Iterative expert-in-the-loop classification of sleep PSG recordings using a hierarchical clustering.

BACKGROUND: The classification of sleep signals is a subjective and time consuming task. A large number of automatic classifiers have been published in the past decade but a sleep community has no strong confidence to use them in clinical practice and still remains using a standard manual scoring according standardized rules. NEW METHOD: We developed a semi-supervised data-driven approach for objective and efficient evaluation of polysomnographic (PSG) data. The proposed algorithm finds a representative set of signal segments that are subsequently scored by a sleep neurologist. The remaining part of the recording is then automatically classified using these templates. RESULTS: The method was evaluated on 36 PSG recordings (18 chronic insomniacs, 18 healthy controls). We show a faster and objective evaluation of PSG data compared to the manual scoring that is over-performing automated classifiers (accuracy increases ∼14%). The classification results are comparable on both datasets. COMPARISON WITH EXISTING METHOD(S): The methodology that we propose has not yet been published in the area of sleep PSG data processing. The performance of our method is comparable to various published automated approaches (a typical published classification accuracy is ∼75-95%). The method allows the evaluation of PSG recordings in more general terms and across different recording devices and standards. CONCLUSIONS: The proposed solution is not based on a single-purpose rules or heuristics and training model is not trained on other patient's sleep recordings. The method is applicable to wide range of similar tasks and various types of physiological signals.

Outcome of resynchronization therapy on superficial and endocardial electrophysiological findings.

Cardiac resynchronization therapy (CRT) has proven efficacious in the treatment of patients with heart failure and dyssynchronous activation. Currently, we select suitable CRT candidates based on the QRS complex duration (QRSd) and morphology with left bundle branch block being the optimal substrate for resynchronization. To improve CRT response rates, recommendations emphasize attention to electrical parameters both before implant and after it. Therefore, we decided to study activation times before and after CRT on the body surface potential maps (BSPM) and to compare thus obtained results with data from electroanatomical mapping using the CARTO system. Total of 21 CRT recipients with symptomatic heart failure (NYHA II-IV), sinus rhythm, and QRSd >/=150 ms and 7 healthy controls were studied. The maximum QRSd and the longest and shortest activation times (ATmax and ATmin) were set in the BSPM maps and their locations on the chest were compared with CARTO derived time interval and site of the latest (LATmax) and earliest (LATmin) ventricular activation. In CRT patients, all these parameters were measured during both spontaneous rhythm and biventricular pacing (BVP) and compared with the findings during the spontaneous sinus rhythm in the healthy controls. QRSd was 169.7+/-12.1 ms during spontaneous rhythm in the CRT group and 104.3+/-10.2 ms after CRT (p<0.01). In the control group the QRSd was significantly shorter: 95.1+/-5.6 ms (p<0.01). There was a good correlation between LATmin(CARTO) and ATmin(BSPM). Both LATmin and ATmin were shorter in the control group (LATmin(CARTO) 24.8+/-7.1 ms and ATmin(BSPM) 29.6+/-11.3 ms, NS) than in CRT group (LATmin(CARTO) was 48.1+/-6.8 ms and ATmin(BSPM) 51.6+/-10.1 ms, NS). BVP produced shortening compared to the spontaneous rhythm of CRT recipients (LATmin(CARTO) 31.6+/-5.3 ms and ATmin(BSPM) 35.2+/-12.6 ms; p<0.01 spontaneous rhythm versus BVP). ATmax exhibited greater differences between both methods with higher values in BSPM: in the control group LATmax(CARTO) was 72.0+/-4.1 ms and ATmax (BSPM) 92.5+/-9.4 ms (p<0.01), in the CRT candidates LATmax(CARTO) reached only 106.1+/-6.8 ms whereas ATmax(BSPM) 146.0+/-12.1 ms (p<0.05), and BVP paced rhythm in CRT group produced improvement with LATmax(CARTO) 92.2+/-7.1 ms and ATmax(BSPM) 130.9+/-11.0 ms (p<0.01 before and during BVP). With regard to the propagation of ATmin and ATmax on the body surface, earliest activation projected most often frontally in all 3 groups, whereas projection of ATmax on the body surface was more variable. Our results suggest that compared to invasive electroanatomical mapping BSPM reflects well time of the earliest activation, however provides longer time-intervals for sites of late activation. Projection of both early and late activated regions of the heart on the body surface is more variable than expected, very likely due to changed LV geometry and interposed tissues between the heart and superficial ECG electrode.

Electrocardiographic outcome of resynchronization therapy.

Cardiac resynchronization therapy (CRT) has proven efficacious in reducing or even eliminating cardiac dyssynchrony and thus improving heart failure symptoms. However, quantification of mechanical dyssynchrony is still difficult and identification of CRT candidates is currently based just on the morphology and width of the QRS complex. As standard 12-lead ECG brings only limited information about the pattern of ventricular activation, we aimed to study changes produced by different pacing modes on the body surface potential maps (BSPM). Total of 12 CRT recipients with symptomatic heart failure (NYHA II-IV), sinus rhythm and QRS width >/=120 ms and 12 healthy controls were studied. Mapping system Biosemi (123 unipolar electrodes) was used for BSPM acquisition. Maximum QRS duration, longest and shortest activation times (ATmax and ATmin) and dispersion of QT interval (QTd) were measured and/or calculated during spontaneous rhythm, single-site right- and left-ventricular pacing and biventricular pacing with ECHO-optimized AV delay. Moreover we studied the impact of CRT on the locations of the early and late activated regions of the heart. The average values during the spontaneous rhythm in the group of patients with dyssynchrony (QRS 140.5+/-10.6 ms, ATmax 128.1+/-10.1 ms, ATmin 31.8+/-6.7 ms and QTd 104.3+/-24.7 ms) significantly differed from those measured in the control group (QRS 93.0+/-10.0 ms, ATmax 79.1+/-3.2 ms, ATmin 24.4+/-1.6 ms and QTd 43.6+/-10.7 ms). Right ventricular pacing (RVP) improved significantly only ATmax [111.2+/-10.6 ms (p<0.05)] but no other measured parameters. Left ventricular pacing (LVP) succeeded in improvement of all parameters [QRS 105.1+/-8.0 ms (p<0.01), ATmax 103.7+/-7.1 ms (p<0.01), ATmin 20.2+/-3.7 ms (p<0.01) and QTd 52.0+/-9.4 ms (p<0.01)]. Biventricular pacing (BVP) showed also a beneficial effect in all parameters [QRS 121.3+/-8.9 ms (p<0.05), ATmax 114.3+/-8.2 ms (p<0.05), ATmin 22.0+/-4.1 ms (p<0.01) and QTd 49.8+/-10.0 ms (p<0.01)]. Our results proved beneficial outcome of LVP and BVP in evaluated parameters (what seems to be important particularly in the case of activation times) and revealed a complete return of activation times to normal distribution when using these CRT modalities.

Combining latent class analysis labeling with multiclass approach for fetal heart rate categorization.

The most common approach to assess fetal well-being during delivery is monitoring of fetal heart rate and uterine contractions-the cardiotocogram (CTG). Nevertheless, 40 years since the introduction of CTG to clinical practice, its evaluation is still challenging with high inter- and intra-observer variability. Therefore the development of more objective methods has become an issue of major importance in the field. Unlike the usually proposed approaches to assign classes for classification methods that rely either on biochemical parameters (e.g. pH value) or a simple aggregation of expert judgment, this work investigates the use of an alternative labeling system using latent class analysis (LCA) along with an ordinal classification scheme. The study is performed on a well-documented open-access database, where nine expert obstetricians provided CTG annotations. The LCA is proposed here to produce more objective class labels while the ordinal classification aims to explore the natural ordering, and representation of increased severity, for obtaining the final results. The results are promising suggesting that more effort should be put into this proposed approach.

Measuring the food and built environments in urban centres: reliability and validity of the EURO-PREVOB Community Questionnaire.

OBJECTIVES: The authors designed an instrument to measure objectively aspects of the built and food environments in urban areas, the EURO-PREVOB Community Questionnaire, within the EU-funded project 'Tackling the social and economic determinants of nutrition and physical activity for the prevention of obesity across Europe' (EURO-PREVOB). This paper describes its development, reliability, validity, feasibility and relevance to public health and obesity research. STUDY DESIGN: The Community Questionnaire is designed to measure key aspects of the food and built environments in urban areas of varying levels of affluence or deprivation, within different countries. The questionnaire assesses (1) the food environment and (2) the built environment. METHODS: Pilot tests of the EURO-PREVOB Community Questionnaire were conducted in five to 10 purposively sampled urban areas of different socio-economic status in each of Ankara, Brno, Marseille, Riga, and Sarajevo. Inter-rater reliability was compared between two pairs of fieldworkers in each city centre using three methods: inter-observer agreement (IOA), kappa statistics, and intraclass correlation coefficients (ICCs). RESULTS: Data were collected successfully in all five cities. Overall reliability of the EURO-PREVOB Community Questionnaire was excellent (inter-observer agreement (IOA) > 0.87; intraclass correlation coefficients (ICC)s > 0.91 and kappa statistics > 0.7. However, assessment of certain aspects of the quality of the built environment yielded slightly lower IOA coefficients than the quantitative aspects. CONCLUSIONS: The EURO-PREVOB Community Questionnaire was found to be a reliable and practical observational tool for measuring differences in community-level data on environmental factors that can impact on dietary intake and physical activity. The next step is to evaluate its predictive power by collecting behavioural and anthropometric data relevant to obesity and its determinants.

Automatic evaluation of intrapartum fetal heart rate recordings: a comprehensive analysis of useful features.

Cardiotocography is the monitoring of fetal heart rate (FHR) and uterine contractions (TOCO), used routinely since the 1960s by obstetricians to detect fetal hypoxia. The evaluation of the FHR in clinical settings is based on an evaluation of macroscopic morphological features and so far has managed to avoid adopting any achievements from the HRV research field. In this work, most of the features utilized for FHR characterization, including FIGO, HRV, nonlinear, wavelet, and time and frequency domain features, are investigated and assessed based on their statistical significance in the task of distinguishing the FHR into three FIGO classes. We assess the features on a large data set (552 records) and unlike in other published papers we use three-class expert evaluation of the records instead of the pH values. We conclude the paper by presenting the best uncorrelated features and their individual rank of importance according to the meta-analysis of three different ranking methods. The number of accelerations and decelerations, interval index, as well as Lempel-Ziv complexity and Higuchi's fractal dimension are among the top five features.

Discrimination of endocardial electrogram disorganization using a signal regularity analysis.

Measures from the theory of nonlinear dynamics were applied on complex fractionated atrial electrograms (CFAEs) in order to characterize their physiological dynamic behavior. The results were obtained considering 113 short term atrial electrograms (A-EGMs) which were annotated by three experts into four classes of fractionation according to A-EGMs signal regularity. The following measures were applied on A-EGM signals: General Correlation Dimension, Approximate Entropy, Detrended Fluctuation Analysis, Lempel-Ziv Complexity, and Katz-Sevcik, Variance and Box Counting Fractal Dimension. Assessment of disorganization was evaluated by a Kruskal Wallis statistical test. Except Detrended Fluctuation Analysis and Variance Fractal Dimension, the CFAE disorganization was found statistically significant even for low significant level alpha = 0.001. Moreover, the increasing complexity of A-EGM signals was reflected by higher values of General Correlation Dimension of order 1 and Approximate Entropy.

Comparison of several classifiers to evaluate endocardial electrograms fractionation in human.

Complex fractionated atrial electrograms (CFAEs) may represent the electrophysiological substrate for atrial fibrillation (AF). Progress in signal processing algorithms to identify CFAEs sites is crucial for the development of AF ablation strategies. A novel algorithm for automated description of atrial electrograms (A-EGMs) fractionation based on wavelet transform and several statistical pattern recognition methods was proposed and new methodology of A-EGM processing was designed and tested. The algorithms for A-EGM classification were developed using normal density based classifiers, linear and high degree polynomial classifiers, nearest mean scaled classifiers, nonlinear classifiers, neural networks and j48. All classifiers were compared and tested using a representative set of 1.5 s A-EGMs (n = 68) ranked by 3 independent experts 100% coincidentialy into 4 classes of fractionation: 1 - organized atrial activity; 2 - mild; 3 - intermediate; 4 - high degree of fractionation. Feature extraction and well performing classification algorithms tested here showed maximal error of 15% and mean classification error across all implemented classifiers 9%, and the best mean classification error 5.9% (nearest mean classifier), and classification error of highly fractionated A-EGMs of approximately 9%.

Examining cross-database global training to evaluate five different methods for ventricular beat classification.

The detection of ventricular beats in the holter recording is a task of great importance since it can direct clinicians toward the parts of the electrocardiogram record that might be crucial for determining the final diagnosis. Although there already exists a fair amount of research work dealing with ventricular beat detection in holter recordings, the vast majority uses a local training approach, which is highly disputable from the point of view of any practical-real-life-application. In this paper, we compare five well-known methods: a classical decision tree approach and its variant with fuzzy rules, a self-organizing map clustering method with template matching for classification, a back-propagation neural network and a support vector machine classifier, all examined using the same global cross-database approach for training and testing. For this task two databases were used-the MIT-BIH database and the AHA database. Both databases are required for testing any newly developed algorithms for holter beat classification that is going to be deployed in the EU market. According to cross-database global training, when the classifier is trained with the beats from the records of one database then the records from the other database are used for testing. The results of all the methods are compared and evaluated using the measures of sensitivity and specificity. The support vector machine classifier is the best classifier from the five we tested, achieving an average sensitivity of 87.20% and an average specificity of 91.57%, which outperforms nearly all the published algorithms when applied in the context of a similar global training approach.

Multivariate analysis of full-term neonatal polysomnographic data.

INTRODUCTION: Polysomnography (PSG) is one of the most important noninvasive methods for studying maturation of the child brain. Sleep in infants is significantly different from sleep in adults. This paper addresses the problem of computer analysis of neonatal polygraphic signals. METHODS: We applied methods designed for differentiating three important neonatal behavioral states: quiet sleep, active sleep, and wakefulness. The proportion of these states is a significant indicator of the maturity of the newborn brain in clinical practice. In this study, we used data provided by the Institute for Care of Mother and Child, Prague (12 newborn infants of similar postconceptional age). The data were scored by an experienced physician to four states (wake, quiet sleep, active sleep, movement artifact). For accurate classification, it was necessary to determine the most informative features. We used a method based on power spectral density (PSD) applied to each EEG channel. We also used features derived from electrooculogram (EOG), electromyogram (EMG), ECG, and respiration [pneumogram (PNG)] signals. The most informative feature was the measure of regularity of respiration from the PNG signal. We designed an algorithm for interpreting these characteristics. This algorithm was based on Markov models. RESULTS: The results of automatic detection of sleep states were compared to the "sleep profiles" determined visually. We evaluated both the success rate and the true positive rate of the classification, and statistically significant agreement of the two scorings was found. Two variants, for learning and for testing, were applied, namely learning from the data of all 12 newborns and tenfold cross-validation, and learning from the data of 11 newborns and testing on the data from the 12th newborn. We utilized information obtained from several biological signals (EEG, ECG, PNG, EMG, EOG) for our final classification. We reached the final success rate of 82.5%. The true positive rate was 81.8% and the false positive rate was 6.1%. DISCUSSION: The most important step in the whole process is feature extraction and feature selection. In this process, we used visualization as an additional tool that helped us to decide which features to select. Proper selection of features may significantly influence the success rate of the classification. We made a visual comparison of the computed features with the manual scoring provided by the expert. A hidden Markov model was used for classification. The advantage of this model is that it determines the future behavior of the process by its present state. In this way, it preserves information about temporal development.

A new approach to automated assessment of fractionation of endocardial electrograms during atrial fibrillation.

Complex fractionated atrial electrograms (CFAEs) may represent the electrophysiological substrate for atrial fibrillation (AF). Progress in signal processing algorithms to identify sites of CFAEs is crucial for the development of AF ablation strategies. A novel algorithm for automated description of fractionation of atrial electrograms (A-EGMs) based on the wavelet transform has been proposed. The algorithm was developed and validated using a representative set of 1.5 s A-EGM (n = 113) ranked by three experts into four categories: 1-organized atrial activity; 2-mild; 3-intermediate; 4-high degree of fractionation. A tight relationship between a fractionation index and expert classification of A-EGMs (Spearman correlation rho = 0.87) was documented with a sensitivity of 82% and specificity of 90% for the identification of highly fractionated A-EGMs. This operator-independent description of A-EGM complexity may be easily incorporated into mapping systems to facilitate CFAE identification and to guide AF substrate ablation.

Automatic detection of sleep stages in neonatal EEG using the structural time profiles.

The new method for automatic sleep stages detection in neonatal EEG was developed. The procedure is based on processing of time profiles computed by adaptive segmentation and subsequent classification of signal graphoelements. The time profiles, functions of the class membership in the course of time, reflect the dynamic EEG structure and may be used for indication of changes in the neonatal sleep stages.

School children dyslexia analysis using self organizing maps.

The main goal of the study is an unsupervised classification of school children dyslexia. Eye movements of 49 subjects were measured using videooculographic technique (VOG) during two non-reading and one reading tasks. A feature selection was performed obtaining data set consisting of 26 features. Next an inductive modelling technique was applied to data set resulting in extraction of six features which were used as the input to self-organizing map (SOM). Three clusters were finally formed by the SOM proving that the proposed methodology is suitable for automatic dyslexia analysis.

Speech recognition methods applied to biomedical signals processing.

The paper focuses on processing of long biological signals used during monitoring procedures like in the case of portable Holter device for arrythmia analysis (ECG), intracranial pressure monitoring (ICP) in intensive care unit or overnight electroencephalogram monitoring (EEG) for sleep apnea detection. Two methods taken from speech processing are proposed: dynamic time warping (DTW) and hidden Markov models (HMM). The unsupervised analysis of ECG and ICP beats is carried out using hierarchical clustering approach. In case of EEG, first the estimation of sleep stages is performed and next the different breathing events are detected by HMM by means of Viterbi inference. We show that for the first two problems DTW outperforms HMM while in the third case the HMM inference capability makes HMM suitable for sleep apnea diagnosis.

Proposed standard definition for child overweight and obesity.

The prevalence of obesity in child population is becoming a serious problem in all advanced countries. Therefore many authors try to find a definition according to which it would be possible to assess overweight and obesity in examined subjects in a simple manner. The simplest way is to plot the subject's data in a percentile zone of reference data according to his BMI value. The problem is dealt with e.g. by Cole et al. (1) who defined internationally acceptable standards of BMI for children and adolescents from 0-18 years. These standards are, however, based only on six national studies and on the assumption that in the population of 18-year-olds there are 10% subjects with BMI values above 25 and 3% subjects with values above 30. The suggested standards thus raise the level of the 90th and 97th percentile, as compared with BMI reference data of 1991 which are used in the Czech Republic. Therefore the ratio of obese children in the Czech Republic and in many other countries would be very small. Internationally acceptable standards should be therefore based on a broader discussion of the professional public.

Medical applications of enhanced rule-based expert systems.

The paper describes several types of efficiency enhancements of 'classical' rule-based diagnostic expert systems. The blackboard control structure enables to explore more knowledge bases of the same syntax in parallel, the taxonomy structures make fast zooming of attention possible and provide additional inference mechanism based on inheritance principles. The applicability of the enhancing techniques is documented by four case studies exploring the extended FEL-EXPERT shell in different tasks of medical decision-making. The authors consider the enhancing techniques as useful steps on the way from 'classical' diagnostic expert systems towards more complex multi-agent decision tools.

Future directions.

Despite progress in promotion and support of breastfeeding over the past decade, the HIV pandemic necessitates new actions based on human rights, such as voluntary and confidential testing and counseling, offering HIV-positive women objective information on the risks and costs of all infant feeding options, and providing appropriate support for their decisions. Implementation of the Baby-Friendly Hospital Initiative and the International Code of Marketing are essential components of a rights-based policy response to HIV and will lessen spillover of replacement feeding among HIV-negative women. Protective effects of nevirapine and exclusive breastfeeding, as well as the listed additional topics, require further research. We have yet to make exclusive breastfeeding easy and common when mothers choose to breastfeed.

Growth of the Czech child population 0-18 years compared to the World Health Organization growth reference.

Nationwide anthropological surveys of 0-18-year-old children were carried out in the Czech Republic (former Czech regions of Czechoslovakia) five times during the last 40 years, at 10-year intervals. The measurements provide reference data for the growth of the Czech child population and at the same time ample material for comparison with other populations. A comparison of the results of such nationwide anthropometric surveys with the World Health Organization (WHO) growth reference implies that the growth trend of Czech children is comparable with the WHO reference. The greatest similarity in the shape of the WHO height curves and the results obtained in Czechoslovakia was in 1971. The results obtained in the nationwide surveys also imply that the secular trend still exerts its effect. In the case of height, this finding is consistent during the whole period of 40 years. Concerning weight in higher age categories, especially in girls, the trend slowed down or stopped. Furthermore, there is no problem of malnutrition in the Czech child population. Similarly, as in any other developed nations, the prevalence of overweight and obesity is the problem that merits attention. Am. J. Hum. Biol. 9:459-468, 1997. © 1997 Wiley-Liss, Inc.