Google is spearheading yet another interesting application of artificial intelligence: to predict the cardiovascular health of an individual. If this happens, it will be a breakthrough in the field of health care.

The methodology

Google has developed an artificial intelligence algorithm which can tell if a person is at a risk of contracting a heart disease by scanning a person’s retina. An article was published recently in the Nature journal Biomedical Engineering describing this research. Google collected retina scans and medical data of around three hundred thousand patients. Google then trained a deep learning algorithm they have developed to predict cardiovascular risks in any patient by feeding the retina scan and medical data to the algorithm.

Deep learning algorithms are self-learning algorithms which make correlations between the data that is fed to them. Based on the retina scan images and the medical data history of the respective patient, the algorithm teaches itself if there is any recurring correlation between the features of the retina visible in the scan and the medical history of the patient.

If you are looking for Deep Learning training course in Hyderabad please visit: https://datamites.com/deep-learning-course-training-hyderabad/

This algorithm was then run on two completely independent datasets, one consisting of almost twelve thousand patients and the other of around thousand patients. The deep learning algorithm was systematically shown the retina scan of patients who had suffered a heart attack in the last five years and those who had a completely normal cardiac history. The algorithm successfully predicted seventy percent of the time the patients who have had a cardiac problem in recent history. This result was encouraging. However, a seventy percent success rate says that the deep learning algorithm is not robust enough yet to be used in day to day practical purposes.

Other applications

Same as retina scan, this methodology can also be applied to make correlations between data from a patients blood and their cardiovascular health. Depending on the blood data, Google artificial intelligence can predict heart diseases, as well as the possibility of an upcoming heart attack or heart stroke. It is also possible to tell if the patient smokes or not, their blood pressure, and their age group.

Lily Peng, an MD and  the lead researcher for this project at Google, is hoping to use artificial intelligence in other fields of medical diagnosis as well. She hopes that soon deep learning algorithms can be used for the prediction of cancer and its possible solutions.

The future of artificial intelligence in medical research

Lily Peng says that we should keep our expectations low at this point of time. A day where artificial intelligence can be used for medical diagnosis with certainty is still a few years away. The present accuracy is fairly low. However the present results show a lot of optimism for the future. Deep Learning algorithms get more and more smarter as more data is fed to them. Ms. Peng admits that the total data which has been fed to the present algorithm is still not enough.  As the number of data points fed to this algorithm increases, it will increase the accuracy of the predictions.

Datamites Institute is one of the learning institutes in INDIA for data science courses. If you are looking for AI Training in Bangalore, Hyderabad and Pune, contact Datamites. You can opt for Classroom or Online classes for AI training course.

Leave a comment