Adopting AI or other technology innovations can be high stakes for healthcare providers
Now that artificial intelligence is living up to the potential that future-minded commentators have touted for a long time, many healthcare providers are considering how to factor AI and big data projects into their processes to improve care and increase efficiency. However, investing in one platform or one focus area can be risky because of the pace of change. Putting millions or billions into one platform or project, which could be obsolete or fall flat in a few years is a huge risk.
Nooman Haque, Managing Director for Healthcare and Life Sciences at Silicon Valley Bank believes the industry needs “runaway successes” to drive wider global adoption.
The key issue for me is around workflow. You’re giving doctors, physicians, payers, regulators all these new tools to work with but they are very much discreet, pilot projects. They’re not fully pervasive or integrated into the workflow.
We recently spoke to Nooman about the challenges and impact in adopting AI and other technologies for healthcare providers. Listen below or on iTunes.
Silicon Valley Bank recently published their mid-year report on healthcare investments.
What’s the Impact From AI?
AI’s most productive capability, the speed at which it can recognize complex patterns in large data sets and iterate its findings, is very suited to the state of medical care, diagnosis and pharmaceutical research today. Here are ways AI is making an immediate impact in medicine
One way that AI is speeding up patient care is how it changes how we go about drug design. This process is normally very hypothesis driven, but with AI a more trial-and-error approach running vast quantities of patient and drug data together is able to generate potential solutions. This means that whereas previously medical researchers needed to have a clear understanding of exactly how a new drug design might work, now they can run AI to look for potential solutions based on vast quantities of cases of how chemical compounds have helped different patients before.
This essentially transforms the search for new formulas from a process constrained by human mental resources to a process that leverages the rapid trial-and-error capability of powerful computers. It is like being able to use a magnet while searching for a needle in a haystack.
While prosthetics might seem like an odd application of AI, researchers have begun fitting prosthetic hands with cameras that use machine learning to adjust grip patterns to be able to better handle different objects. This means that a prosthetic arm will learn the nuances of grasping over 500 different objects more securely, instead of the more two-dimensional grip mechanics that most use currently.
Doctors are using AI in a new way to leverage deep-learning in order to tailor disease treatment that focuses on the patient’s bio markers and genetics. Whereas conventionally there is a focus on deciding treatment based on all available patient data, now with AI we can cross-reference this with the patients DNA (which is known as proteomics) to narrow down the choice of best treatment.
Speed to Diagnosis
Aside from the aforementioned precision treatment, AI is also speeding up diagnosis. Stanford researchers have designed a solution to more quickly recognise cancerous moles, flagging high-risk identifiers. Cancer in particular is so complex and time-sensitive that it becomes difficult for doctors to address it quickly enough, so AI is ideal for sifting through the mountains of relevant clinical and patient data.
Health and Fitness
Cafewell is an app that could fill the role of dietitian and personal trainer. Given data about the user and their goals, Cafewell uses IBM AI algorithms to evaluate from previous user data to develop a workout and eating regime with the highest probability of success. And based on the user’s feedback, the app refines its evaluations for the next user.
Considering how complex the eye is, the power of AI is set to be useful in detecting the minor signs of the onset of blindness. Google’s DeepMimd is being used to crunch millions of data points when presented with an eye scan, helping doctors diagnose the onset of sight loss before it gets too far along.
VAs and chatbots
When you’re sick, the sooner you can get advice from your health provider the better. The way AI can help is by automating first-line medical interaction to rapidly advise patients online what steps to take based on cross-referencing their symptoms and plan with those of millions of other patients.
It’s all about data
In essence, AI is transforming healthcare by comparing large amounts of data points with millions of other similar cases. This gives both quantitative improvement by speeding up treatment decisions but also qualitative benefits as in the case of prosthetics, enabling solutions that previously did not exist.
We are still a long way off eliminating most diseases with AI. Research such as that which is going on at Stanford, Google’s DeepMind and IBM’s Watson are showing that while currently AI is mainly assisting doctors with treatment, new applications of AI are being found constantly that will soon be able to surpass what doctors can do now.