Healthcare Club Newsletter Week of 5/4 - The Rise of Healthcare AI

Today I’d like to talk about technology and healthcare. Specifically, AI and healthcare. And by ‘AI’, I mean ‘Artificial Intelligence’. And by ‘Artificial Intelligence’, I really mean ‘Machine Learning’ or ML.

First, we should clear up any confusion between ‘AI’ and ‘ML’. I spent nearly 4 years selling IBM’s AI product portfolio to a variety of customers eager to solve their business problems with the magic of AI. 90% of the time, I spent my first meeting with a customer explaining what AI actually was and debunking any misleading marketing myths about AI.

#1 AI is not an out-of-the-box solution. Utilizing AI in a business requires training and model adjustment over time.

#2 AI is not a technology; it is the idea that machines are capable of being able to carry out services and tasks in a way that we would consider ‘smart’.

#3 We achieve AI through Machine Learning.

#4 Machine Learning is the current technique being utilized to drive us towards the notion of Artificial Intelligence.

#5 Skynet is not in our immediate future.

That should cover the bases for now.

Researchers turned to using AI in attempt to tackle some of the most pressing questions around the spread and origin of Covid-19. I have selected a few such applications to highlight here:

1.       Tracking the spread of the virus– BlueDot (https://bluedot.global/)

Using Natural Language Processing and Machine Learning, Blue Dot knew about the Covid-19 outbreak before anyone knew what Covid-19 was. The software program uses hundreds of thousands of sources (public health organization announcements, digital media, livestock health reports, etc.) to detect infectious disease threats as early as possible. BlueDot was able to predict the spread of the Zika virus to Florida six months before it happened in 2016.

2.       Understanding the way Covid-19 operates – DeepMind, the AI arm of Google’s parent company Alphabet (https://deepmind.com/)

DeepMind has built a deep-learning library called AlphaFold. Much like how the human brain uses neural networks to process information, so too does this program which is built to predict how the proteins in an organism will ‘curve’ or ‘crinkle’ based on their genome. Once the structure of a protein is known, then the shape of the receptors can be determined. After training the program on large genomic datasets, researchers are hoping to use the tool to get a sense of the types of drugs that might have an effect on Covid-19.

3.       Imaging: identifying signs of infection – Darwin AI (https://www.darwinai.com/)

Darwin AI trained a program using image recognition applications to identify signs of Covid-19 in patient X-Rays. When testing kits were in short supply, this program was used as a diagnostic tool by doctors to get a better sense of what the patient was really dealing with.

And this is only the beginning. Healthcare is a highly regulated space and the primary obstacles are the ones that we all know and love…regulation, privacy, trust in the technology, etc. The industry still is in the process of figuring out how they are going to position themselves.

If you would like to learn more about applications of AI in healthcare, I highly recommend subscribing to ‘The Algorithm’ newsletter by the MIT Technology Review. They cover all industry applications but have a good assortment of healthcare-specific articles to peruse