5 of the 21 morphological characteristics used by experts. This shows promise for the usefulness of a fuller scale system. While an automated system would not replace the expert, it would make the task easier by providing a means for pre-processing the large amount of data with which the expert must contend.

Algorithms↗

PubMed2004

Making a structured psychiatric diagnostic interview faithful to the nomenclature.

Psychiatric diagnostic interviews to be used in epidemiologic studies by lay interviewers have, since the 1970s, attempted to operationalize existing psychiatric nomenclatures. How to maximize the chances that they do so successfully has not previously been spelled out. In this article, the authors discuss strategies for each of the seven steps involved in writing, updating, or modifying a diagnostic interview and its supporting materials: 1) writing questions that match the nomenclature's criteria, 2) checking that respondents will be willing and able to answer the questions, 3) choosing a format acceptable to interviewers that maximizes accurate answering and recording of answers, 4) constructing a data entry and cleaning program that highlights errors to be corrected, 5) creating a diagnostic scoring program that matches the nomenclature's algorithms, 6) developing an interviewer training program that maximizes reliability, and 7) computerizing the interview. For each step, the authors discuss how to identify errors, correct them, and validate the revisions. Although operationalization will never be perfect because of ambiguities in the nomenclature, specifying methods for minimizing divergence from the nomenclature is timely as users modify existing interviews and look forward to updating interviews based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, and the International Classification of Diseases, Eleventh Revision.

Algorithms↗

PubMed2000

[Artificial neural network for the identification of infrared spectra].

An Artificial Neural Network(ANN) was used to identify unknown infrared spectra. The Neural Network consisted of three layers was trained by a back-propagation algorithm. In the first step of the experiment, the training set was pure spectra information, the neural network can only identify correctly the spectra without noise or with relatively low noise, in the second step, the training set was spectra information with relatively low noise, the identification results of the test sets was better than that of the first step. The results showed that artificial neural network can be used as a powerful tool in solving classification and identification problems.

Algorithms↗

PubMed2004

DemTect: a new, sensitive cognitive screening test to support the diagnosis of mild cognitive impairment and early dementia.

OBJECTIVES: To design a new, highly sensitive psychometric screening to identify patients with mild cognitive impairment (MCI) and patients with dementia in the early stages of the disease. METHODS: Five tasks were included in the DemTect: a word list, a number transcoding task, a word fluency task, digit span reverse, and delayed recall of the word list. The normation was performed with 145 healthy control subjects (CG). Furthermore, 97 MCI patients and 121 patients with possible Alzheimer's disease (AD) were tested with the DemTect and the MMSE. Classification rates for both tests were analysed. RESULTS: On the basis of the CG data, age-dependant transformation algorithms for the DemTect subtests were defined, and an education correction was provided for the total transformed score. The patient groups scored significantly below the CG in both the DemTect and the MMSE. Compared to the MMSE, classification rates of the DemTect were superior for both the MCI and the AD group, with high sensitivities of 80% and 100%, respectively. CONCLUSIONS: The DemTect is short (8-10 minutes), easy to administer, and its transformed total score (maximum 18) is independent of age and education. The DemTect helps in deciding whether cognitive performance is adequate for age (13-18 points), or whether MCI (9-12 points) or dementia (8 points or below) should be suspected.

Age Factors↗