New building algorithm to identify problems with the heart behind the signs. Chotirokh znіmkіv disguise appeared sufficiently, that the computer algorithm with high efficiency diagnosing pathology, inform Ukr. media.

The research shows that selfies can be accessible and simple way to detect cardiovascular disease. And the significance of their visibility is to help the computer algorithm of deep learning, which analyzes photographs.

“As far as we know, this is the first robot, as it demonstrates that piece intelligence can be victorious for the analysis of features with the method of revealing heart disease,” says one of the authors of the robot, Zhe Zheng. “Our subsumkova meta is a development of a self-assessment supplement for assessing the morbidity of the diagnosis of heart disease before the clinic is recognized by people, if it lies up to a high risk group (for cardiovascular pathologies)”.

It has been known for a long time that the songs of the roses of the state of affairs are connected with the rise of the risk of heart disease. Among such signs are early sivine and scum, a fold on the urine of the ear, xanthelasma (introduction of cholesterol under the skin, sound like a sling) and lipoidny arches of the horn; folds from the infiltrate of fatty speech). However, it is easy for people to count the risk of heart disease in order to predict and assess the risk of heart disease.

The author's method of investigation was to create a tool that would allow to carry out primary diagnostics on the basis of resurveying a sign. For whom, in the period from mid-summer 2017 to mid-summer 2019, fakhivtsy conducted a treatment of 5796 patients from eight doctors in China. We tried procedures for the follow-up of blood vessels, such as coronary angiography or coronary computed tomographic angiography. Also, we made a photo of a skin patient based on a photo of a skin patient: a full-face head, two pictures in profile, as well as a view of the upper part of the head.

Based on the results of angiograms (signs of vessels) of patients, the steps of a healthy cardiovascular system were assessed, depending on whether the number of blood-bearing vessels sounded 50% or more, and also on the basis of. This information was tested for the purpose of creation, training and re-verification of the deep learning algorithm.

Later on, the researchers protested the implementation of the algorithm on 1,013 patients from nine medical clinics in China, as for medical treatment during the period from April to April 2019. It turned out that this algorithm, for its effectiveness, outperformed the known methods for predicting the risk of heart disease. Sensitivity to the method (the frequency of correctly identified positive findings in diagnosis) accounted for 80% on the validation and test selections, and specificity (the frequency of correctly identified negative findings) accounted for 61% and 54%.

aspect. net.