Wieslaw Wajs

Expert system supporting an early prediction of the bronchopulmonary dysplasia

This work presents a decision support system which uses machine learning to support early prediction of bronchopulmonary dysplasia (BPD) for extremely premature infants after their first week of life. For that purpose a knowledge database was created based on the historical data gathered including data on 109 patients with birth weight less than or equal to 1500g. The core of the database consists of support vector machine and logit regression classification results calculated specifically for that system, and obtained by considering 2(14) different combinations of 14 risk factors. Based on the results obtained and user demands, the system recommends the best methods and the most suitable parameter subset among those currently available to the user. The program is also able to estimate the accuracy, sensitivity and specificity together with their standard deviations. The user is also given information on which additional parameter it is worth adding to his measurement system most and what an increase in prediction efficiency it is expected to trigger. The BPD can be predicted by the system with the accuracy reaching up to 83.25% in the best-case scenario, i.e. higher than for most of the models presented in the literature. This work presents a set of examples illustrating the difficulties in obtaining one single model that can be widely used, and thus explaining why an expert system approach is much more useful in day-to-day clinical practice. In addition, the work discusses the significance of the parameters used and the impact of a chosen method on the sensitivity and specificity.

Bronchopulmonary Dysplasia prediction using Support Vector Machine and LIBSVM

The paper presents BPD (Bronchopulmonary Dysplasia) prediction for extremely premature infants after their first week of life. SVM (Support Vector Machine) algorithm implemented in LIBSVM[1] was used as classifier. Results are compared to others gathered in previous work [2] where LR (Logit Regression) and Matlab environment SVM implementation were used. Fourteen different risk factor parameters were considered and due to the high computational complexity only 3375 random combinations were analysed. Classifier based on eight feature model provides the highest accuracy which was 82.60%. The most promising 5-feature model which gathered 82.23% was reasonably immune to random data changes and consistent with LR results. The main conclusion is that unlike Matlab SVM[2] implementation, LIBSVM can be successfully used in considered problem, but it is less stable than LR. In addition, the article discusses influence of the model parameters selection on prediction quality.

Bronchopulmonary Dysplasia Prediction Using Support Vector Machine and Logit Regression

The paper presents BPD (Bronchopulmonary Dysplasia) prediction for extremely premature infants after their first week of life. SVM (Support Vector Machine) and LR (Logit Regression) are used as classifiers. Data was collected thanks to the Neonatal Intensive Care Unit of The Department of Pediatrics at Jagiellonian University Medical College and includes 109 patients with birth weight less than or equal to 1500g. Fourteen different risk factor parameters were considered and all 214 combinations were analyzed. Classifier based on six feature LR model provides accuracy up to 82%, while SVM one turns out to be generally much worse, providing in best case scenario 80% of accuracy. In addition, the article discusses the influence of the model parameters selection on prediction quality.

Intelligent information system for interpretation of dermatoglyphic patterns of down's syndrome in infants

The paper describes design of an intelligent information system for assessment of dermatoglyphic indices of Downs syndrome in infants. The system supports medical diagnosis by automatic processing of dermatoglyphic prints and detecting features indicating presence of genetic disorders. Application of image processing and pattern recognition algorithms in pattern classification of fingerprints and prints of hallucal area of the sole is described. Application of an algorithm based on multi-scale pyramid decomposition of an image is proposed for ridge orientation calculation. A method of singular points detection and calculation of ATD angle of the palm print is presented. Currently achieved results in dermatoglyphic prints enhancement, classification and analysis are discussed. Scheme used in classification of dermatoglyphic prints is described. RBF and triangular kernel types are used in the training of SVM multi-class systems generated with one-vs-one scheme. Results of experiments conducted on the database of Collegium Medicum of the Jagiellonian University in Cracow are presented.

Bronchopulmonary Dysplasia Prediction using Logistic Regression

The prognostic value of some continuos measurement parameters calculating Bronchopulmonary Dysplasia predictor is the main goal of the paper. The most important question is, if the continuous measurement of parameters can help us to build a better algorithm solving Bronchopulmonary Dysplasia prediction problem?

Artificial immune system for medical data classification

The article presents application of artificial immune algorithms in classification of vectorized medical data sets. Artificial immune network was created and trained for the purpose of arterial blood gasometry parameters (pH, pCO2, pO2, HCO3) classification. Training data originates from the Infant Intensive Care Unit of the Polish – American Institute of Pediatry, Collegium Medicum, Jagiellonian University in Cracow.

A Method for Microprocessor External Digital Circuits Using Programmable Devices

Abstract A method for microprocessor external digital circuits using programmable device is proposed. Each node of the Local Operating Network (LON) is independently working device, and except network communication port, contains I/O port. The paper describes how to build a host interface to the Neuron Chip microprocessor that uses the Microprocessor Interface Program. The Microprocessor Interface Program establishes a fast direct link from the host processor to the Neuron Chip network processor without the Neuron Chip being involved in any application processing. The Neuron Chip parallel I/O object pennits bi-directional data transfer at rates of up to 3.3Mbps.

Medical Decision Support System for Assessment of Dermatoglyphic Indices and Diagnosis of Down’s Syndrome

The paper describes design of an intelligent information system for assessment of dermatoglyphic indices of Down’s syndrome in infants. The system supports medical diagnosis by automatic processing of dermatoglyphic prints and detecting features indicating presence of genetic disorders. Application of image processing and pattern recognition algorithms in pattern classification of fingerprints and prints of the hallucal area of the sole is described. Application of an algorithm based on multi-scale pyramid decomposition of an image is proposed for ridge orientation calculation. A method of singular points detection and calculation of the ATD angle of the palm print is presented. Currently achieved results in dermatoglyphic prints enhancement, classification and analysis are discussed.

Linear Programming and Risk Analysis Methods for Municipal Solid Waste Decision Support System

Abstract Mathematical modelling is used to manage the Municipal Solid Waste (MSW) This paper develops a linear programming (LP) model for solid waste management systems. The integration of the Mathematical Programming Model and Risk Analysis Methods is demonstrated through application to MSW management systems in Niepolomice town, not far from Krakow, Poland. Experimental results and the simulation values are obtained through using Cris tall Ball®, a forecasting and risk analysis program for effective remediation of contaminated soil and waste landfill.

Simulation of Electrohydrodynamic Phenomenon using Computational Intelligence Methods

Abstract The electrohydrodynamic effect occurring during an underwater electrical explosion of a copper wire and method of its simulation using computational intelligence algorithms are presented in the paper. Principles of operation of the seismic wave EHD generator are explained. Differential equations describing a mathematical model of changes in a release circuit of the EHD generator are provided. The results of computations of differential equations, using Runge -Kutta algorithm, are provided. The simulation is based on current intensity and pressure data sets recorded during field experiments and data obtained by solving differential equations of the theoretical model. The relationships between electrical parameters of the electrohydrodynamic discharge circuit, material parameters of the exploded wire and result- ing pressure values of the shock wave are discussed. A regression model is created using computational intelligence algorithms for prediction of pressure values of a seismic wave resulting from the explosion of the wire. The results of the prediction of the pressure wave using neural networks are presented in the paper and verified by comparing the measured and calculated values of the pressure wave.