Philips is rapidly embracing artificial intelligence, incorporating the tools into many of its medical imaging systems.
The company launched a new MRI scanner, which features a variety of AI tools, in November. In an interview, Shez Partovi, chief innovation and strategy officer, said the lightweight scanner potentially expands access to care in rural communities and reduces the need for hospitals to build expensive MRI suites.
Related: Mayo Clinic, Philips collaborate to speed up cardiac MRI with AI
The interview has been edited for length and clarity.
What impact will AI have on healthcare in the next five to 10 years?
Many physicians tell us that automation is something they're looking for, so they don't have to do the things that are not patient-directed. We focus a lot on how we can use AI to give time back to clinicians so they can spend it with patients.
When you look at the past 10 years, electronic medical records went from paper to digital, medical imaging has gone from films to digital and now pathology is going from slides to digital. You have a tsunami of data, and the industry keeps shoving more data in front of physicians and nurses, but they really want to know the meaning behind the data.
This insights-driven decision making is one of the most fundamental things that we believe will change the way care is delivered. Instead of just giving data to clinicians, we need to actually extract meaning from it and give insights to the physician so they can make the right decision at the right time.
When we think of the most pivotal thing in reducing cost and improving quality, it’s likely going to be AI.
How can AI improve access to care?
We've designed a MRI scanner that doesn't need helium refills and it means you can put it anywhere. Our BlueSeal MRI scanner is lightweight and can be put on a truck, in smaller community hospitals and in any outpatient location.
We have embedded AI in it that can help render diagnoses. For example, if a patient with Alzheimer’s gets an MRI, with our partner, Icometrix, we can look at the scan and determine if the Alzheimer's is getting better or if it's getting worse.
A traditional MRI scanner is not available everywhere in every community, let alone the radiologist to interpret this sophisticated scan of the Alzheimer's.
You have on the one hand, treatment that's available. Recently, a drug was announced for Alzheimer’s and it needs four to six MRI scans a year for that patient. On the other hand, it is an access limitation if you can't put the scanner everywhere with technology to read and interpret those Alzheimer’s scans.
How does Philips overcome issues with AI and generative AI including bias and hallucination?
In order to look at how to build really great models, there are the three V's of data. You must have a high volume of data, a variety and diverse amount of data and a veracity of data, so it’s clean and accurate. That’s how you build models that avoid bias.
We partner with health systems where we build models together that can be used in products. We focus on working with partners with large, diverse amounts of data and we make sure it's the cleanest, most accurate data.
When there's AI embedded in a technology that we build, we want to make sure the clinicians know the part of the solution that is AI.
What do you think is to come in AI?
AI is going to be in a world where we have post-market surveillance to see if there's drift and decay in the algorithm’s performance. We're already looking and working with partners on technology to help monitor algorithms post-release.
I think we're still in the early dawn of how this technology is going to deliver benefits and the entire end-to-end is going to become much more sophisticated as we go forward.
Phillips is actively involved with the National Academy of Medicine, with regulators and in every country, to make sure these algorithms do what they say they’ll do, they're unbiased, their impact is measurable and scientifically proven and they do no harm.