Artificial intelligence is a major buzzword these days. But what is the state of AI in healthcare, specifically? The long answer is complicated and changing daily. But the short answer is simply this: AI is already in healthcare, and its presence is growing.
Contents
Deep Learning, Machine Learning, and Artificial Intelligence
So What is artificial intelligence exactly? How about machine learning? And what on earth is deep learning?
To put it simply, deep learning is a way to develop machine learning, and machine learning is a way to develop AI.
The technology website Built-In explains that “machine learning feeds a computer data and uses statistical techniques to help it ‘learn’ how to get progressively better at a task, without having been specifically programmed for that task”.
Built-in goes on to say that deep learning uses a biologically-inspired neural network architecture to process data through multiple “hidden” layers.
This ‘deep’ processing assists the machine in making connections and weighing the value of data.
These technologies are entering the world of healthcare in many different ways, but three of the main areas of application are
- Precision healthcare
- Research
- Early diagnosis
1) Precision Healthcare
The basic premise of precision healthcare is that each individual patient is different, even when they share a diagnosis.
They have unique genes, unique lifestyles, and unique environments. To treat effectively, all of those individual variables must be considered. The problem is, that’s too much data for a human to analyse.
This is where machine learning steps in. Some companies claim to have included billions of patient records in their datasets. They used that data to train an artificial intelligence that can handle human variability.
Using this method to perfect precision health care would functionally end trial and error. And while that goal may be a bit too lofty, even getting close would mean a much-improved experience for patients and professionals alike.
2) Research
Finding viable candidates for drug trials has always been tough. So too has been reaching patients who may benefit from new or little-known treatments.
The traditional process has lacked the data resources necessary to connect all possible patients with all possible providers.
As a result, patients that might be perfect candidates for these new or trial therapies never even hear about them
However, AI can sort through the mountain of data and identify the right patients, regardless of geography or resources.
In fact, multiple companies are already using AI to do just that. Patients that may have once been overlooked are now identifiable and at a fraction of the former effort.
3) Early Diagnosis
This is arguably the most exciting application of AI in healthcare. Multiple research teams have shown AI being substantially better at early diagnosis than human clinicians.
And when you think about it, it makes sense. Early diagnosis is all about observing subtle patterns, and so is machine learning.
To illustrate just how good the AI is getting when it comes to diagnosis, take a look at the developing research into mammography and breast cancer detection.
Numbers Don’t Lie
Just this month, a study was released in Lancet Digital Health. It had a sample of more than 170,000 mammogram images, drawn from South Korea, the USA, and the UK.
Its findings indicate AI is demonstrably better at detecting breast cancer than radiologists. The detection sensitivity was 90% for the AI, as opposed to 78% for radiologists.
Another study, published in Nature, showed similar results. In that case, the Google AI subsidiary Deep mind worked with multiple institutions to test a model for screening for breast cancer.
This was another study with impressive sample size. More than 91,000 women were included.
When comparing the results, the AI produced fewer false positives and false negatives. This was in spite of the fact that radiologists had access to additional information like patient histories, while the AI had only X-Rays.
Machine Learning and Wound Care
Machine learning has already changed the way that patients and medical professionals monitor healing wounds. Vohra’s AI-powered wound care predictor is a tool that employs data from millions of patients to predict wound healing rates with an 80% certainty.
This knowledge empowers patients and their healing teams to better advocate for themselves and to know when intervention is required. All for free and through the internet.
The Future of Healthcare
There’s no stopping AI’s integration into healthcare. Millions and millions of dollars have already been invested in AI projects, and many of those projects are already being implemented.
Going forward, the healthcare field will have a great deal of new technology to navigate. However, if it’s all done right, the benefits for patients and practitioners are limitless.
More Interesting:
How Blockchain is Shaping the Healthcare Industry
Augmented Reality in Medical Education and Training
Healthcare Hurdles: How to Overcome Roadblocks to Adopting AI
