| Abstrakt: |
According to the latest World Heart Federation report for 2023, more than half a billion people around the world continue to be affected by cardio-vascular diseases (CVD), which also account for one third of the deaths in 2021 worldwide; a constant rise in the CV morbidity and mortality is observed during the years. In the last decade subtypes of artificial intelligence (AI) are increasingly used in the diagnostic, prognosis and treatment of cardio-vascular diseases. AI is an area of computer sciences, which uses different technological approaches that aim to imitate human brain processes - brain's ability to learn, accumulate knowledge and make assumptions. Different subtypes of AI are already used in cardiovascular medicine with the aim of better early detection and diagnosis of the diverse genotypes and phenotypes of CV diseases, improving the quality of patient care, improving the training of medical personnel, improving the effectiveness and, accordingly, the cost of medical care, refinement of ECG and imaging diagnostics, early detection of arrhythmias, early diagnosis of myocardial infarction or initial atherosclerosis, reduction of subsequent hospitalizations and calculation of the risk of morbidity and mortality in various clinical situations. Over the past two decades, several different types of "machine learning," "deep learning," "cognitive computing," and artificial neural networks (ANN) have been introduced to the diagnosis and treatment of CVD. In 2021, Saira A. et al. from Saudi Arabia published in the journal "Nature" an AI-based ECG reading and classification system using the Shaoxing People's Hospital database in China, which integrated their 12-lead ECG database data of 10,646 patients. In comparison, the previously introduced to public ECG-based systems had at their disposal only dozens to hundreds patient numbers. This AI-based database incorporates 11 types of main heart rates and arrhythmias and 67 additional CV rate rhythm-anomalies. Its initial version is published in an Android app for smartphone, a demo version could be found online under https://www.youtube.com/watch?v=3tfin4sSBFQ. Wang K. Et al. from Stanford University, USA, use ANN AI type for more precise assessment of heart volumes and sizes, using 2D cardiac MRT images of 500 patients and 200 controls, 30 images taken from 10 different axes creating an AI-based model of maximal precision to measure the telesystolic and teledyastolic volumes of the left ventricle. Similar AI-based ANN is used in 2015 from Yaniv B. et al. to discover CV pathology as pleural effusion, dilated hearts in X-ray thorax images. Meng L-B. et al. create AI-based ANN to find and classify the genes, which are differently expressed in patients with some form of atherosclerosis and respectively in healthy subjects. The Gene Expression Omnibus database is used. 234 potentially suspected for the development of atherosclerosis gene variations are described, mainly coding actin filaments, interconnections, sooth-muscle cells and genes, coding cytokine communication. As a total of 13 genes are determined to be the so called „hub genes"- main genes, accountable for atherosclerosis - a future potential target for treatment has the gene for tropomyosin 2. Motwani M. et al. propose an AI-based machine learning model for prediction of the mortality risk in 10,030 subjects with different types of ischemic heart disease, followed for 5 years in the CONFIRM registry. The prognoses are that in the near future with the rising use of AI in CV medicine, an enormous improvement in the precision of diagnostics and treatment is to be expected, similarly - an improvement in the quality-price ratio. The potential risks, arising from the not regulated use of different subtypes of AI are discussed in different AI forums worldwide. [ABSTRACT FROM AUTHOR] |
