Neuronet for ECG assessment

2019: Announcement

In the middle of November, 2019 the AI technology capable to predict failures of a warm rhythm was provided and to precisely predict risk of death at patients, even when independent cardiologists cannot distinguish the same risk factors.

The team of researchers of Geisinger company prepared neural network for assessment of electrocardiograms to predict risk of development of failures of a rhythm, first of all fibrillations of auricles. For training of a neuronet the results of 1.77 million ECGs from nearly 400,000 patients collected for the last 30 years in archives were used. The trained AI predicted long-term outcomes and precisely identified patients from risk group. The model could also foretell at what patients the risk of failure of a warm rhythm is increased, even when doctors interpreted test results as normal. At the same time the neuronet analyzed 15 segments consisting from more than 30,000 data points on each ECG.

the AI technology capable to predict failures of a warm rhythm was provided and to precisely predict risk of death at patients

The data collected using AI and neural networks and also from digital wearable devices, even more often are considered as the important tool helping doctors to set risk factors at patients at an early stage. Researchers hope that the AI-forecasting model of failure of a rhythm and warm death can be used at an early stage of diagnostics. In particular, identification of fibrillation of auricles will allow to reduce significantly risk of a stroke, so, and death. So far researchers work on in the best way to adapt data retrieveds to different groups of the population.

New approach can completely change a view of interpretation of the ECG in the future, - Brandon Fornwalt, one of heads of laboratory of cardiological visualization Geisinger and the senior author of researches explained.

Notes