The continuum model of development and subsequent aging of the human body or a continuum of transitional health states, depending on many risk factors, is characterized by the absence of clear boundaries between normality and pathology, that is, in a mathematical sense, we are talking about a continuous function that determines the health state of the developing organism and depends on many arguments. In the process of life, a regular reassessment of the relative role and influence of both individual factors and their combinations should be carried out, taking into account the interference of positive and negative influences of external factors. The purpose of constructing a dynamic mathematical model is to analyze the transitional pathological states of the body over time and evaluate the effectiveness of protective effects of a preventive nature. The continuum (quasi-continuum) of transitional states of an organism is considered as a random process with discrete states. The creation of a model of transitional health states meets the goal of managing personal health by preventing or delaying the manifestation of diseases on the basis of targeted counter-effects on certain modifiable risk factors for the development of pathological processes. A model is considered that includes imperatives as restrictions and operators as an abstraction of measures aimed at changing the patient’s condition. This model is planned for inclusion in an intelligent recommendation system.

The article examines a method for constructing an expert system based on frames. Special emphasis is placed on accounting for fuzziness/uncertainty in the interpretation of linguistic variables, which is highly relevant in situations where data is ambiguous, i.e., it may indicate different stages of a disease or are in the boundary area for classified states. To solve this problem, a method is proposed that allows to form and output hypotheses to the user about contiguous situations, taking into account the presence of fuzzy initial data. The described approach is especially important for poorly structured subject areas, in example for medical diagnostics, where the accuracy of interpretation of indicators directly affects the quality of decisions. In addition, it is important that the processing of linguistic variables makes the system more intuitive for the doctor and, accordingly, increases the confidence in it when it is introduced into practice. The system is tested on the task of diagnosing the stages of progression of Duchenne muscular dystrophy, which belongs to rare diseases. The advantage of the proposed approach is its universality, it can be adapted for other medical tasks in which there is uncertainty in data and decision making. The results of the study may also be useful for the development of knowledge-based systems in other subject areas where uncertainty in data occurs.

A particularity of intelligent recommender systems in the domain of medicine is the need to take into account a diversity of numerous features, and the limitation is conditioned by the need to control recommendations made by a physician. Recommendations are not directly transferred to the user because it is necessary to ensure the safety of the patient in their fulfillment. The absence of health deviations unknown to the individual at the input of the system can cause irreversible consequences. This must be taken into account in the architecture of recommender systems for health protection.

The paper presents an analysis of the main approaches to creating artificial intelligence systems-an approach based on knowledge and data. The main advantages and limitations of each approach are highlighted. It is noted that, despite the great popularity of the data-driven approach, researchers pay insufficient attention to the creation of methods and approaches for working with small data. The important role of expert knowledge for the creation of intelligent systems is noted. The paper argues that it is the integration of the two approaches that is promising for the successful solution of a wide range of intelligent problems and will allow solving the problems of explainability for different levels of end users of intelligent systems.

The number of genetically related diseases is over 7000, of which most are extremely rare. Early diagnosis of these diseases is associated with several difficulties, including a variety of clinical forms, polymorphism (multivariation) of phenotypic manifestations, and lack of personal experience in observing patients with these pathologies. The use of physician-assisted computer systems may allow to overcome these difficulties. For this purpose, an expert system was developed to support diagnostic decisions in hereditary lysosomal diseases. Knowledge extraction took place in two stages—from literature sources and from experts. The knowledge base is implemented on a cloud platform in the form of an ontological network. The mathematical model of the disease allows a complex assessment of the signs based on, the modality coefficient and confidence measures of manifestation and degree of expression suggested by the experts. The comparative analysis algorithm compares the new case to the reference variants of the known clinical forms of the integral model and then ranks the hypotheses put forward. The explanation block allows to present the data that served as a basis for the hypothesis because of the features: confirming the hypothesis, missing to confirm the hypothesis, or irrelevant to the diagnosis. The results of clinical testing of the system showed high (above 88

The article discusses the principles of constructing an intelligent recommendation system for monitoring critical infrastructure operators that is focused on assessing their psychoemotional state in real time and periodically assessing risk factors for chronic diseases. Information flows of different types of information and modules for preprocessing and analysis of heterogeneous data are described. The paper presents the architecture of the system, which provides information about deviations in the condition of operators for decision makers at the facility and for medical workers.

Fuzzy situational control allows to identify current fuzzy situations in conditions of incomplete information about the state and operation of an object. The multi-stage diagnosis and treatment process involves a huge number of situations. Various transient states of an object (human body) represent its temporal extent. Situational control can and should be used at different stages of the medical and technological process. It involves identifying fuzzy risk situations that can affect the development of a pathology, diagnosing the developed disease, as well as analyzing data from monitoring systems to detect pre-critical conditions and providing guidelines for object (human) exposure to transform a dangerous situation into a new, more favorable one. This paper examines an approach to designing a situational management system for the medical and technical process. We present a variant of the patient's clinical pathway in a traditional setting and under conditions where a complex medical and technological process system potentially operates.

Artificial intelligence technologies are increasingly being applied in a variety of medical disciplines. After reviewing 278 publications from 1985 to 2023, 99 articles were selected from the databases elibrary, PubMed, Medline, WoS, Nature, Springer, and Wiley J Database to present the main approaches and a modern picture of the application of artificial intelligence methods and technologies in pediatric surgery and intensive care. The article examines many facets of artificial intelligence systems for medical uses, namely, computer decision support systems or supporting the surgeon throughout the surgical intervention procedure. Computer analysis of 3D visualization and 3D anatomical modeling of images obtained from computed tomography and magnetic resonance imaging investigations can be used to plan operations. Because of the possibilities of sufficiently accurate 3D models and methods for organs and pathological processes, various methodologies and software tools for preoperative planning and intraoperative support of surgical intervention have been developed. Computer (technical) vision analyzes high-quality medical images and interprets them in multimodal three-dimensional images for computer diagnoses and operations under visual control, including augmented reality methods. Robotic surgery involves manipulators, including remotely controlled ones, and intellectualized complexes that independently perform specific actions of the second assistant surgeon. In intensive care, artificial intelligence technologies are being investigated to merge data from bedside monitors and other information about patients conditions to identify critical situations and control mechanical ventilation. Simultaneously, several obstacles impede the adoption of artificial intelligence in surgery. The nature and standardization of the initial data required for their integration, taking into consideration atypical cases, the possibility of bias in the sample used, and the transparency of the decision-making process in machine learning models are examples. The explanation of solutions presented in machine learning models and the transition to full-fledged validation of the systems being built define the prospects for developing and using artificial intelligence systems.

The widest range of scientific interests of Dmitrii Aleksandrovich Pospelov included numerous problems of artificial intelligence of a methodological and applied nature, management of large systems, the search for new unconventional approaches in computer architecture, and much more. Artificial intelligence is considered as a synthetic science at the intersection of computer science, applied mathematics, systems theory, control theory, logic, philosophy, psychology, and linguistics. To make decisions in intelligent systems, he proposed deductive, inductive, and plausible models that take into account the peculiarities of human reasoning. Consideration of the gyromat as an elementary model of expedient behavior, capable of adapting to the conditions of the problem being solved, was significantly ahead of the multiagent systems that appeared later. In technical systems, he considered it necessary to use various ways of organizing human activity. He analyzes cognitive graphics in the context of correlating texts and visual pictures through a general representation of knowledge. He draws attention to the role of images in human decision-making and the need to reflect them in intelligent systems. Of particular importance was the development of pseudo-physical logic to describe human perception of processes occurring in the real world, which can be represented, in particular, by the logic of relations and logic on fuzzy metric and topological scales. They showed that the semantics of operations on expert assessments on scales strongly depends on the context, while scales are formative models of the world. In applied semiotics, he examines the issues of using signs and sign systems in systems for representing, processing, and using knowledge in solving various problems; such semiotic systems are open, focused on working with dynamic knowledge bases, and implementing various aspects of the logic of reasoning. One of the most important achievements was a set of methods for constructing control systems, which are based on semiotic models for representing control objects and describing control procedures. Pospelov’s foresight of the future of artificial intelligence and the identification of growth points are reflected in many publications, and the modern development of artificial intelligence confirms much of what he outlined. As a science organizer, he led international projects, headed the UNESCO International Laboratory for Artificial Intelligence, was co-director of the International Basic Laboratory for Artificial Intelligence, and organized numerous international and European conferences.

In the rapidly developing artificial intelligence, the explainability of the proposed hypotheses and confidence in the outstanding solutions remain important problem areas. The article discusses various approaches to explainability for users of the recommendations of computer systems that they receive. The differences in the concepts of transparency and explainability are pointed out. The concepts of interpretation of results in a formal form and meaningful explanation are compared. Particular attention is paid to the need for a directed explanation for users of different levels of decision-making. The problem of trust in artificial intelligence systems is presented from various positions, which should collectively formulate the integral trust of users to the solutions obtained. Briefly, promising areas of development of artificial intelligence discussed at a Russian conference on artificial intelligence are indicated.

Aim. To improve the efficiency of diagnosis of hereditary lysosomal storage diseases using an intelligent computerbased decision support system.Materials and methods. Descriptions of 35 clinical cases from the literature and depersonalized data of 52 patients from electronic health records were used as material for clinical testing of the computer diagnostic system. Knowledge engineering techniques have been used to extract, structure, and formalize knowledge from texts and experts. Literary sources included online databases and publications (in Russian and English). On this basis, for each clinical form of lysosomal diseases, textological cards were created, the information in which was corrected by experts. Then matrices were formed, including certainty factors (coefficients) for the manifestation, severity, and relevance of signs for each age group (up to 1 year, from 1 to 3 years inclusive, from 4 to 6 years inclusive, 7 years and older). The knowledge base of the expert system was implemented on the ontology network and included a disease model with reference variants of clinical forms. Decision making was carried out using production rules.Results. The expert computer system was developed to support clinical decision-making at the pre-laboratory stage of differential diagnosis of lysosomal storage diseases. The result of its operation was a ranked list of hypotheses, reflecting the degree of their compliance with reference descriptions of clinical disease forms in the knowledge base. Clinical testing was carried out on cases from literary sources and patient data from electronic health records. The criterion for assessing the effectiveness of disease recognition was inclusion of the verified diagnosis in the list of five hypotheses generated by the system. Based on the testing results, the accuracy was 87.4%.Conclusion. The expert system for the diagnosis of hereditary diseases has shown fairly high efficiency at the stage of compiling a differential diagnosis list at the pre-laboratory stage, which allows us to speak about the possibility of its use in clinical practice.

Diagnosis of orphan (rare), in particular hereditary diseases, is associated with difficulties due to the diversity of pathology and polymorphism (multi-variance) of phenotypic manifestations. This is the source for frequent errors at the stage of primary diagnosis. In this regard, there is a need to improve the accuracy of differential diagnosis at the pre-laboratory stage. To support medical decisions, it is advisable to use intelligent systems with developed knowledge bases. The rarity of hereditary diseases is the basis for the use of expert knowledge. In the system of hereditary lysosomal storage diseases, knowledge extraction was two-stage. At the first stage, knowledge about the clinical manifestations of diseases was extracted from literary sources. At the second stage, expert determined the confidence measure in various attributes (characteristics) of signs. The knowledge base is implemented since a fuzzy disease model that uses ontologies to integrate diverse information and includes a comprehensive expert assessment of signs (modality, manifestation, degree of expression). Based on the comparative analysis algorithm, the new case is compared with the reference variants of the integral model corresponding to diagnostic hypotheses, including their subsequent ranking. The explanation block allows to present the data that served as the basis for the hypothesis based on signs confirming the hypothesis put forward, missing to confirm the hypothesis, or not related to the diagnosis. The expert system is implemented in the ontological environment of a specialized cloud platform. The results of clinical testing of the system showed a high (above 85%) efficiency of differential diagnosis.

Role-play promotes a deeper understanding of the methods of building knowledge-based systems. Students are trained to interview experts to extract knowledge for intelligent systems and, at the same time, to argue their opinion. Role-play allows you to master the necessary skills of a knowledge engineer (cognitologist). Parsing errors allows you to point out opportunities that were not used when extracting knowledge. The role-playing game can involve not only two participants (student-"cognitologist" and student-"expert", but all members of the student group. The teacher has ample opportunities to adjust the course of the role-playing game.

Authors present an expert system based on a model with operators that characterize various states of chronic cerebral ischemia. Risk factors (or predictors) that can affect the health, are considered operators in the model. The system of interactions between arguments and counterarguments measured by expert assessments characterizes the changes in the state of the body. Predictors are arguments and protectors are counterarguments. The production rules of the knowledge base are formed on the basis of attribute combinatorics, where predictors and protectors are regarded as attributes. The expert system, which includes multiple combinations of predictors and protectors, will make it possible to form personalized prognostic hypotheses for patients at different periods of time with regard to the changes in risk factors. The coefficients may be re-evaluated if required when new knowledge emerges. Expanding the knowledge base is possible by creating rules that include new factors.

Modern decision support systems (DSS) should not work autonomously, but be embedded in information systems (particularly in healthcare, in electronic health records). This will allow you to extract the necessary data in automatic mode and then supplement them in a dialogue with the user. In dynamic systems, when monitoring certain indicators, it is necessary to process the data flow in real time, taking into account the boundary conditions. This will allow providing situational decision-making control, which is especially important in emergency conditions. In addition, there is a need to feed back the formed hypotheses to the information system and explain the proposed solutions. At the same time, information that was obtained additionally in a dialogue with the user of the intelligent system must be transmitted to the database and recorded in certain fields.

Objective. To create a computer support system for pre-laboratory diagnosis of hereditary metabolic diseases with mental pathology. Material and methods. The authors used numerous literary sources to extract data for intelligent system. The initially formed database was supplemented by expert assessments for certain categories of age groups of children. Results. The experts determined the modality, confidence factors for the timing of manifestation and severity of signs for each clinical form at certain age periods. In the course of the study the authors built the models for a comprehensive assessment of signs and an integral assessment of diseases. The diagnostic algorithm is included in the database. It is used to compare the proposed hypotheses. The authors implemented a prototype of an expert diagnostic system based on a mucopolysaccharidosis model, which demonstrated 90% efficiency in a control sample of 20 patients.

This article proposes an approach to a comprehensive assessment of expert knowledge with using the model. Implemented the ability to account for a fuzzy and incomplete clinical picture of diseases. Based on the hypotheses, differential diagnostic series and comparison of reference models with personal models of new cases are formed, that allows to rate the degree of similarity and identify the disease. A comparative analysis of diagnostic hypotheses was carried out using special algorithms. The study was carried out on the mucopolysaccharidoses as an example, which belong to the class of orphan inherited lysosomal diseases.

So far, the concept of image row or tuples in the development of intelligent systems has been discussed in relation to the role of phenotypic (external) manifestations of diseases in diagnostics. This study introduces the idea of neuroimaging tuples as a tool to make a prognosis of the course of chronic cerebral ischemia. The phenomenon of leukoaraiosis is analyzed as a radiological feature of chronic brain ischemia and a predictor of stroke. Image tuples are formed from the results of computed tomography, computed tomography angiography, magnetic resonance imaging, of 85 patients with chronic cerebral ischemia. Native computed tomography images were processed with adaptive filtering methods. Computed tomography angiography results were processed through a vesselness filter that allows development of 3D reconstructions of vasculature in leukoaraiosis areas. The problem of fuzzy images, the principles of comparative analysis of images and the possibility of using confidence factors in the image tuples are discussed in the article. A scheme of a hybrid intelligent system that combines traditional logic-linguistic rules and images based on primary information and reconstruction of the original DICOM images in the knowledge base was developed. The sphere of the application is stroke risk prediction using an intelligent system.

The fuzzy of symptoms (including visual images), representations and assessments in medicine correspond to the peculiarities of the picture of the world of the patient and the physician taking into account the influence of reflection. The continuum of intermediate characteristics of the signs creates serious difficulty for their assessment by physicians. Experts’ confidence factors not only for linguistic features, but also for visual images can help increase the hypothesis quality in intelligent medical diagnostic systems.