Artificial Intelligence, also known as AI, is as ubiquitous as the air we breathe. We trust AI to suggest movies to us, to correct spelling errors in our email or Slack messages, and to help guide us home through GPS.
In other words, we have come to place unequivocal trust in AI.
And why not? In many ways, AI outperforms humans in intellectually demanding tasks. The smartphone in your pocket can access an AI which can trounce the strongest grandmasters in a chess game.
Ok, AI can move pieces around a board and suggest the best restaurants for me to try out. So what?
Well, it turns out AI is capable of doing so much more. In fact, physicians and researchers are deploying more and more AI technology to advance medicine!
What does this technology look like and to what extent is it developed? Let’s dig a little deeper with some examples.
AI Solves a Protein Folding Problem
Protein folding is the process by which a protein structure takes shape. To function as expected, proteins must be folded correctly as they are being formed. Abnormally folded proteins can cause adverse medical conditions such as Parkinson’s disease and Alzheimer’s disease.
Understanding how a protein folds provides researchers the knowledge to understand the molecular processes of many of these conditions. This understanding can be pivotal to finding cures to these diseases.
While protein folding has been studied using computers for a while now, AlphaFold, an AI running on Google’s DeepMind technology, introduces an almost quantum leap in improvement relative to the current methods.
AlphaFold has learned to “predict the structure of dozens of proteins with a margin of error of just 1.6 angstroms—that’s 0.16 nanometers, or atom-sized”. AlphaFold’s achievement is noteworthy because it matches the accuracy of lab results, without the accompanying time and resources.
AlphaFold accomplishes this using deep learning, which is a concerned with algorithms inspired by the structure and function of the human brain.
AI In Medical Image Detection
There AI is being developed to assist doctors in examining medical images.
One area where AI has experienced some serious improvement recently is in detecting anomalies in medical images. Researchers from a South Korean University visualized an AI called DLAD to examine chest radiographs for abnormalities, such as potential cancers. The algorithm’s performance was compared to multiple physician’s detection abilities on the same images and outperformed 17 of 18 doctors!
Google’s AlphaFold isn’t the only effort the tech giant has produced to advance medicine. Google’s AI Healthcare team visualized an AI module, LYNA, which analyzed histology slides stained tissue samples to identify metastatic breast cancer tumors from lymph node biopsies. Amazingly, this algorithm could identify suspicious regions undistinguishable to the human eye in the samples. LYNA was tested on two datasets and was shown to accurately classify a sample as cancerous or noncancerous correctly 99% of the time. Wow!
Most importantly, when given to doctors to use in conjunction with their typical analysis of stained tissue samples, LYNA halved the average slide review time.
The less time doctors spend on reviewing slides, the more time and energy they can devote to performing procedures and caring for their patients.
Machines Versus Paperwork
The monotonous burden of paperwork is an encumbrance which doctors have to experience to practice medicine. It comes with the territory. In the United States, it’s estimated that 70% of physicians spend 10 or more hours a week on paperwork.
There’s a place here for AI, too!
As mentioned earlier, the less time doctors spend on monotonous tasks, the more time they’ll have to care for patients.
To conquer the mountains of paperwork, businesses and researchers are deploying technology infused with machine learning.
According to expert.ai, “Machine learning provides systems the ability to automatically learn and improve from experience without being explicitly programmed”. In other words, the technology is structured in a way such that it teaches itself.
So, how exactly is machine learning scaling all that paperwork? The MIT Clinical Machine Learning Group is developing machine learning algorithms to help with things like documentation, clinical decisions, and personalized treatment suggestions. This AI is being further developed to understand natural language.
Need functioning software now? Companies like Quotient Health and Ciox Health already have solutions distributed to medical offices.
