: This thesis applies neural network feature selection techniques to multivariate time series data to improve prediction of a target time series. Two approaches to feature selection are used. First, a subset enumeration method is used to determine which financial indicators are most useful for aiding in prediction of the S&P 500 futures daily price. The candidate indicators evaluated include RSI, Stochastics and several moving averages. Results indicate that the Stochastics and RSI indicators result in better prediction results than the moving averages. The second approach to feature selection is calculation of individual saliency metrics. A new decision boundary-based individual saliency metric, and a classifier independent saliency metric are developed and tested. Ruck's saliency metric, the decision boundary based saliency metric, and the classifier independent saliency metric are compared for a data set consisting of the RSI and Stochastics indicators as well as delayed closing price values. The decision based metric and the Ruck metric results are similar, but the classifier independent metric agrees with neither of the other metrics. The nine most salient features, determined by the decision boundary based metric, are used to train a neural network and the results are presented and compared to other published results. (AN)

/pdf/nonlinear-time-series-analysis-1j3a35n4y3.pdf

Nonlinear Time Series Analysis.

Photoplethysmography and its application in clinical physiological measurement

In previous chapters, we have studied the model 

$$y = A\beta + \epsilon, $$

where the mean Ey = Aβ depends linearly on the parameters β, the errors are normal (Gaussian), and the errors are additive. We have also seen (Chapter 7) that in some situations, a transformation of the problem may help to correct some departure from our standard model assumptions. For example, in §7.3 on variance-stabilising transformations, we transformed our data from y to some function g(y), to make the variance constant (at least approximately). We did not there address the effect on the error structure of so doing. Of course, $g(y) = g(A\beta + \epsilon )$ as above will not have an additive Gaussian error structure any more, even approximately, in general.

Generalised Linear Models

The number of publications investigating heart rate variability (HRV) in psychiatry and the behavioral sciences has increased markedly in the last decade. In addition to the significant debates surrounding ideal methods to collect and interpret measures of HRV, standardized reporting of methodology in this field is lacking. Commonly cited recommendations were designed well before recent calls to improve research communication and reproducibility across disciplines. In an effort to standardize reporting, we propose the Guidelines for Reporting Articles on Psychiatry and Heart rate variability (GRAPH), a checklist with four domains: participant selection, interbeat interval collection, data preparation and HRV calculation. This paper provides an overview of these four domains and why their standardized reporting is necessary to suitably evaluate HRV research in psychiatry and related disciplines. Adherence to these communication guidelines will help expedite the translation of HRV research into a potential psychiatric biomarker by improving interpretation, reproducibility and future meta-analyses.

/pdf/guidelines-for-reporting-articles-on-psychiatry-and-heart-5grjuekelz.pdf

Guidelines for Reporting Articles on Psychiatry and Heart rate variability (GRAPH): recommendations to advance research communication

The Internet of Medical Things (IoMT) designates the interconnection of communication-enabled medical-grade devices and their integration to wider-scale health networks in order to improve patients’ health. However, because of the critical nature of health-related systems, the IoMT still faces numerous challenges, more particularly in terms of reliability, safety, and security. In this paper, we present a comprehensive literature review of recent contributions focused on improving the IoMT through the use of formal methodologies provided by the cyber-physical systems community. We describe the practical application of the democratization of medical devices for both patients and health-care providers. We also identify unexplored research directions and potential trends to solve uncharted research problems.

/pdf/internet-of-medical-things-a-review-of-recent-contributions-2ft727ty1v.pdf

Internet of Medical Things: A Review of Recent Contributions Dealing With Cyber-Physical Systems in Medicine

Monitoring of uterine contraction activity is an important diagnostic tool used during both pregnancy and labour. The strain the pregnant uterus exerts on the maternal abdomen is measured via external tocography. However, limitation of this approach has caused the development of another technique-electrohysterography--which is based on the recording of electrical uterine activity. A computer-aided system is presented, which allows the recording of electrohysterographic signals from the maternal abdomen and their on-line analysis both in time and frequency domains. As a research material, we acquired 108 traces during a 24 h period before labour from a group of patients between 37 and 40 weeks of gestation. The comparison study between electrohysterography and tocography was carried out thanks to the possibility of simultaneous recording of mechanical and electrical uterine activities. The obtained results show that both methods demonstrate high agreement in relation to the number of contractions recognized as being consistent. However, their agreement in relation to the quantitative description of recognized patterns has appeared to be unacceptable to consider these methods as fully alternative. The appropriate way of further development of electrohysterography seems to be spectral analysis. Several spectral parameters describing electrophysiological properties of uterine muscle can be obtained by the use of electrohysterographic signals.

https://iopscience.iop.org/article/10.1088/0967-3334/26/5/014/pdf

Quantitative analysis of contraction patterns in electrical activity signal of pregnant uterus as an alternative to mechanical approach.

A method for comparison of two acquisition techniques that are applied in clinical practice to provide information on fetal condition is presented. The aim of this work was to evaluate the commonly used Doppler ultrasound technique for monitoring of mechanical activity of fetal heart. Accuracy of beat-to-beat interval determination together with its influence on indices describing the fetal heart rate (FHR) variability calculated automatically using computer-aided fetal monitoring system were examined. We considered the direct fetal electrocardiography as a reference technique because it ensures the highest possible accuracy of heart interval measurement, and additionally all the definitions of popular time domain parameters quantifying FHR variability formerly have been created using the fetal electrocardiogram. We evaluated the reliability of various so called short-term and long-term variability indices, when they are calculated automatically using the signal obtained via the Doppler US from a fetal monitor. The results proved that evaluation of the acquisition technique influence on fetal well-being assessment cannot be accomplished basing on direct measurements of heartbeats only. The more relevant is the estimation of accuracy of the variability indices, since analysis of their changes can significantly increase predictability of fetal distress

Comparison of Doppler ultrasound and direct electrocardiography acquisition techniques for quantification of fetal heart rate variability

The currently used fetal monitoring instrumentation that is based on Doppler ultrasound technique provides the fetal heart rate (FHR) signal with limited accuracy. It is particularly noticeable as significant decrease of clinically important feature - the variability of FHR signal. The aim of our work was to develop a novel efficient technique for processing of the ultrasound signal, which could estimate the cardiac cycle duration with accuracy comparable to a direct electrocardiography. We have proposed a new technique which provides the true beat-to-beat values of the FHR signal through multiple measurement of a given cardiac cycle in the ultrasound signal. The method consists in three steps: the dynamic adjustment of autocorrelation window, the adaptive autocorrelation peak detection and determination of beat-to-beat intervals. The estimated fetal heart rate values and calculated indices describing variability of FHR, were compared to the reference data obtained from the direct fetal electrocardiogram, as well as to another method for FHR estimation. The results revealed that our method increases the accuracy in comparison to currently used fetal monitoring instrumentation, and thus enables to calculate reliable parameters describing the variability of FHR. Relating these results to the other method for FHR estimation we showed that in our approach a much lower number of measured cardiac cycles was rejected as being invalid. The proposed method for fetal heart rate determination on a beat-to-beat basis offers a high accuracy of the heart interval measurement enabling reliable quantitative assessment of the FHR variability, at the same time reducing the number of invalid cardiac cycle measurements.

/pdf/a-novel-technique-for-fetal-heart-rate-estimation-from-1a27udw97w.pdf

A novel technique for fetal heart rate estimation from Doppler ultrasound signal

Bioelectrical fetal heart activity being recorded from maternal abdominal surface contains more information than mechanical heart activity measurement based on the Doppler ultrasound signals. However, it requires extraction of fetal electrocardiogram from abdominal signals where the maternal electrocardiogram is dominant. The simplest technique for maternal component suppression is a blanking procedure, which relies upon the replacement of maternal QRS complexes by isoline values. Although, in case of coincidence of fetal and maternal QRS complexes, it causes a loss of information on fetal heart activity. Its influence on determination of fetal heart rate and the variability analysis depends on the sensitivity of the heart-beat detector used. The sensitivity is defined as an ability to detect the incomplete fetal QRS complex. The aim of this work was to evaluate the influence of the maternal electrocardiogram suppression method used on the reliability of FHR signal being calculated.

The influence of coincidence of fetal and maternal QRS complexes on fetal heart rate reliability

Cardiotocography (CTG) is a biophysical method of fetal condition assessment based mainly on recording and automated analysis of fetal heart activity. The computerized fetal monitoring systems provide the quantitative description of the CTG signals, but the effective conclusion generation methods for decision process support are still needed. Assessment of the fetal state can be verified only after delivery using the fetal (newborn) outcome data. One of the most important features defining the abnormal fetal outcome is low birth weight. This paper describes an application of the artificial neural network based on logical interpretation of fuzzy if-then rules neurofuzzy system to evaluate the risk of low-fetal birth weight using the quantitative description of CTG signals. We applied different learning procedures integrating least squares method, deterministic annealing (DA) algorithm, and e-insensitive learning, as well as various methods of input dataset modification. The performance was evaluated with the number of correctly classified cases (CC) expressed as the percentage of the testing set size, and with overall index (OI) being the function of predictive indexes. The best classification efficiency (CC = 97.5% and OI = 82.7%), was achieved for integrated DA with e-insensitive learning and dataset comprising of the CTG traces recorded as earliest for a given patient. The obtained results confirm efficiency for supporting the fetal outcome prediction using the proposed methods.

Predicting the Risk of Low-Fetal Birth Weight From Cardiotocographic Signals Using ANBLIR System With Deterministic Annealing and ${\bm \varepsilon}$ -Insensitive Learning

Janusz Wrobel

Papers

Quantitative analysis of contraction patterns in electrical activity signal of pregnant uterus as an alternative to mechanical approach.

Comparison of Doppler ultrasound and direct electrocardiography acquisition techniques for quantification of fetal heart rate variability

A novel technique for fetal heart rate estimation from Doppler ultrasound signal

The influence of coincidence of fetal and maternal QRS complexes on fetal heart rate reliability

Predicting the Risk of Low-Fetal Birth Weight From Cardiotocographic Signals Using ANBLIR System With Deterministic Annealing and ${\bm \varepsilon}$ -Insensitive Learning