Self-Learning AI Robot Beats Doctors at Predicting Heart attack - Can Replace Cardiologists!


Usually health systems depends upon the doctors for the Prediction of Patient Health, As it requires Manipulatioon of the medical knowledge to do Correct diagnosis or the Prediction. Human body is complex enough to develop an algorith or software to detect Upcoming or current Health Issues. Especially heart attacks are very hard to detect.
Now scientists had shown that computers are capable of teaching themselves the medical science. These computers can perform better than standard medical guidline, Making heart attacks more predictable by machines.
These AI self learnt medical standard Implemented in General Life it can save lives of millions of people per year.

Elsie Ross, a vascular surgeon at Stanford University in Palo Alto, California

I can’t stress enough how important it is, and how much I really hope that doctors start to embrace the use of artificial intelligence to assist us in care of patients.

Death Caused by Cardiovascular Problems Worldwide

  • Every year more than 21 million people Dies because of heart and vascular diseases. Some of them dies because of Heart attack, some because of Arteries blockage and other circulatory system problems. To predict these issues there are standard guidlines by American Heart Association (aha), most of physicians worldwide follow these instructions. The main rist factors in these instructions are Age, cholestrol levels and blood pressure.

How the Study was Performed With AI ?

  1. Dr. Weng In UK, and his colleagues compared the ACC/AHA guidline with Machine Learning Algorithms.
  2. Four techniques were used - Random forest, logistic regression, gradient boosting, and neural networks. Which will create a predictive tool without Human Instructions.
  3. In this study data was from 378,256 Medical records of Patients in UK.
  4. The goal of it was to find the Pattern In the records. Which were related to Cardivascular events.

How artificial Intelligence Could Save the lives ?

  1. First the AI algorithms train their Own algorithm from the clinical data available now. They used 78% of the data (about 295,267 records). They used this data and and tried to find the patterns.
  2. Once they Got the patterns, they tried to make their own guidlines from this pattern.
  3. After that These AI health robots tested these guidlines on the remaining 22% data for the confirmation.
  4. Using the 2005 data, they predicted which patients would have their first cardiovascular event over the next 10 years, and then checked the guesses against the available 2015 records.
  5. ACA/AHA (American Heart Association) guidlines Uses 8 factors to predict the same, But The Machine-Learning AI Took into account 22 more data Points, Including
  • Ethnicity
  • Arthritis
  • Kidney Diseases,etc.
  1. All these 4 (Random forest, logistic regression, gradient boosting, and Neural Networks) AI methods performed much better than AHA guidelines.

How We can Confirm that AI Health Bots Were Performing Better ?

  • A statistic Called AUC (Area under the curve - In AUC a score of 1.0 means 100% accuracy) - In this statistic AHA Guidlines gives 0.728 Score, But the AUC score of AI robots was 0.745 TO 0.764, Which is quite High.
  • Team Reported that These neural Networks Predicted 7.6% More events than the AHA/ACC Method.
  • It was found that from 83,000 patients there were about 355 additional patients whose life could have been saved. By possible means of diet and cholestrol lowering Methods.

But here is someone who does not completly agree with it

  • Stephen Weng, an epidemiologist at the University of Nottingham in the United Kingdom -

But that’s too simplistic to account for the many medications a patient might be on, or other disease and lifestyle factors. There’s a lot of interaction in biological systems. Some of those interactions are counterintuitive: A lot of body fat can actually protect against heart disease in some cases. That’s the reality of the human body.
Weng adds - “What computer science allows us to do is to explore those associations.”

What Data scientists Says about this Work ?

Evangelos Kontopantelis, a data scientist at the University of Manchester - "This is high-quality work, dedicating more computational power or more training data to the problem could have led to even bigger gains.”

Found With AI which was not In the American Heart Association Guidlines.

  1. Many new risk factors were discovered with Machine learning algorithms which were actually Not present In AHA/ACC guidlines.
  • Several Mental Illness
  • Taking Oral Corticosteroids
  1. However Surprisingly - None of the 4 algorithms considered diabetes as a predictor, which is on the ACC/AHA list, to be among the top 10 predictors.
  2. Doctor Weng hopes to include lifestyle and genetic factors in Machine algorithm for futher accuracy.

Machine Learning Algorithms are Black Boxes - Risky.

Evangelos Kontopantelis Notes one limit to this study that is - Machine Learning algos are blind boxes for Humans, As humans see data goes in and decision comes out, But we can not find out the logic behind the decision or what happens in between. This makes the humans difficult to tweak the algos.

Will physicians soon adopt similar machine-learning methods in their practices?
Will these AI Robots Replace the Cardiologists Someday ?

What do you think about this study Please Reply US.

Doctors really pride themselves on their expertise, Ross says. “But I, being part of a newer generation, see that we can be assisted by the computer.”


There is endless potential for the AI Robots to work alongside Cardiologists and less room for error. Think of the medical breakthroughs that could come from such a partnership. Co-existence should be the goal, rather than turning it into a competition. Obviously, not everyone will agree at first but studies can speak for themselves. This is a fascinating turn point in medicine if put into practice further.

Copyright 2018