Almir Badnjevic

Artificial neural networks in the discrimination of Alzheimer's disease using biomarkers data

This paper presents the results of a study developing artificial neural network system (ANN) for classification of Alzheimers disease (AD) and healthy patients. The classification is done using biomarkers, from cerebrospinal fluid: albumin ratio (CSF/Serum and/or Plasma), Aβ40 (CSF), Aβ42 (CSF), tau-total (CSF) and tau-phospho (CSF). Neural network input parameters are datasets from Alzbiomarkers database. Independent t-test is used to calculate statistical difference between input parameters. Developed neural network was validated with 80 subjects from Alzbiomarkers database. Out of 45 AD subjects, 43 were correctly classified as AD patients, obtaining a sensitivity of 95.5%, and out of 35 healthy subjects 32 were correctly classified obtaining specificity of 91.43%.

Classification of asthma using artificial neural network

This paper presents a system for classification of asthma based on artificial neural network. A total of 1800 Medical Reports were used for neural network training. The system was subsequently tested through the use of 1250 Medical Reports established by physicians from hospital Sarajevo. Out of the aforementioned Medical Reports, 728 were diagnoses of asthma, while 522 were healthy subjects. Out of the 728 asthmatics, 97.11% were correctly classified, and the healthy subjects were classified with an accuracy of 98.85%. Sensitivity and specificity were assessed, as well, which were 97.11% and 98.85%, respectively. Our system for classification of asthma is based on a combination of spirometry (SPIR) and Impulse Oscillometry System (IOS) test results, whose measurement results were inputs to artificial neural network. Artificial neural network is implemented to obtain both static and dynamic assessment of the patients respiratory system.

Integrated software suite for diagnosis of respiratory diseases

Respiratory diseases can be very difficult to diagnose because their symptoms are sometimes very similar to each other. If we analyze asthma and COPD (Chronic Obstructive Pulmonary Disease), diseases which are targeted in our research, we come to the conclusion that early detection of airway impairment can greatly assist in an early diagnosis. In this paper we present an integrated software suite that can help doctors make correct diagnosis of diseases such as asthma and COPD, using lung function tests. The software is based on object oriented methodology. The parameters of spirometry, IOS (Impulse Oscillometry System) and body plethysmography, used in the diagnosis of respiratory disease, will be included in the neuro-fuzzy system in order to help the software suggest proper diagnosis of asthma, COPD or normal lung condition. In order to meet all the conditions that are necessary for the proper and complete diagnosis of respiratory diseases, there is also information about the symptoms, allergies and auscultation of the patient included. In cases where it is not possible to determine the diagnosis on the basis of symptoms, spirometry and IOS test, the software indicates BDT (bronchial dilation test) and BPT (bronchial provocation test), after which new tests are required for spirometry, IOS and body plethysmography in order to get a complete diagnosis.

Classification of stress recognition using Artificial Neural Network

This paper presents the results of a study developing expert system to support stress recognition based on Artificial Neural Network (ANN). Developed ANN is trained using data from Physionet database and collected data from other researchers. The implemented system for stress recognition uses drivers ECG signal, Galvanic Skin Response and Respiration Rate as parameters. Developed neural network was validated with 77 samples. Samples are obtained from subjects using Pasco sensors in 7D cinemas. Out of 77 samples, in 71% of subjects higher level of stress is recognized, while 29% of subjects are classified as subjects with normal vital functions. An accuracy of 99% and specificity of 98% is obtained.

Classification of Asthma Utilizing Integrated Software Suite

The aim of this study was to investigate problems in detecting and diagnosing asthma and based on it, to develop method for classification of asthma. This method is implemented through integrated software suite developed to assist physicians in the analysis and interpretation of pulmonary function test results to improve detection, diagnosis and treatment of asthma. The software determines and classifies the asthma based on fuzzy rules and trained neural network. More than one thousand report samples were used for training data. A total of 289 patients, previously diagnosed with asthma or normal lung conditions by physicians, were tested with this tool. The software performed the classification of patients with asthma in 97.22% and with normal lung function in 98.61% cases in concordance with physician’s report.

Classification of Chronic Obstructive Pulmonary Disease (COPD) Using Integrated Software Suite

Chronic Obstructive Pulmonary Disease (COPD) is a respiratory disorder characterized by chronic and recurrent airflow obstruction, which increases airway resistance. About 75% of COPD patients do not have established diagnosis, most of them in mild degree, but also 4% in severe and 1% in very severe degree of COPD. The reason for that are slow progression of symptoms as cough and exercise intolerance, as well as development of disease in the elderly. Integrated software suite is developed to assist clinicians in the analysis and interpretation of pulmonary function tests data to better detect, diagnose and treat COPD conditions. A total sum of 385 patient reports with previously diagnosed COPD or normal lung conditions by clinicians was tested with this tool. With diagnosed COPD by clinicians there were 252 patients, and even in 92% the software has performed the classification of COPD in the same way as doctors. The software classification of patients with normal lung function was 90.97%.

Medical devices in legal metrology

In addition to knowledge and experience of medical doctors, correct diagnosis and appropriate patient treatment largely depend on accuracy and functionality of medical devices. In a large number of serious medical situations proper functionality of medical devices is crucial for patients. Therefore it is necessary to carry out as strict and independent testing of functionalities of medical devices as possible and to obtain the most accurate and reliable diagnosis and patient treatment. This paper presents the results of study conducted by the Institute of Metrology of Bosnia and Herzegovina (IMBIH) that highlight the necessity of introducing metrology into medicine and defining standard regulations for inspections of medical devices. As it has been previously done for other kinds of devices that are under jurisdiction of the Institute of Metrology of BH, this research provides a foundation for the introduction of medical devices into the legal metrology system with precisely defined units of measurement, their ranges and errors. The study was based upon data collected through three clinical centers, 25 hospitals, 63 health centers and 320 private health institutions in BH over the course of one year. As a result of this study, the medical devices that have been introduced into the legal metrology system in BH include ECG devices, defibrillators, patient monitors, respirators, anesthesia machines, dialysis machines, pediatric and neonatal incubators, therapeutic ultrasounds, infusion pumps and perfusors. Furthermore, standard inspection regulations for the aforementioned medical devices are also defined. Additionally, a national laboratory for the inspection of medical devices was established and it currently operates under the ISO 17020 standard. With the introduction of medical devices into the legal metrology system and with the establishment of a fully operational national laboratory for inspection of medical devices, we expect that the reliability of medical devices in diagnosis and patient care will increase and that the costs of the health care system in BH will be reduced.

Testing of mechanical ventilators and infant incubators in healthcare institutions

The medical device industry has grown rapidly and incessantly over the past century. The sophistication and complexity of the designed instrumentation is nowadays rising and, with it, has also increased the need to develop some better, more effective and efficient maintenance processes, as part of the safety and performance requirements. This paper presents the results of performance tests conducted on 50 mechanical ventilators and 50 infant incubators used in various public healthcare institutions. Testing was conducted in accordance to safety and performance requirements stated in relevant international standards, directives and legal metrology policies. Testing of output parameters for mechanical ventilators was performed in 4 measuring points while testing of output parameters for infant incubators was performed in 7 measuring points for each infant incubator. As performance criteria, relative error of output parameters for mechanical ventilators and absolute error of output parameters for infant incubators was calculated. The ranges of permissible error, for both groups of devices, are regulated by the Rules on Metrological and Technical Requirements published in the Official Gazette of Bosnia and Herzegovina No. 75/14, which are defined based on international recommendations, standards and guidelines. All ventilators and incubators were tested by etalons calibrated in an ISO 17025 accredited laboratory, which provides compliance to international standards for all measured parameters.The results show that 30% of the tested medical devices are not operating properly and should be serviced, recalibrated and/or removed from daily application.

Dynamical model of Tuberculosis-Multiple Strain Prediction based on artificial neural network

This paper presents implemented artificial neural network (ANN) for diagnosing pulmonary tuberculosis progression and dynamics. Tuberculosis is an infectious disease caused in most cases by microorganism, called Mycobacterium tuberculosis. Tuberculosis is a huge problem in most low-income countries, and also in the Balkan region. The design of the artificial neural network is based on two strains of tuberculosis bacteria and multiple strains of tuberculosis. Training data sets contain 1000 reports for this artificial neural, 800 of them are used for estimation and 200 for validation. The ANN system is validated on 1400 patients from the Clinical Centre University of Sarajevo in the two years period. Out of 1315 patients, 99.24% are correctly classified as tuberculosis related patients. System was 100% successful on 85 patients were diagnosed with normal lung function. Sensitivity of 99.24% and specificity of 100% in tuberculosis classification are obtained. Our artificial neural network is a promising method for predicting diagnosis and possible treatment routine for tuberculosis disease.

Diagnosis of Epilepsy from EEG Signals Using Global Wavelet Power Spectrum

Epilepsy diagnosis using EEG signals represents important segment in general clinical practice. However, EEG signal features such as amplitude, are not helpful in order to visually distinguish between healthy and epileptic patients; and therefore additional signal processing and results analysis is needed. In this paper, the analyses and the results of the properties of EEG signals of healthy subjects and patients with an epileptic syndrome without seizure, using global wavelet power spectrum (GWS) are presented. The results of the analysis of the 200 EEG signals confirm that this approach can enable a simple recognition of epileptic EEG signals in a standard clinical practice. The results indicate that the magnitudes of the EEG signal components for the patients with an epileptic syndrome are considerably different to the EEG signal components of the healthy subjects. Also, the GWS dominant values for selected signals of patients with an epileptic syndrome are found in the delta and theta frequency bands.