With an increasing shortage of radiologists, the debate around the use of artificial intelligence (AI) and the impact it could have on increasing efficiencies in radiology care is becoming more pertinent. Simon McGuire, health systems lead, Philips UK and Ireland – who spoke at this year’s National Conference for Radiology Managers – offers his thoughts.
What is your opinion of AI?
AI is a big topic of discussion across our industry. The systematic shift it presents could revolutionise the way we diagnose, treat and care for patients.
The AI premise is exciting and it’s easy to get swept up in conversations about some of the hypothetical outcomes. As an innovation-led organisation we would be remiss to not consider the future. However, we also make conscious decisions to approach new technologies with pragmatism and ask how we can use it to improve care today, through specific applications that can be concretely linked to someone living a healthier life, for longer.
So rather than the term Artificial Intelligence, shorthand for technology and machine learning, at Philips we talk about Adaptive Intelligence which we define as combining AI technologies, such as those that can prioritise data, with clinicians’ expertise. This is because we want to acknowledge that AI technology alone can’t solve every clinician and patient problem.
With all this in mind I think AI is vital for supporting an industry that is seeing increasing patient demand amid recruitment challenges. For example, radiography vacancies currently sit at 15% and 12.5% for radiologists despite a 25% increase in patient referrals for diagnostic tests compared to five years ago.
What experience do you have in implementing ‘Adaptive Intelligence’?
When AI is introduced, it’s important to plan for progress in manageable steps to ensure the infrastructure is in place to really capitalise on the full potential of these technologies.
Philips has partners in the UK and around the world where we are examining and documenting the positive impacts of the digitisation of health data combined with AI. Since the end of 2018, we have been working with Glasgow’s Queen Elizabeth University Hospital as part of project iCAIRD (Industrial Centre for Artificial Intelligence Research in Digital Diagnostics) to fast track the digitisation of NHS pathology samples. Making this type of information digitally accessible to researchers and physicians could support innovations in cancer across the UK and improve current practices. This is something we have already explored in Oxford at smaller scale, where AI combined with digitised pathology information is being used by clinicians to support earlier and more accurate diagnosis for patients at Oxford University Hospitals (OUH) NHS Foundation Trust.
Likewise, we are using AI technologies to streamline and automate prostate treatment planning workflows because planning workflows can be very labour intensive and therefore, to manage the growing number of cancer patients, we need to increase efficiencies without compromising the quality of the care we deliver. One way to do this is by saving time on repetitive routine tasks.
Our RTdrive MR Prostate technology automates a number of processes from imaging to treatment planning, such as creating digital images of at-risk organ contours and then auto-exporting them. This is quicker than using manual methods and eradicates one source of human error directly from the process.
Our hope through projects such as iCAIRD is to enable more and more hospitals to experience the benefits of digitisation. The more time you save in the sharing of important information across teams and by automating repetitive, more process driven, tasks, the more time physicians have to provide a better quality of care. This could be spending more time describing treatment procedures so patients feel more at ease, or focusing more time on complex cases to ensure that first-time right diagnosis.
What is your perspective on the future of AI within healthcare?
So far, Philips has been focusing on advancing the infrastructure that will enable the digital transformation for our health services. However, long term, AI could create a huge leap in advancing precision diagnostics, through a concept that I, along with others, believe is the future of healthcare – the ‘digital twin’.
As a society, we are already collecting data from doctor appointments and hospital visits, via health apps and wearable devices. In a 10-year period, as much as eight trillion bytes of data will be collected from an average patient. By harnessing this huge pool of meaningful data across radiology, pathology and genomic data, among other sources, we could create a digital replica of ourselves. Not only could that let us view our heart rate, for example, but also examine how our heart rate impacts on our overall health, or predict a life-style related occurrence before it happens. This will enable us to take greater control and action to prevent lifestyle-related conditions. In the hospital setting, clinicians could test treatment options on this digital model to predict outcomes for even the rarest diseases and use the ‘digital twin’ to measure the impact of surgical procedures in real time.
Developing a ‘digital twin’ is not confined to the healthcare space. In fact, a number of industries are already using the technology to pilot processes, from improving the customer experience within retail to exploring next generation aircrafts for NASA. Therefore, the huge benefits of this type of technology care already becoming clear. We just need the building blocks in place to get us there.