Artificial Intelligence Methods in Assessing the Severity and Differential Diagnosis of Bronchoobstructive Syndrome

Abstract

Respiratory muscles strength is the main indicator of their functional state. The study of respiratory muscles strength is becoming increasingly prevalent in clinical pulmonology, especially in case of chronic obstructive pulmonary disease (COPD) and asthma. However, respiratory muscles strength is used neither for COPD stratification nor for differential diagnosis of COPD and asthma related to the broncho-obstructive syndrome. The aim of the study was to develop models that support medical decision making in broncho-obstructive syndrome diagnostics. Material and methods. 214 patients who were hospitalized with COPD exacerbation (115 people), severe uncontrolled asthma (56 people), and their combination (43 people). Respiratory muscles strength indicators (MEP, MIP and SNIP), 9 anthropometric parameters, spirometry and blood gas parameters, modified medical research council dyspnea scale, COPD assessment test data were recorded. Data processing was carried out by means of Mann-Whitney, Fisher and Tukey tests and correlation analysis. Respiratory muscles strength models were performed by linear and nonlinear regression methods. COPD stratification and differential diagnosis of COPD and asthma models were performed by artificial neural networks. Results. Respiratory muscles strength models of healthy individuals and COPD patients allowed to estimate the effects of various factors on the respiratory muscles functional status. Comparative analysis of COPD severity verification showed that models accuracy increased when we had added a respiratory muscles strength indicator. The most informative indicators were MIP, total body mass, partial pressure of carbon dioxide and fibrinogen. Moreover, MIP increased the accuracy of all the models. Conclusion. Practical application of artificial neural networks models in telemedicine projects allows developing information services to support real-time assessment of the patient's condition.

Computer-Aided Detection of Respiratory Sounds in Bronchial Asthma Patients Based on Machine Learning Method

“…), communication means and software allows developing new alternative methods of computer-aided diagnostics based on analysis of the patients’ breath sounds. These methods include machine learning and deep machine learning (deep learning) [ 8 , 14 , 15 ]. Recently, the possibility of applying these methods in various fields of medicine was actively studied [ 8 , 14 , 16 – 20 ].…”