Detection of Right Ventricular Dysfunction Using LogNNet Neural Network Model Based on Pulmonary Embolism Data Set

The high association of right ventricular dysfunction (RVD) with mortality in patients with acute pulmonary embolism (PE) remains an important health problem. In this respect, rapid, economical and highly-accurate detection of risk factors for early diagnosis of RVD in patients with PE is expected to greatly benefit the diagnosis and treatment of the disease and contribute significantly to the reduction of mortality. The aim of this study is to identify the most effective features from the PE dataset for RVD diagnosis, using a special-algorithm for the LogNNet reservoir neural-network. The cohort of patients diagnosed with acute PE in the last five years in our hospital was retrospectively analyzed and the data in accordance with our criteria were recorded. A total of 163 patients' data were acces sed and the patients had 20 characteristics. RVD was diagnosed in 27 of these patients. 78-79 years of age was found to be an important threshold for the diagnosis of RVD. The LogNNet model revealed that older age, comorbidities and coronary-heart disease greatly increased the risk of RVD. The model also found that individuals with diabetes and COPD were at higher risk of RVD, while individuals with malignancies were at lower risk of RVD. However, the model found that unilateral-thrombus increased the risk of RVD more than bilateral-thrombus. The risk of RVD is high in PE patients with unilateral-thrombus. In addition, PE patients with comorbidities such as coronary heart disease, diabetes and COPD are at high-risk for RVD and should be followed closely. © 2024, Yuzuncu Yil Universitesi Tip Fakultesi. All rights reserved.