FORMATION AND STUDY OF FUZZY MODELS FOR DETERMINING THE LEVEL OF HUMAN ANXIETY
Authors:
1.
Kazanskiy nacional'nyy issledovatel'skiy tehnicheskiy universitet imeni A.N. Tupoleva
2.
Kazan National Research Technical University named after A.N. Tupolev
Type:
Article
Pages:
from 102
to 106
Status:
Published
Received:
31.01.2026
Accepted:
31.01.2026
Published:
31.01.2026
Language:
Russian
Keywords:
HUMAN MENTAL STATE, ANXIETY LEVEL, NEURO-FUZZY MODEL, FUZZY NEURAL NETWORK, DATA SET, KNOWLEDGE BASE, FUZZY PRODUCTION RULES
Abstract:
This article addresses the problem of determining a person's anxiety level based on the development and analysis of fuzzy models. Elevated anxiety negatively impacts cognitive functions, sleep, performance, emotional state, social relationships, and physical health. Therefore, regular self-monitoring and anxiety assessment allow for the timely detection of changes in the psycho-emotional state and the necessary measures to stabilize it. An approach to developing fuzzy models for analyzing a person's mental state is proposed. Anxiety level is chosen as the target indicator for analysis. A neuro-fuzzy system is used to develop fuzzy models. Its use required obtaining and preparing data for analysis, specifying the number of fuzzy gradations of input neurons in the fuzzy neural network, training it, and developing and evaluating the fuzzy model. The publicly available "Social Anxiety Dataset," hosted on the Kaggle platform, was selected for training the fuzzy neural network and developing the fuzzy model. A dataset of over 11,000 records includes 18 input features and 1 output feature (anxiety level, expressed as a number from 1 to 9). Of the 18 input features, 6 were selected as the most significant based on correlation analysis. Outliers in the dataset were then identified and eliminated. The resulting data was used to develop and test fuzzy models for determining human anxiety levels. In the first stage of the study, the optimal number of three output classes (low, medium, and high anxiety) was determined. Experiments were then conducted to determine the effect of the number of input variable gradations on classification accuracy. The best classification results were achieved using five fuzzy gradations. In the final stage of the study, experiments were conducted to evaluate the accuracy of the constructed fuzzy model when working with noisy data. The results confirmed the model's robustness to noise and input data variability, as well as its applicability in real-world conditions where input values may be inaccurate. As a result of the research, a fuzzy model with a classification accuracy of 83.99% on the test data set was constructed. The model demonstrated a high level of readiness for practical use for preliminary self-diagnosis of a person's psycho-emotional state.
This article addresses the problem of determining a person's anxiety level based on the development and analysis of fuzzy models. Elevated anxiety negatively impacts cognitive functions, sleep, performance, emotional state, social relationships, and physical health. Therefore, regular self-monitoring and anxiety assessment allow for the timely detection of changes in the psycho-emotional state and the necessary measures to stabilize it. An approach to developing fuzzy models for analyzing a person's mental state is proposed. Anxiety level is chosen as the target indicator for analysis. A neuro-fuzzy system is used to develop fuzzy models. Its use required obtaining and preparing data for analysis, specifying the number of fuzzy gradations of input neurons in the fuzzy neural network, training it, and developing and evaluating the fuzzy model. The publicly available "Social Anxiety Dataset," hosted on the Kaggle platform, was selected for training the fuzzy neural network and developing the fuzzy model. A dataset of over 11,000 records includes 18 input features and 1 output feature (anxiety level, expressed as a number from 1 to 9). Of the 18 input features, 6 were selected as the most significant based on correlation analysis. Outliers in the dataset were then identified and eliminated. The resulting data was used to develop and test fuzzy models for determining human anxiety levels. In the first stage of the study, the optimal number of three output classes (low, medium, and high anxiety) was determined. Experiments were then conducted to determine the effect of the number of input variable gradations on classification accuracy. The best classification results were achieved using five fuzzy gradations. In the final stage of the study, experiments were conducted to evaluate the accuracy of the constructed fuzzy model when working with noisy data. The results confirmed the model's robustness to noise and input data variability, as well as its applicability in real-world conditions where input values may be inaccurate. As a result of the research, a fuzzy model with a classification accuracy of 83.99% on the test data set was constructed. The model demonstrated a high level of readiness for practical use for preliminary self-diagnosis of a person's psycho-emotional state.
Keywords:
HUMAN MENTAL STATE, ANXIETY LEVEL, NEURO-FUZZY MODEL, FUZZY NEURAL NETWORK, DATA SET, KNOWLEDGE BASE, FUZZY PRODUCTION RULES
HUMAN MENTAL STATE, ANXIETY LEVEL, NEURO-FUZZY MODEL, FUZZY NEURAL NETWORK, DATA SET, KNOWLEDGE BASE, FUZZY PRODUCTION RULES
