Search

Machine learning (ML) has been used in different ways in the fight against COVID-19 disease. ML models have been developed, e.g., for diagnostic or prognostic purposes and using various modalities of data (e.g., textual, visual, or structured). Due to the many specific aspects of this disease and its evolution over time, there is still not enough understanding of all relevant factors influencing the course of COVID-19 in particular patients. In all aspects of our work, there was a strong involvement of a medical expert following the human-in-the-loop principle. This is a very important but usually neglected part of the ML and knowledge extraction (KE) process. Our research shows that explainable artificial intelligence (XAI) may significantly support this part of ML and KE. Our research focused on using ML for knowledge extraction in two specific scenarios. In the first scenario, we aimed to discover whether adding information about the predominant COVID-19 variant impacts the performance of the ML models. In the second scenario, we focused on prognostic classification models concerning the need for an intensive care unit for a given patient in connection with different explainability AI (XAI) methods. We have used nine ML algorithms, namely XGBoost, CatBoost, LightGBM, logistic regression, Naive Bayes, random forest, SGD, SVM-linear, and SVM-RBF. We measured the performance of the resulting models using precision, accuracy, and AUC metrics. Subsequently, we focused on knowledge extraction from the best-performing models using two different approaches as follows: (a) features extracted automatically by forward stepwise selection (FSS); (b) attributes and their interactions discovered by model explainability methods. Both were compared with the attributes selected by the medical experts in advance based on the domain expertise. Our experiments showed that adding information about the COVID-19 variant did not influence the performance of the resulting ML models. It also turned out that medical experts were much more precise in the identification of significant attributes than FSS. Explainability methods identified almost the same attributes as a medical expert and interesting interactions among them, which the expert discussed from a medical point of view. The results of our research and their consequences are discussed.

The focus of this study, and the subject of this article, resides in the conceptually funded usability evaluation of an application of descriptive models to a specific dataset obtained from the East Slovak Institute of Heart and Vascular Diseases targeting cardiovascular patients. Delving into the current state-of-the-art practices, we examine the extent of cardiovascular diseases, descriptive data analysis models, and their practical applications. Most importantly, our inquiry focuses on exploration of usability, encompassing its application and evaluation methodologies, including Van Welie’s layered model of usability and its inherent advantages and limitations. The primary objective of our research was to conceptualize, develop, and validate the usability of an application tailored to supporting cardiologists’ research through descriptive modeling. Using the R programming language, we engineered a Shiny dashboard application named DESSFOCA (Decision Support System For Cardiologists) that is structured around three core functionalities: discovering association rules, applying clustering methods, and identifying association rules within predefined clusters. To assess the usability of DESSFOCA, we employed the System Usability Scale (SUS) and conducted a comprehensive evaluation. Additionally, we proposed an extension to Van Welie’s layered model of usability, incorporating several crucial aspects deemed essential. Subsequently, we rigorously evaluated the proposed extension within the DESSFOCA application with respect to the extended usability model, drawing insightful conclusions from our findings.

Emotions are an integral part of human life. We know many different definitions of emotions. They are most often defined as a complex pattern of reactions, and they could be confused with feelings or moods. They are the way in which individuals cope with matters or situations that they find personally significant. Emotion can also be characterized as a conscious mental reaction (such as anger or fear) subjectively experienced as a strong feeling, usually directed at a specific object. Emotions can be communicated in different ways. Understanding the emotions conveyed in a text or speech of a human by a machine is one of the challenges in the field of human-machine interaction. The article proposes the artificial intelligence approach to automatically detect human emotions, enabling a machine (i.e., a chatbot) to accurately assess emotional state of a human and to adapt its communication accordingly. A complete automation of this process is still a problem. This gap can be filled with machine learning approaches based on automatic learning from experiences represented by the text data from conversations. We conducted experiments with a lexicon-based approach and classic methods of machine learning, appropriate for text processing, such as Naïve Bayes (NB), support vector machine (SVM) and with deep learning using neural networks (NN) to develop a model for detecting emotions in a text. We have compared these models’ effectiveness. The NN detection model performed particularly well in a multi-classification task involving six emotions from the text data. It achieved an F1-score = 0.95 for sadness, among other high scores for other emotions. We also verified the best model in use through a web application and in a Chatbot communication with a human. We created a web application based on our detection model that can analyze a text input by web user and detect emotions expressed in a text of a post or a comment. The model for emotions detection was used also to improve the communication of the Chatbot with a human since the Chatbot has the information about emotional state of a human during communication. Our research demonstrates the potential of machine learning approaches to detect emotions from a text and improve human-machine interaction. However, it is important to note that full automation of an emotion detection is still an open research question, and further work is needed to improve the accuracy and robustness of this system. The paper also offers the description of new aspects of automated detection of emotions from philosophy-psychological point of view.

Deep learning methods have proven to be effective for multiple diagnostic tasks in medicine and have been performing significantly better in comparison to other traditional machine learning methods. However, the black-box nature of deep neural networks has restricted their use in real-world applications, especially in healthcare. Therefore, explainability of the machine learning models, which focuses on providing of the comprehensible explanations of model outputs, may affect the possibility of adoption of such models in clinical use. There are various studies reviewing approaches to explainability in multiple domains. This article provides a review of the current approaches and applications of explainable deep learning for a specific area of medical data analysis—medical video processing tasks. The article introduces the field of explainable AI and summarizes the most important requirements for explainability in medical applications. Subsequently, we provide an overview of existing methods, evaluation metrics and focus more on those that can be applied to analytical tasks involving the processing of video data in the medical domain. Finally we identify some of the open research issues in the analysed area.

This paper discusses the necessity of fake news detection and how selected features can show differences between trustworthy and fake news. To demonstrate this concept, we first identified a set of features, that we believe can distinguish between fake and trustworthy news. We used these features to analyse two real datasets and evaluated our results in various ways. We first used visual analysis by means of boxplots and evidenced the significance of differences by means of the Wilcox singed-rank test. As next, we used three different classification algorithms to train models for distinguishing between trustworthy and fake news using all important features. Finally, we used Principal Component Analysis (PCA) to visualize relations between identified features.

Introduction Currently, just a few major parameters are used for cardiovascular (CV) risk quantification to identify many of the high-risk subjects; however, they leave a lot of them with an underestimated level of CV risk which does not reflect the reality. Material and Methods The submitted study design of the Kosice Selective Coronarography Multiple Risk (KSC MR) Study will use computer analysis of coronary angiography results of admitted patients along with broad patients’ characteristics based on questionnaires, physical findings, laboratory and many other examinations. Results Obtained data will undergo machine learning protocols with the aim of developing algorithms which will include all available parameters and accurately calculate the probability of coronary artery disease. Conclusions The KSC MR study results, if positive, could establisha base for development of proper software for revealing high-risk patients, as well as patients with suggested positive coronary angiography findings, based on the principles of personalised medicine.

Deep neural network models have produced significant results in solving various challenging tasks, including medical diagnostics. To increase the credibility of these black-box models in the eyes of doctors, it is necessary to focus on their explainability. Several papers have been published combining deep learning methods with selected types of explainability methods, usually aimed at analyzing medical image data, including ECG images. The ECG is specific because its image representation is only a secondary visualization of stream data from sensors. However, explainability methods for stream data are rarely investigated. Therefore, in this article we focus on the explainability of black-box models for stream data from 12-lead ECG. We designed and implemented a perturbation explainability method and verified it in a user study on a group of medical students with experience in ECG tagging in their final years of study. The results demonstrate the suitability of the proposed method, as well as the importance of including multiple data sources in the diagnostic process.

The aim of this article is to present how we processed and analysed data from a logistics company using various Business Intelligence tools. The theoretical part of the article is therefore focused on defining Business Intelligence concepts and data warehouses that are relevant to the issue. The practical part of the article focuses on editing data, creating dimension tables and facts. The data were collected from the Dynafleet system and originated from a shipping company. The data provided are for the period from 2013 to 2016. We design user scenarios to help the company's manager in making an efficient assignment of drivers to planned delivery routes. The research is focused on design and creation of a logistics system based on data analytics that can continuously analyze the incoming data and generate current decision support reports. The created user scenarios have a wide range of uses and can also be helpful in assessing the performance of individual drivers and their workloads. Using a logistics system, the logistics manager can get the valuable and useful information needed to effectively operate the business.

The medical procedures for disease diagnostics are significantly demanding and time-consuming. Data mining methods can accelerate this process and assist doctors in making decisions in complex situations. In case of Parkinson´s disease (PD), the diagnostics of the initial disease stage is the primary issue since the symptoms are not so unambiguous and easily observable. Therefore, this article is focused on determining the actual stage of PD based on the data recording signals of patient´s speech using decision trees (C4.5, C5.0 and CART). Methods such as RandomForest, Bagging and Boosting were also employed to improve the existing classification models. Estimation of model accuracy was achieved by using k-fold cross-validation and validation with omission of one record (Leave-one-out). In addition, experiments were also performed to remove collinearity in data by computing the Variance inflation factor (VIF) in order to increase the accuracy of the models.

Abstract Background There is potential for medical research on the basis of routine data used from general practice electronic health records (GP eHRs), even in areas where there is no common GP research platform. We present a case study on menopausal women with hypertension and metabolic syndrome (MS). The aims were to explore the appropriateness of the standard definition of MS to apply to this specific, narrowly defined population group and to improve recognition of women at high CV risk. Methods We investigated the possible uses offered by available data from GP eHRs, completed with patients interview, in goal of the study, using a combination of methods. For the sample of 202 hypertensive women, 47–59 years old, a data set was performed, consisted of a total number of 62 parameters, 50 parameters used from GP eHRs. It was analysed by using a mixture of methods: analysis of differences, cutoff values, graphical presentations, logistic regression and decision trees. Results The age range found to best match the emergency of MS was 51–55 years. Deviations from the definition of MS were identified: a larger cut-off value of the waist circumference measure (89 vs 80 cm) and parameters BMI and total serum cholesterol perform better as components of MS than the standard parameters waist circumference and HDL-cholesterol. The threshold value of BMI at which it is expected that most of hypertensive menopausal women have MS, was found to be 25.5. The other best means for recognision of women with MS include triglycerides above the threshold of 1.7 mmol/L and information on statins use. Prevention of CVD should focus on women with a new onset diabetes and comorbidities of a long-term hypertension with anxiety/depression. Conclusions The added value of this study goes beyond the current paradigm on MS. Results indicate characteristics of MS in a narrowly defined, specific population group. A comprehensive view has been enabled by using heterogenoeus data and a smart combination of various methods for data analysis. The paper shows the feasibility of this research approach in routine practice, to make use of data which would otherwise not be used for research.

Today, there are many parameters used for cardiovascular risk quantification and to identify many of the high-risk subjects; however, many of them do not reflect reality. Modern personalized medicine is the key to fast and effective diagnostics and treatment of cardiovascular diseases. One step towards this goal is a better understanding of connections between numerous risk factors. We used Factor analysis to identify a suitable number of factors on observed data about patients hospitalized in the East Slovak Institute of Cardiovascular Diseases in Košice. The data describes 808 participants cross-identifying symptomatic and coronarography resulting characteristics. We created several clusters of factors. The most significant cluster of factors identified six factors: basic characteristics of the patient; renal parameters and fibrinogen; family predisposition to CVD; personal history of CVD; lifestyle of the patient; and echo and ECG examination results. The factor analysis results confirmed the known findings and recommendations related to CVD. The derivation of new facts concerning the risk factors of CVD will be of interest to further research, focusing, among other things, on explanatory methods.

The paper presents a new approach to effectively support the adaptation phases in the case-based reasoning (CBR) process. The use of the CBR approach in DSS (Decision Support Systems) can help the doctors better understand existing knowledge and make personalized decisions. CBR simulates human thinking by reusing previous solutions applied to past similar cases to solve new ones. The proposed method improves the most challenging part of the CBR process, the adaptation phase. It provides diagnostic suggestions together with explanations in the form of decision rules so that the physician can more easily decide on a new patient’s diagnosis. We experimentally tested and verified our semi-automatic adaptation method through medical data representing patients with cardiovascular disease. At first, the most appropriate diagnostics is presented to the doctor as the most relevant diagnostic paths, i.e., rules—extracted from a decision tree model. The generated rules are based on existing patient records available for the analysis. Next, the doctor can consider these results in two ways. If the selected diagnostic path entirely covers the actual new case, she can apply the proposed diagnostic path to diagnose the new case. Otherwise, our system automatically suggests the minimal rules’ modification alternatives to cover the new case. The doctor decides if the suggested modifications can be accepted or not.

The paper deals with the decision of small and medium-sized software companies in transition to SaaS model. The goal of the research is to design a comprehensive methodic to support decision making based on actual data of the company itself. Based on a careful analysis, taxonomy of costs, revenue streams and decision-making criteria are proposed in the paper. On the basis of multi-criteria decision-making methods, each alternative is evaluated and the alternative with the highest score is identified as the most appropriate. The proposed methodic is implemented as a web application and verified through case studies.

In this paper we report on a study of 1541 articles from three different journals (Journal of Informetrics, Information Processing and Management, and Computers and Electrical Engineering) from the period 2007–2014. We analyzed their dates of submission and of final decision to accept and investigated whether the difference between these two dates (the so-called “time to accept”) is smaller for the articles authored by the corresponding journal’s editorial board members and whether longer times to accept yield higher citation counts. The main results are that we found significantly shorter times to accept editorial board member’s articles only in Journal of Informetrics and not in the other two journals, and that articles in any of these journals that took longer to be accepted did not receive markedly more citations.

One of the high interesting areas in the distributed systems is the problem ofbuilding and running pervasive computing services for mobile users. The central point forbuilding such a services is to appropriately model the physical and data environment. In thispaper we concentrate on the data part of such an environment that would enable creating ageneral scheme for a category of flexible services for mobile users. We define the basicprofile of the users and methods, how the services should deal with the profiles. Formodeling the presence of the user in the data environment (even if he/she is off-line) theagent-based solution was chosen, so the distributed system is build as a multi-agent system.

In this paper we describe some approaches to text mining, which are supported by an original software system developed in Java for support of information retrieval and text mining (JBowl), as well as its possible use in a distributed environment. The system JBowl1 is being developed as an open source software with the intention to provide an easily extensible, modular framework for pre-processing, indexing and further exploration of large text collections. The overall architecture of the system is described, followed by some typical use case scenarios, which have been used in some previous projects. Then, basic principles and technologies used for service-oriented computing, web services and semantic web services are presented. We further discuss how the JBowl system can be adopted into a distributed environment via technologies available already and what benefits can bring such an adaptation. This is in particular important in the context of a new integrated EU-funded project KP-Lab2 (Knowledge Practices Laboratory) that is briefly presented as well as the role of the proposed text mining services, which are currently being designed and developed there.