Artificial intelligence technology could soon be a part of electrocardiogram readouts (ECG). The use of AI in healthcare is not a nascent trend. The induction of AI has overhauled several healthcare devices. A team of scientists have trained an artificial intelligence tool to predict an individual’s age and sex. The tool uses ECG readouts to predict the aforementioned characteristics. The scientists are working toward converting the primitive version of the tool to an advanced model. The advanced model shall be able to predict various healthcare parameters for individuals.
Scientists from the Mayo Clinic College of Medicine and Science, Rochester integrated convolutional neural networks (CNN) with AI technologies. Further, researchers used ECG readouts for over 500,000 individuals draw valuable inferences.
The research methodology and findings of the scientists were published in Circulation: Arrhythmia and Electrophysiology journal. The research paper describes several methods employed by the scientists for determination of age and sex from ECGs.
Analysing Samples and Deviances
Results from the former batch of 275,000 people suggested that CNN was better at predicting sex. The system exhibited 90% accuracy in predicting sex, and was 72% accurate in predicting age. The scientists then considered a sample of 100 people who had ECG readouts for 20 years or more. Further, they used the predictions of AI-based tool patients with heart conditions. The tool overestimated the chronological age of people who had suffered heart conditions. The prediction was more accurate in case of people with no history of heart diseases.
Key Inferences and Takeaways
The scientists inferred that the tool was more efficient in predicting the biological age of people. This means that the AI-based tool considered health parameters of patients while making predictions. Hence, the newly developed tool can be worked upon to make estimations related to human health.