Mieczysław Jagodziński

The future paradigm of early cardiac diagnostics is shifting the focus towards heart attack preventive medicine based on non-invasive medical imaging with the support of artificial intelligence. It is necessary to preventively detect its increased risk early and respond with preventive drugs before moving on to more effective, but also more invasive, forms of therapy. The main motivation of our study was to improve existing and develop new AI-based solutions for cardiac preventive medicine, with particular emphasis on the prevention of heart attacks. This is due to the fact that the epidemic of lifestyle diseases (including cardiologic ones) has been stopped but not reversed; hence, automatically supervised prevention using AI seems to be a key opportunity to introduce progress in the above-mentioned areas. This can have major effects not only scientific and clinical in nature, but also economic and social. The aim of this article is to develop and test an AI-based tool designed to predict the occurrence of a heart attack for the purposes of preventive medicine. It used the combination and comparison of multiple AI methods and techniques to determine a personalized heart attack probability based on a wide range of patient characteristics and, from a computational point of view, determine the minimum set of characteristics necessary to do so. When applied to a specific patient, this represents progress in this field of research, resulting in improvements in preclinical care and diagnostics, as well as predictive accuracy in preventive medicine. After an initial selection based on the authors’ knowledge and experience, four solutions turned out to be the best: linear support vector machine (Linear SVC), logistic regression, k-nearest neighbors algorithm (KNN, k-NN), and random forest. A comparison of the models developed in the study shows that models based on logistic regression proved to be the most accurate, although their predictive value is moderate, but sufficient for the initial screening diagnosis—selecting patients who require further, more accurate testing. In addition, this can be performed based on a reduced set of parameters, particularly heart rate, age, BMI, and cholesterol. This allows the development of a prevention strategy based on modifiable factors (e.g., in the form of diet, activity modification, or a hybrid combining different factors) combined with the monitoring of heart attack risk by the proposed system. The novelty and contribution of the described system lies in the use of AI for a widely available, cheap, and quick predictive analysis of cardiovascular functions in a group of patients classified as at risk, and over time in all patients as a standard periodic examination qualifying them for further, more advanced diagnosis of heart diseases.

Unmanned aerial vehicles (UAVs), also known as drones, are increasingly common and popular due to their relatively low prices and high mobility. The number of areas for their practical applications is rapidly growing. The most promising are: last-mile delivery, emergency response, the inspection of technical devices and installations, etc. In these applications, it is often necessary to solve vehicle routing problems, formulated as a variant of the vehicle routing problems with drones (VRPD). This study presents a proactive approach to a modified and extended VRPD, including: the dynamic selection of drone take-off points, bidirectional delivery (delivery and pick up), various types of shipments, allocation of shipments to drones and drones to vehicles, the selection of the optimal number of drones, etc. Moreover, a formal model of constraints and questions for the extended vehicle routing problem with drones (EVRPD) and exact and approximate methods for solving it have been proposed. The proposed model can be the basis for supporting proactive and reactive decisions regarding last-mile delivery, particularly the selection of the necessary fleet, starting points, the identification of specific shipments that prevent delivery with available resources, etc. The study also includes the results of numerous computational experiments verifying the effectiveness of the implementation methods. The time to obtain a solution is at least 20 times shorter for the proposed DGA (dedicated genetic algorithm) than for the mathematical programming solvers such as Gurobi or LINGO. Moreover, for larger-sized data instances, these solvers do not allow obtaining any solution in an acceptable time, or they obtain worse solutions.

The paper presents a method based on algebraic-logical meta-model for solving flow-shop scheduling problem with time limits, the quality control, removing of manufacturing defects on an additional repairing machine and re-treatment of job in technological route. The appearance of the defect is unexpected event and would influence the total job processing time when the schedule is executed. Since a priori quality control results are unknown we deal with stochastic uncertainties. To be exact, the job processing time is stochastic. Flow shop scheduling problems are NP-hard and tend to become more complex when unexpected events are taken into consideration. The presented method is called a switching of algebraic-logical models. The switching method allows to present a problem by using models of subproblems and switching function. The switching function specifies the rules of using models of subproblems and these models are simplest then a model of initial problem. This method allows to solve dynamical nondeterministic problems.