Artificial intelligence (AI), machine learning (ML) and the cloud are already transforming healthcare by automating workflows, improving processing speeds and enhancing image quality, leading to faster, more accurate diagnoses. Seamless integration of these technologies within devices, programmes and services enables healthcare professionals to make more informed decisions, while also enhancing productivity and throughput. At present, deep learning-based algorithms account for a majority of ‘smart’ digital solutions, automatically performing image perception, recognition, segmentation or classification tasks. They are increasingly being used in radiology departments, and as we move to digitalised services, to interpret medical images – including cancers, in video colonoscopy and endoscopic ultrasound – as well as in ophthalmology.
Identifying barriers
Uptake of new technologies is obviously a real challenge in a field as closely regulated and complex as healthcare, but there are also basic logistical concerns and more complicated barriers that need to be addressed to enable more widespread adoption. The COVID-19 pandemic has also no doubt accelerated the demand for these digital solutions while, conversely, restricting the availability of resources to deploy them, further complicating efforts.
Dr Rowland Illing, Chief Medical Officer and Director of International Government Health for Amazon Web Services (AWS), explained how technology providers are helping to tackle some of the issues healthcare organisations currently face: “The main barriers are the lack of skills and access to AI and ML learning tools, as well as having adequate computing infrastructure and ensuring clinical confidence. At AWS, we have developed a range of integrated development environments to help diagnostics companies and healthcare providers address these issues, such as the Amazon SageMaker that GE Healthcare uses on its Edison™ platform. This provides access of AI and ML via the cloud, making it easier to build, train and deploy models in a stepwise manner for end users who are not familiar with the technology. Healthcare organisations can take full advantage of the capabilities – which mimic human cognition, and have voice and image recognition, natural language processing and interactive chat bots – without having in-depth technical knowledge. There are plenty of opportunities to infuse ML into operational workflows”
Dr Hugh Harvey, Managing Director of clinical digital consultancy Hardian Health, continued: “The biggest barriers are clinical validity and evidence, health economics and getting the IT infrastructure right. It’s vital to produce guidance on how we properly report a prospective efficacy and validation study. This takes time and is expensive. Fortunately, there are many excellent grant funding opportunities available but, in essence, we need to make sure the underlying technology can constantly improve in a very safe way. Most importantly, we need to demonstrate that the tools we are building can give us a return on investment, not just for a single department, but for the healthcare organisation as a whole. This needs to be assessed through proper health economics analysis, and I think this a massive challenge. We also need to ensure that there’s back-compatibility and tight integration of these technologies with potentially outdated equipment.”
Developing partnerships
Accelerating digital innovation in healthcare requires a truly collaborative environment, and this is exactly what GE Healthcare’s Edison Ecosystem does. This solution is well positioned to overcome hurdles and creates a fast track to accelerate mainstream adoption of AI and ML like never before. The holistic approach brings together stakeholders from five key areas – applications, devices, IT solutions, developer services and healthcare providers – while ensuring that the patient is at the heart of the conversation. Dr Evis Sala, Professor of Oncological Imaging, University of Cambridge, commented: “If patients endorse AI, it’s more powerful for the stakeholders in the hospital to actually accept it. Patients are more resilient and open minded than we are, so getting them involved from the beginning in a stepwise, task-by-task manner is essential. For example, we asked ovarian cancer patients how they felt about using AI to do the image segmentation, and they didn’t have a problem with that. There are, of course, a lot of other questions that we still need to ask to ensure patients are comfortable with the technology being more widely introduced in healthcare. It’s also important to understand that it takes time generate the required evidence, and that we need to continually work with vendors to develop and improve these solutions.”
Rowland added: “When patients can actually feel satisfied about the digital system, it’s a very powerful thing. We have a great company currently running in AWS called Axial3D, which makes printed models based upon the radiology, allowing surgeons to show patients exactly what an operation involves. Engaging with patients in the process is critical, and if you can demonstrate how the AI has helped, it gains greater acceptance from them.”
The real gamechanger for the future
We are at an early stage for AI adoption, with the market expected to grow rapidly over the next few years. The data generated by these technologies today is only a fraction of what will exist in the future, and hospitals need to start looking at their data – how it is currently stored and managed – as quality of this data will become increasingly important with the growth of smart digital solutions. The only way forward is if healthcare providers have the correct tools – carefully designed for the purpose – in their arsenal to extract value from their investment.
Jan Beger, Senior Director of Digital Ecosystem Europe at GE Healthcare, summarised: “Companies innovating AI in healthcare need to be prepared to invest time, effort and expertise to create models that focus on the needs of healthcare professionals and patients. Close collaboration between AI developers, providers and patients is crucial for the successful adoption of digital solutions in global healthcare settings. Large scale deployment is only possible if the technology shows a clear return on investment, both financially and through improved quality of care. The first leap will be seamlessly embedding these innovations into general hospital settings to augment and improve current medical practices, and to achieve higher levels of performance, speed and convenience. The real gamechanger over the next five years will be the prevalence of ML taking place in the cloud.”
For more information go to: https://www.gehealthcare.co.uk/products/healthcare-digital
Footnotes: Dr. Roland IIling and GE Healthcare have no specific contractual relationship.
GE Healthcare and AWS recently entered into a collaboration for the provision of cloud services. (https://www.gehealthcare.com/article/for-radiologists%2c-reading-from-home-is-the-new-working-from-home).
Dr. Hugh Harvey, Managing Director of Hardian Health is an advisory board member for EdisonTM Accelerator in EMEA
Professor Dr.Evis Sala is Chief Medical Officer and co-founder of Lucida Medical, recently announced as a cohort participant for EdisonTM Accelerator in EMEA.
