The future of AI in the NHS - part two

Earlier this year, I completed the ‘Artificial Intelligence in Health Care’ course at the MIT Sloan School of Management. I found it to be fascinating, thought provoking and very worthwhile. Specifically, it took a subject I was generally interested in and helped me focus on the precise issues of relevance to not only my practice but the future of NHS health care overall.

In this series of blog posts, I am repurposing my written assignments as articles so colleagues in the NHS and globally can share in that thought process. You will see that, at this stage, there are more questions than answers, but hopefully together we can work to move the theoretical towards the practical.

How can AI improve disease diagnosis and patient monitoring in the NHS and what are the challenges involved?

As stated in my first post, I work as a primary care physician within the British National Health Service (NHS). The Department of Health (DH) together with the NHS are responsible for providing national health screening programs in addition to management of acute and chronic diseases. The most prominent screening programs include bowel, breast and cervical cancers, abdominal aortic aneurysm and diabetic retinopathy. The UK, like other developed nations, is facing increasing demand for healthcare, an aging population and limited healthcare resources. Social care is also witnessing increasing pressure to manage patients in their own homes and those in care homes with increasing numbers of dementia sufferers. Primary care, in turn, is responsible for diagnosing and monitoring chronic conditions (diabetes, hypertension, ischemic heart disease etc.) using less-than-ideal tools with limited capabilities.

AI is emerging as a technology that promises to help address some of the challenges above. I addressed the potential benefits of AI in managing chronic health conditions in my previous post. Here, I propose questions on the challenges of using AI technology in disease diagnosis and monitoring.

1. Is an off-the-shelf AI model suitable for processing data to help diagnose certain diseases? The NHS constantly generates, processes, and handles a significant amount of healthcare data. Therefore, I believe that an off-the-shelf AI model for processing a vast quantity of data may not be adequate. I recall an American company, Palantir, was invited during the COVID-19 pandemic to use NHS data to help the decisionmakers identify the resources required to combat the outbreak and manage the vaccination program. Palantir was successful to a certain extent in managing the task. However, this was a unique scenario. Therefore, I believe the UK healthcare system requires a very specific model to handle its data and provide AI-based solutions to the issues it faces.

2. How do we tackle the public perception that the use of AI in disease diagnosis and monitoring is not a replacement of the coveted relationship between the patients and their physicians? It is human nature to resist change. I believe the introduction of AI in healthcare represents more than change; it is a seismic shift and a new paradigm in managing healthcare. Therefore, it will likely generate significant amount of pushback from the public and, to a certain extent, from some healthcare workers – especially those who did not grow up and train in the era of heavy computer use and AI. Furthermore, medical practice is considered a traditional profession that relies on human interaction (the doctor – patient relationship). As such, there is a concern that the introduction of AI into the equation will be met with skepticism from patients and medics alike. However, I believe that demonstrating the advantages of AI, particularly managing healthcare with limited resources, can help sell the perception of AI as a mere tool that is being used, rather than as a replacement for the physician. The NHS could mount a public campaign to explain the benefits of AI and to dispel any myths.

3. How can we address the economics of constantly updating AI technology to match the return on investment? One of the limitations of AI is the actual platform of compute that executes the algorithms and allows deep machine learning. The NHS provides healthcare to a country of nearly 75,000,000 citizens and, it is no secret, that service all too often suffers with underfunding and is subject to political whims. Therefore, I can foresee the NHS struggling with the cost of constantly upgrading the compute platform to keep up with the demand of the machine learning and sophisticated algorithms without inviting the private sector. Such an invitation will generate further skepticism in the public mind, as has been the case on numerous occasions the British public became aware that the private sector is increasingly being invited to contract services with the NHS.

4. Can “invisible” technology, such as the Emerald device, be a reliable means of monitoring patients in their own environment without any wearables or sensors, while safeguarding against abuse of collected data by nefarious parties? As stated elsewhere, the UK, like most other developed countries is facing an ever-increasing demand for healthcare services with an aging population and a national focus on keeping and caring for individuals at their own home as long as possible. Technology like the Emerald device will undoubtedly revolutionize remote monitoring. However, the data it generates could be sold and used by third parties. Therefore, we must envisage safeguarding against such data use and potential abuse. In terms of boosting efficacy, one consideration could be to amalgamate the invisibles with other wearable technological devices that are commonly utilized these days, including smart watches, smart rings, smart phones, and so on. Such a combination will still maintain the principle of monitoring patients in their own home and could potentially result in more medically useful data.

5. How do we address concerns of healthcare professionals that AI is a potential threat to their own jobs and career progression? I have encountered several colleagues who have expressed dismay at the rise of AI and argued that it is a threat to their jobs. Some of them were glad to be reaching the age of retirement before AI ‘takes over’. As stated above, we can try to advance the perception of AI as a tool rather than a replacement as a way to combat this fear. However, I believe that the argument requires further support and evidence to convince the skeptics of this technology. As with the public perception, I believe that there is much to be done to produce indisputable evidence that AI in healthcare will improve detection, diagnosis, and long-term management of diseases.
   

These questions are some of the many reasons I took this course: to be equipped with the evidence and knowledge needed to engage in productive discussions with my colleagues on the potential benefits of AI. Sometimes, though, having the answers really just means asking the right questions.

Next week, I will continue my deep dive into the NHS case study, with a focus on natural language processing and data analytics in health care. In the meantime, connect with us on social media, share this article with your followers and let’s get the conversation going!

Connect with Contempo and join the discussion!