Interview: Dr. Mathias Goyen, Chief Medical Officer – Europe, GE Healthcare

After years of talk, artificial intelligence is here now and being deployed across health care. So what will be its near-term impact? We talk to Mathias Goyen about specifically how and where it will make a difference.

HBI: When we think of AI and healthcare one tends to think about hundreds of start-ups using it to interpret medical images. Or of how Ada and others are using it as a diagnostics tool in primary care. How do you see it?

MG: I think it already goes much deeper than that. Within our equipment we are incorporating it pre-scan and in the scanning process. But as well as at this individual level, we are also deploying it in workflow at the departmental level, and at the hospital level.

Its role in diagnosis has already gone beyond primary care. As an example, in partnership with Roche Diagnostics we aim to launch a product that helps clinicians make more personalized treatment decisions for cancer patients with NAVIFY Tumor Board and we are working on another product for the early detection of sepsis, monitoring acute care patients. Within these, we are using AI to analyse in-vivo imaging and in-vitro lab tests and to suggest patient pathways and treatment options.

HBI: And how far along is all this?

MG: It is already forming part of many of our latest generation of products.

With our partnership with Roche that I just mentioned, the latest release of NAVIFY Tumor Board should go live in the coming months and we have another product that will help with the early prediction of sepsis that we also hope to launch soon.

Our AI-powered Command Centres are already in a number of hospitals in the US and we recently announced the first in Europe, at Bradford, in the UK.

And we are also deploying this within scanning equipment as mentioned.

HBI: So let’s start by looking at how you are deploying it in equipment

MG: As an example, today a radiographer spends a lot of time setting up a scan. You have to match the equipment to the anatomy of the patient, to tell the scanner the location of the patient’s brain in the bore, for instance. This requires the technologies to “set planes”, or “prescribe slices”, which is a highly manual process.  With built-in AI capabilities the equipment can automatically detect the anatomy. So we automate a lot of the set up.

That saves time and it also increases the consistency of scans – they become operator independent. So you can have direct comparisons between 2-3 images which may be 3-6 months apart. You can confidently track patient progress longitudinally.

HBI: And during the scan?

MG: Imaging blood vessels with ultrasound is complex, skilled and time-consuming. But with AI, you can press a button and AI can identify and label arteries and veins.. The system knows the difference between veins and arteries and can make exam-specific decisions. Again, it leads to consistency.

After the scan we can use AI to interpret images. If you have a collapsed lung early diagnosis is essential. But often radiologists have 6-8 hour queues of images. So we’ve incorporated AI into X-rays. Take the X-ray at 2 in the morning and it will flash red if it thinks it has spotted a collapsed lung. So the image can be prioritised. That has the potential to save lives.

HBI: Then we have AI in hospital management.

MG: Yes here we have built command centres which enable hospitals to deploy resources far more effectively. Humber River hospital in Canada was one of the first to go live here. It has 500+ beds and predictive analytics to forecast patient flow has allowed it to create capacity equivalent to more than 20 extra beds, without adding physical beds. It has cut clean-bed waiting times by 45%.

HBI: To me the most revolutionary application is your work with Roche. The aim here is to use AI and cloud computing to help doctors diagnose and manage patients, focusing first on oncology.  Now this really is a game changer, isn’t it? Because you are not only merging imaging and lab tests on the same platform but also drawing on a massive medical research database to suggest the right treatment paths for patients.

MG: Yes it is. I remember when I was a hospital radiologist and we’d have a  tumor board meeting for 20 patients and for each we would have to physically assemble the data – images, tests etc – often on paper or faxes. It would take a long time and then we would have to interpret it based on our pooled medical knowledge. With the new product NAVIFY Tumor Board, much of this can be done digitally, or automatically.

HBI: And that’s not just for oncology is it?

MG: We are working on a separate product for the early detection of sepsis in patients. AI-enabled “Virtual Collaborator”*to integrate data from the EMR and other patient-information systems and provide insights into the status and trends of patients who are at-risk for sepsis-related deterioration.

While most current applications of AI are reactive in nature, with the technology responding to questions or inputs, this starts to infer the “why” behind clinicians’ questions and serves relevant information beyond what was asked.

HBI: So all of this radically affects how doctors and other medical staff do their jobs, doesn’t it? Won’t it lead to mass redundancies?

MG: The work of the radiologist will likely change significantly in the coming years. AI will very likely become part of our daily routine in diagnosing simpler cases and performing repetitive tasks. But this is no reason to feel threatened, because the radiologist’s work involves more than just image interpretation. Radiologists sit in tumor boards, treat illnesses (think of the huge and growing field of interventional radiology) and use imaging, in combination with the patient’s history, to develop findings.

AI is helpful in operations that it has learned and simulated a thousand-fold. But as soon as something unexpected happens, it is by no means certain that the AI will make the right decision.  While we are making great advances in artificial intelligence it still sometimes requires a touch of old-fashioned real common sense! We can make AI not a force to fear, but a helping hand that makes our lives easier and healthier

*Technology in development that represents ongoing research and development efforts. These technologies are not products and may never become products. Not for sale. Not cleared or approved by the U.S. FDA or any other global regulator for commercial availability
We would welcome your thoughts on this story. Email your views to Max Hotopf or call 0207 183 3779.