Machine learning overview
In 1959, Arthur Samuel defined machine learning as “the field of study that gives computers the ability to learn without being explicitly programmed.” Machine learning has shown superior performance compared to direct programming for a vast number of extremely complex tasks, such as spam filtering, speech recognition, and recommender systems. Machine learning is used to rank advertisements by online stores including Amazon, to improve results by search engines like Google, and also for fraud detection by financial institutions. Today, machine learning achieves close-to-human or even better-than-human performance in an increasing number of tasks such as image recognition and medical diagnostics.
Some tremendous breakthroughs that brought machine learning to the attention of the media and the public were the Watson computer system developed by IBM, which won the quiz show Jeopardy! against the human Jeopardy! champions Brad Rutter and Ken Jennings. In 2015, the computer program AlphaGo, developed by Google, defeated human champion Lee Sedol at the ancient board game of Go, a game of profound complexity.
Importance of hERG activity
The hERG (human Ether-à-go-go-Related Gene) protein is a potassium ion channel protein. It plays an essential role in regulating the electrical activity of the heart. hERG is susceptible to drug binding and inhibition of its normal activity. Some drugs that block hERG have been found to cause long QT syndrome. Correctly predicting the hERG inhibition of an investigational drug has been recognized as an important part of the drug development process. Our novel machine-learning model offers a new way to predict hERG blockage that can then be confirmed functionally in our novel human cardiomyocyte cell-line expressing hERG.