The MIT Technology Review has published an article saying that for the first time a Machine Learning algorithm has matched the performance of human cardiologists.

Curb Your Enthusiasm

All science articles must be read in reverse order because that’s where the caveats lie. This article is no exception.

A sentence ends with “But it is not perfect, of course” and is followed by:

One potential problem is that the data set used here is relatively small. Machine-learning algorithms generally require huge annotated data sets to learn well. Creating bigger data sets of the heart attack recordings will be time consuming and hard. But only with bigger data sets can clinicians be sure the algorithms will be accurate in the wide range of chaotic environments that doctors work in.

How small is the data?

Moving up a few paragraphs, we see the researchers used only “148 ECG records from patients with myocardial infarction and 52 healthy controls.”

The nice thing about the article is that it also succinctly describes that the whole Machine Learning approach to solving problems as only “pattern recognition”. See below:

[I]n recent years, neural networks have made significant progress in pattern recognition problems.

When the only tool you have is a hammer every problem resembles a nail. — anon

Footnotes

MIT Technology Review, published Jul 2, 2018 https://www.technologyreview.com/s/611541/algorithm-matches-human-cardiologists-in-detecting-heart-attacks/. Accessed Jul 5, 2018 02:45am UTC.