Papers

Visit my Research Gate or Google Scholar profile for more information about my papers.

 

Publications:

  1. M Soliński, J Gierałtowski, J Żebrowski, “Modeling heart rate variability including the effect of sleep stages”, Chaos: An Interdisciplinary Journal of Nonlinear Science 26 (2), 023101 (2016) – see also in Reasearch Gate

    Abstract

    We propose a model for heart rate variability (HRV) of a healthy individual during sleep with the assumption that the heart rate variability is predominantly a random process. Autonomic nervous system activity has different properties during different sleep stages, and this affects many physiological systems including the cardiovascular system. Different properties of HRV can be observed during each particular sleep stage. We believe that taking into account the sleep architecture is crucial for modeling the human nighttime HRV. The stochastic model of HRV introduced by Kantelhardt was used as the initial starting point. We studied the statistical properties of sleep in healthy adults, analyzing 30 polysomnographic recordings, which provided realistic information about sleep architecture. Next, we generated synthetic hypnograms and included them in the modeling of nighttime RR interval series. The results of standard HRV linear analysis and of nonlinear analysis (Shannon entropy, Poincaré plots, and multiscale multifractal analysis) show that—in comparison with real data—the HRV signals obtained from our model have very similar properties, in particular including the multifractal characteristics at different time scales. The model described in this paper is discussed in the context of normal sleep. However, its construction is such that it should allow to modelheart rate variability in sleep disorders. This possibility is briefly discussed. 
  2. J Gierałtowski, K Ciuchciński, I Grzegorczyk, K Kośna, M Soliński, P Podziemski, “RS slope detection algorithm for extraction of heart rate from noisy, multimodal recordings”, Physiological measurement 36 (8), 1743 (2015) – see also in Reasearch Gate

    Abstract

    Current gold-standard algorithms for heart beat detection do not work properly in the case of high noise levels and do not make use of multichannel data collected by modern patient monitors. The main idea behind the method presented in this paper is to detect the most prominent part of the QRS complex, i.e. the RS slope. We localize the RS slope based on the consistency of its characteristics, i.e. adequate, automatically determined amplitude and duration. It is a very simple and non-standard, yet very effective, solution. Minor data pre-processing and parameter adaptations make our algorithm fast and noise-resistant. As one of a few algorithms in the PhysioNet/Computing in Cardiology Challenge 2014, our algorithm uses more than two channels (i.e. ECG, BP, EEG, EOG and EMG). Simple fundamental working rules make the algorithm universal: it is able to work on all of these channels with no or only little changes. The final result of our algorithm in phase III of the Challenge was 86.38 (88.07 for a 200 record test set), which gave us fourth place. Our algorithm shows that current standards for heart beat detection could be improved significantly by taking a multichannel approach. This is an open-source algorithm available through the PhysioNet library .

Conference papers:

  1. Iga Grzegorczyk, Mateusz Soliński, Michał Łepek, Anna Perka, Jacek Rosiński, Joanna Rymko, Katarzyna Stępień, Jan Gierałtowski, “PCG classification using a neural network approach”, Computing in Cardiology Conference (CinC), 1129-1132 (2016)

    Abstract & Full Text

    Phonocardiography (PCG) is the one of noninvasive ways to diagnose condition of human heart. The mechanics of heart muscle contractions and closure of the heart valves generates vibrations audible as sounds and murmurs, which can be analysed by qualified cardiologists. Developing an accurate algorithm to determine whether patients’ heart works properly or should be referred to an expert for further diagnosis would significantly improve the quality of healthcare system. It would allow to perform less unnecessary, expensive and time consuming examinations. The analysed data consisted of PCG recordings from the training set provided by the organizers of the PhysioNet Challenge 2016. Its length variedfrom several to 120 seconds. We propose the machine learning algorithm based on neural networks. The segmentation of the PCG signals is performed with algorithm based on Hidden Markov Model. Whereas, the features necessary to define whether the signal looks normal or should be further analysed were carefully chosen by our team and belonged to time domain, ordinate axis or frequency domain group. The great emphasis was put on the statistical features representing the characteristics of the signal. Their optimal values were found during the process of learning of our algorithm. The best overall score we achieved in the official phase of the PhysioNet Challenge 2016 is 0.79 with specificity 0.76 and sensitivity 0.81.
  2. Jan J Gierałtowski, Iga Grzegorczyk, Kamil Ciuchciński, Katarzyna Kośna, Mateusz Soliński, Piotr Podziemski, “Algorithm for life-threatening arrhythmias detection with reduced false alarms“, Computing in Cardiology Conference (CinC), 1201-1204 (2015)

    Abstract & Full Text

    With the increasing quality and costs of health services a lot of attention is put to provide excellent care on the ICU. However, the amount of false alarms of cardiac episodes still outnumbers the true ones significantly. The advanced analysis of multiple signals registered by monitoring system might enable reduction of the false alarms. We analyzed 750 multi-channel recordings from the PhysioNet Challenge 2015 labeled either ‘true alarm’ or ‘false alarm’. In our algorithm there are multiple methods enabling to determine the location of R peaks in ECG signal, basing mostly on RS slopes. Similar slope detection method is performed for other channels provided. In case of signals where it is not possible to detect QRS complexes direct signal morphology assessment is used. These steps allowed us to obtain information needed to verifY if the alarm was true or false. The final scores of PhysioNet Challenge 2015 were: 57. 72 for Real-time event and 63.69 for the Retrospective one.
  3. M Soliński, J Gierałtowski, J Żebrowski, “Modeling of human heart rate variability enhanced using stochastic sleep architecture properties”, Computing in Cardiology Conference (CinC), 513-516 (2014)

    Abstract & Full Text

    Human sleep consists of four characteristic phases: light (L), deep (D), REM sleep and almost-awake state (W) with additional arousal episodes (Exc). All of these elements create a nontrivial, complex structure, the statistical properties of which were studied here carefully. We observed a different behavior of heart rate variability for each phase. Thus, we should take these specific properties of sleep architecture into consideration while modeling heart rate variability.We analyzed 34 simultaneous heart rate variability and 30 EEG nighttime recordings from healthy adults. EEG provides accurate information about sleep phases. The main idea behind our sleep architecture reconstruction is to consider two properties: probabilities of transitions between all possible pairs of phases and probability distribution of phase durations. We calculated the probabilities of transition between each pair of phases and we aggregated them into two probability matrices (separately for each half of the sleep period because the character of the inter-phase transitions is diferent in early and late sleep). We found also that the probability distribution of L, D and REM sleep duration are described by a gamma distribution and that of the W phase – by a Pareto distribution. To generate the RR intervals for every sleep phase, we use the model described in [1j. We consider three variants: (a) periodic sleep cycles with the sequences of phases: L, D, REM, W in each cycle, (b) a randomized distribution of phases, (c) the architecture based on our model. The results show that variant (c) gives 50% of the time series indistinguishable from real data using all standard linear and nonlinear HRV assessment methods while for variants (a) and (b) we obtain 41% and 3% accordingly (see full text).
  4. J Gierałtowski, K Ciuchciński, I Grzegorczyk, K Kośna, M Soliński, P Podziemski, “Heart rate variability discovery: Algorithm for detection of heart rate from noisy, multimodal recordings”, Computing in Cardiology Conference (CinC), 253-256 (2014) 

    Abstract & Full Text

    Starting point for the heart rate variability analysis is the ECG signal, which ensures the most precise way of detecting heartbeats. However, very often devices used to record ECG also record at the same time many other physiological signals containing useful information about heart rate. In the case of the poor ECG quality or its absence information about beats is lost. This raises the need for robust algorithms which could locate heartbeats in continuous long-term data from bedside monitors, allowing reliable, automatic analysis (see full text).
  5. J Gierałtowski, K Ciuchciński, I Grzegorczyk, K Kośna, M Soliński, P Podziemski, “Heart rate variability discovery: Algorithm for detection of heart rate from noisy, multimodal recordings”, Computing in Cardiology Conference (CinC), 253-256 (2014) 

    Abstract & Full Text

    Starting point for the heart rate variability analysis is the ECG signal, which ensures the most precise way of detecting heartbeats. However, very often devices used to record ECG also record at the same time many other physiological signals containing useful information about heart rate. In the case of the poor ECG quality or its absence information about beats is lost. This raises the need for robust algorithms which could locate heartbeats in continuous long-term data from bedside monitors, allowing reliable, automatic analysis (see full text).

Other papers:

  1. [PL] M. Solinski, J. Gierałtowski, Różne oblicza modeli matematycznych zmienności rytmu serca, Artykuł dla Centrum Zastosowań Matematyki na Wydziale Fizyki Technicznej i Matematyki Stosowamej Politechniki Gdańskiej, 2015 (ISBN 978-83-942807-3-4)

    Abstract & Full Text

    Złożoność zmienności rytmu serca człowieka sprawiła, że modelowanie tego zjawiska stanowi niezwykle wymagające wyzwanie. W niniejszej pracy zostaną przedstawione kolejno trzy koncepcje, znacząco różniące się pod względem matematycznych założeń, dotyczące takich zagadnień jak fraktale, transformata Fouriera oraz procesy stochastyczne. Ten szeroki wachlarz stosowanego aparatu matematycznego przyczynił się do dużej różnorodności generowanych serii interwałów RR, szczególnie pod względem ich dynamicznych właściwości. Wraz z opisem modeli zostaną przedstawione zarówno ich mocne jak i słabe strony opierając się na wybranych wynikach analizy sygnałów. W ramach wniosków podane zostają wskazówki dążące do unifikacji pewnych elementów powyższych koncepcji mogących prowadzić do uzyskania optimum w opisie zjawiska zmienności rytmu serca człowieka (see full text).