Wyniki wyszukiwania - BazTech

1

Instance selection techniques in reduction of data streams derived from medical devices

Byczkowska-Lipińska L., Wosiak A.

Przegląd Elektrotechniczny

|

2017

|

R. 93, nr 12

115--118

EN

The research described in this paper concerns the reduction of streams of data derived from medical devices, i.e. ECG recordings. Experimental studies included three instance selection techniques: thresholding method, bounds checking and frequent data reduction . It was shown that application the instance selection techniques may reduce data stream by over 90% without losing anomalies or the measurements that are key values for the medical diagnosis.

PL

W ramach niniejszej pracy przeprowadzona została redukcja strumienia danych pozyskanych z urządzeń medycznych. Badania eksperymentalne obejmowały zastosowanie trzech technik selekcji przypadków: metody eliminacji progowej, weryfikacji zakresu oraz redukcji obiektów częstych. W pracy zostało wykazane, że zastosowanie selekcji przypadków pozwala na redukcję strumienia danych o ponad 90% bez utraty wartości kluczowych dla postawienia diagnozy medycznej.

2

ANN as justified granular computing mechanism for medical data classification

Bernas M.

Journal of Medical Informatics & Technologies

|

2015

|

Vol. 24

85--90

EN

The medical data and its classification have to be treated in particular way. The data should not be modified or altered, because this could lead to false decisions. Most state-of-the-art classifiers are using random factors to produce higher overall accuracy of diagnosis, however the stability of classification can vary significantly. Medical support systems should be trustworthy and reliable, therefore this paper proposes fusion of multiple classifiers based on artificial Neural Network (ANN). The structure selection of ANN is performed using granular paradigm, where granulation level is defined by ANN complexity. The classification results are merged using voting procedure. Accuracy of the proposed solution was compared with state-of-the-art classifiers using real medical data coming from two medical datasets. Finally, some remarks and further research directions have been discussed.

3

Influence of missing data imputation method on the classification accuracy of the medical data

Orczyk T., Porwik P.

Journal of Medical Informatics & Technologies

|

2013

|

Vol. 22

111--116

EN

Aim of this study is to show the dangers of filling missing data - particularly medical data. Because there are many dedicated medical expert systems and medical decision support systems, a special attention must be paid on the construction of classifiers. Medical data are almost never complete, and completion of the missing data requires a special care. The safest approach of dealing with missing data would be removing records with missing parameters and/or removing parameters that are missing in the records. Unfortunately reducing data set that is already very small is not always an option. Dangers coming out from data imputation are shown in the article, which presents the influence of selected missing data filling algorithms on the classification accuracy.

4

On statistical analysis methods improving epidemiological studies

Wosiak A.

Journal of Applied Computer Science

|

2013

|

Vol. 21, nr 2

209--219

EN

Nowadays almost all the principles of diagnosis and treatment are determined by the statistical analysis of the observations made by practitioners. The analysis of medical data is therefore one of the most important elements that aﬀect the level of modern medical care. The research described in this paper aims to determine the characteristics of cardiac parameters in healthy children and in children with a diagnosis in arterial hypertension. Studies include children in the Lodz region. The purpose of the analysis is to determine risk factors of arterial hypertension and thus early diagnosis of children and as quickly as possible the inclusion of an appropriate treatment or observation. The analysis applies a number of methods, including descriptive analysis, grouping and statistical inference. The choice of methods is consistent with the requirements of the USMLE (The United States Medical Licensing Examination) and depends on the evaluated parameters. The research is carried out using professional statistical packages and the computer system designed and developed for this study. The use this system allows for the continuous process of research staging of new cases, which is a deﬁnite advantage when conducting epidemiological studies.

5

Grade Approach to the Analysis of Questionnaires and Clinical Scales Data

Wolińska-Welcz A., Welcz H.

Biocybernetics and Biomedical Engineering

|

2009

|

Vol. 29, no. 2

59-71

EN

Grade approach has been applied to visualize and reinvestigate a dataset from diagnostic questionnaires and scales used in psychological and psychiatric clinical practice. Grade exploration reveals important hidden variables strongly influencing other variables and splits questionnaire data into more regular monotone dependent blocks. Two sets of homogeneous and ordered clusters of patients, formed on the basis of first line variables and described also by second line variables, expose links between neurosis symptoms and Internal or External Locus of Control as well as prove useful for specifying and verifying diagnoses. All analyses were performed using the GradeStat program.

6

Selecting relevant diagnosis variables using integer linear programming techniques

Castillo E., Marino S.I., Tressens S.G.

Biocybernetics and Biomedical Engineering

|

2001

|

Vol. 21, no. 1

5-24

PL

Artykuł przedstawia metodę selekcji cech pozwalającą na wybór minimalnego podzbioru S(o) symptomów znaczących z zadanego zbioru symptomów S. Wybór minimalnego podzbioru S(o) dokonywany jest w ten sposób, by zapewnić identyfikowalność danej grupy chorób nawet w warunkach, gdy ustalona liczba symptomów jest utracona. Zmienne binarne i całkowitoliczbowe programowanie liniowe użyto w rozwiązaniu problemu selekcji cech. Poszczególne zmienne binarne wskazują czy dany symptom powinien być włączony do minimalnego podzbioru S(o). Po wybraniu podzbioru S(o) wybierane są zestawy cech służące różnicowaniu pomiędzy daną chorobą a sumą pozostałych chorób. Rezultaty działania proponowanej metody pokazano w artykule na przykładach dwu zbiorów danych: sztucznym i rzeczywistym. Przykład rzeczywistych danych medycznych obejmuje 143 symptomy i 90 chorób. Zbiór ten został znacząco zredukowany w rezultacie zastosowania procedury selekcji cech.

EN

The paper presents a method for selecting a minimum set S(o) of relevant symptoms, from an initially given set S of symptoms, to identify a given set of diseases, even when a fixed number of arbitrary symptoms is missing. Integer linear programming techniques (ILPT) are used, where binary variabies indicate whether a given symptom must be included in the set S(o) of relevant (minimum) symptoms. Once So has been selected, the minima sets of relevant symptoms for distinguishing each disease from the rest of diseases are obtained, also using ILPT. A pedagogical example illustrates the concepts and the proposed methods. Finally, a real example with 143 symptoms and 90 diseases is used to show that substantial reductions of information can be obtained using the proposed method.

7

Wavelet transform in analysis of high-resolution ECG

Lewandowski P., Maniewski R.

Biocybernetics and Biomedical Engineering

|

2000

|

Vol. 20, no. 2

31-37

EN

The method for analysis of high-resolution ECG signals with use of wavelet transform is presented. As a wavelet a modified Morlet function was used. The method was verified on test signals composed of sinus waves as well as ECG test signal with simulated late potentials. Wavelet transform method was studied on two postinfarction patients groups: with and without ventricular tachycardia. A quantitative parameter WF (wavelet factor) based on area under the time-frequency representation was proposed for results evaluation. Mean values of this parameter for studied groups were significantly different.

PL

Przedstawiono analizę sygnałów EKG o wysokiej rozdzielczości za pomocą transformaty falkowej z wykorzystaniem funkcji Morleta. Metodę wstępnie sprawdzono za pomocą sygnałów testowych (sinusoidalnych oraz EKG z symulowanymi potencjałami późnymi), a następnie zweryfikowano w badaniach grup chorych po zawałach bez arytmii i z występowaniem częstoskurczu komorowego. Do oceny arytmii zaproponowano parametr obliczany z pola pod krzywą rozkładu czasowo-częstotliwościowego. Stwierdzono statystyczne różnice tego parametru dla badanych grup chorych po zawałach.

8

Two non-conventional methods for visualization of multivariate two-group data

Bartkowiak A., Szustalewicz A.

Biocybernetics and Biomedical Engineering

|

2000

|

Vol. 20, no. 4

5-20

EN

We introduce two graphical methods which permit to visualize graphically multivariate data belonging to two groups, e.g. disease and control groups. The methods are: a display of two canonical variates, and dynamic projections by the grand tour. Both methods are exploratory: they help in visualization of multivariate data points belonging to two groups of data and learning what is the shape of the two groups. The proposed methods permit also to identify atypieal data points and outliers. The considerations are illustrated by a medical example concerned with patients suffering from ventilatory disorders and control patients in which ventilatory disorders were not stated.

PL

Opisujemy dwie metody graficzne, które pozwalają zobrazować dane podzielone na dwie grupy, np. na chorych i na grupę kontrolną. Są to: wykres dwóch zmiennych kanonicznych i metoda dynamicznych projekcji, nazywana grand tour. Obydwie metody są eksploratywne, tzn. pomagają w wizualizacji punktów należących do dwóch różnych grup oraz w rozpoznawaniu kształtów tych dwóch grup danych. Proponowane metody mogą być również przydatne przy znajdywaniu tzw. odstających obserwacji. Rozważania są ilustrowane danymi medycznymi, dotyczącymi pacjentów cierpiących na niewydolność układu wentylacyjnego i pacjentów kontrolnych, u których nie stwierdzono tej choroby.