Healthcare systems around the world are already struggling to meet the demands of an ageing population, the rising incidence of chronic diseases, escalating healthcare costs and a deficit of healthcare workers.

For example, in the United States (US) approximately 50% of the population have chronic conditions, which are responsible for 86% of the country’s healthcare expenditure. In the United Kingdom (UK), NHS waiting lists continue to grow, impacting patients’ access to timely diagnosis and treatment. And according to the World Health Organization, one in six people worldwide will be 60 years or older by 2030, a year in which a global shortage of 10 million health workers is also projected.

The stark reality of the world’s current healthcare crisis is ineluctable but could Artificial Intelligence (A.I) be a panacea for healthcare’s most pressing issues?

What is A.I?

A.I utilises computers and machines to perform intellectual tasks ordinarily completed by humans. Although this may conjure thoughts akin to robots with a competency to act consciously, we’re thankfully quite far off from that reality.

Currently, all A.I systems are defined as narrow (or weak) as they are intended to perform specific tasks and create solutions in specialised areas. Generally, A.I is used to automate processes, analyse vast quantities of data and recognise patterns and trends, which can then be used to inform decisions, personalise experiences and solve problems. The application of A.I encompasses a method called machine learning. Using inputted data (the more the better) an A.I system continuously ‘trains’ itself to uncover patterns and trends for a particular task – which would take humans an inordinate amount of time.

How is A.I being used in healthcare?

In the last decade, A.I application in healthcare has gained considerable momentum due to huge technological developments. There are 240 A.I tools in the pipeline for health and social care, but only 25% are mature enough to be used and only 20% have regulatory approval. Despite this, A.I is already being tested or used in many areas of healthcare, for example:

• Within oncology, A.I is being used to provide patients with a bespoke treatment plan of all suitable options available. The entire genetic framework of their cancer is analysed, and the information is used by A.I software to recommend molecular targets most susceptible to successful treatment intervention.

• Moorfields Eye Hospital (UK) is involved in the development of A.I software to analyse eye scans and detect an array of eye conditions. As eye scans are currently examined by specialists, it can be a slow process and possibly preclude early treatment.

• The humble stethoscope is undergoing a digital transformation and being augmented with visual capabilities. This upgrade could allow clinicians to detect heart failure faster and more reliably than current diagnostic testing, potentially having a huge positive impact on prognostic outcomes.

What does the future hold?

A.I will likely progress to allow healthcare professionals (HCPs) ‘see the invisible’ – identifying ailments that are undetectable with current tools, as well as accurately predict an individual’s risk of disease. Facilitating early detection, prevention and encouraging self-care is an essential factor in helping to ease the pressures placed on healthcare systems.

A.I software could propel the drive towards self-care, which is an intrinsic element of primary care and in many cases prevents individuals from even having to see a HCP. An estimated 20 – 40% problems seen in general practice could be addressed through self-care. Wearable devices already feature applications that can assist with self-care activities and provide useful data to supplement standard care.

Ensuring HCPs are being utilised effectively is critical for the future of healthcare systems. HCPs are encumbered by repetitive administrative responsibilities, losing up to 33% of their time on such tasks. Employing A.I software can alleviate the burden of day-to-day admin and maximise efficiency to ensure their time is spent on crucial activities, such as seeing patients.

Building forecasting tools through A.I software could also assist healthcare services to predict potential spikes in A&E admissions, hospital length of stays, and optimize the allocation of hospital beds. In addition, potential treatment decisions and outcomes could be modelled to help select the most effective option for each patient.

Why are safety and ethics important?

A.I is developed on the data we have available, but what if that data is not representative of the care that should be provided? In emergency departments in the US, Black Americans are 40% less likely to receive any pain medication for acute pain compared to White patients. Although the reasons for the disparity are complex, bias is a contributing factor. And if this current standard of care is forming the data for A.I systems, what are the potential implications in a clinical setting whereby decisions are made on the recommendations of A.I? Developing A.I software is an iterative process, but it’s crucial that the data used to train it is scrutinised because widening the health inequalities seen today could continue jeopardizing patient care.

In addition, data protection is of paramount importance. In 2017 the Royal Free Hospital, UK, failed to adequately comply with the Data Protection Act when it supplied Google’s DeepMind with 1.6 million sets of patient data for A.I purposes, resulting in a massive breach of patient confidentially and trust.

Conclusion

The current and future applications of A.I herald a new era of healthcare and its delivery. However, as A.I depends on data, it can only successfully assist in situations that occur frequently, for which there exists a plethora of data. More complex areas and issues in healthcare lack the data needed to effectively employ A.I. The technology is also still in its infancy and further progress is required to ensure safe and ethical conduct in wider healthcare scenarios. So, while A.I may not be the sole solution to the healthcare crisis, in the wise words of cardiologist Eric Topol, perhaps it can make healthcare human again – which is a great place to start.

By Divya Shah, Page & Page and Partners