Artificial Intelligence (AI) is fast transforming many areas of society, including healthcare. According to the Department of Business and Trade, the UK is now home to over 1,300 AI companies – a sixfold increase in the past decade – with a combined turnover of nearly £1.47bn.
Whilst there are many benefits that come from this new technology, with AI being used in a variety of ways in healthcare, from administration and scheduling to diagnostics and treatment plans, there are also a number of challenges which need to be addressed to ensure that concerns are mitigated, and a high level of patient care is maintained and, in many cases, improved.
As healthcare providers come to grips with the implications of the AI Opportunities Action Plan, we outline the opportunities and obstacles for both healthcare buyers and suppliers in the implementation of AI.
Opportunities For Healthcare Buyers
Healthcare buyers are the organisations such as hospitals, NHS trusts and doctors’ surgeries that require the procurement of healthcare services. AI provides several opportunities for healthcare buyers.
Reducing Staff Workload and Burnout
One great advantage of AI is that it can be very effective at freeing up staff time for other tasks. It is well known that many NHS staff are overworked, and this can lead to human error due to tiredness and having too many things to do or remember at once.
With AI taking on some tasks that would usually be done by humans, this reduces time pressure on staff, freeing them to make more considered decisions. Additionally, the use of AI can help with staff burnout, with a 2023 NHS England staff survey indicating that approximately 30% of NHS staff experience it.
Burnout often occurs when staff take on too many responsibilities and feel overwhelmed with the level of pressure, a significant problem for many hospitals as staff absence can have a significant impact on the level of care that can be provided. AI enables hospital managers to decrease the workload of staff members to more manageable levels.
Diagnostics
A key area that AI can be very helpful for is in diagnostics as it can be used for analysing scans and spotting risk factors. The BMA reports that emerging evidence from Somerset NHS Foundation Trust demonstrates how AI software can accelerate the diagnostics pathway, cutting the wait for a CT scan after a chest x-ray from seven days to under three.
AI can often work quicker than humans, therefore completing a greater workload. This can lead to quicker identification of medical concerns and more efficient diagnosis. This leads to better patient outcomes and ultimately saves lives, which is particularly the case when it comes to cancer where time is of the essence to identify and diagnose life-threatening conditions.
Telemedicine
Another area that AI is very useful for is telemedicine. This allows people who are unable to access healthcare, such as those in remote areas or areas with few medical facilities, to receive medical attention including remote consultations, monitoring and diagnostics. Without telemedicine, some demographics would not have access to efficient healthcare and thus would suffer worse health outcomes.
The rise of healthcare wearables can also enable members of the public to gain greater insight into their own health metrics and enable them to monitor their own health. The BMA states that research by Healthwatch, National Voices, and the Good Things Foundation highlights how digital services can improve accessibility for certain groups, such as carers, individuals with reduced mobility, and those who are immunosuppressed or shielding. Additionally, some evidence suggests that digital mental health services help overcome access barriers to traditional care, including stigma.
Challenges For Healthcare Buyers
Job Loss
There are also several challenges that healthcare buyers face when it comes to AI implementation and use. Firstly, the use of AI, whilst freeing up staff time, can lead to job losses as AI systems take over tasks previously done by humans. If managed correctly, staff can be redeployed into other tasks or areas, however, concerns remain about humans being replaced. Job loss affects not only the individual but also their family and wider community.
AI systems can often complete tasks more efficiently than humans and sometimes with greater accuracy due to lack of human error such as a loss of concentration; however, the need for humans to check the accuracy of AI outputs remains. This can be seen in the use of medical professionals to check AI outputs regarding breast cancer scans. We should look to AI to transform healthcare jobs by streamlining routine tasks and boosting efficiency, rather than replacing roles.
Manmade Biases
Another issue is manmade biases. After all, AI is created by humans, so it is vulnerable to being taught the same biases as humans are. Biases can be human made or reflect historical or social inequalities and trends. This can include minority, racial or gender biases.
A key example is how many centuries of excluding women from healthcare research have led to diseases that primarily affect women being overlooked, misdiagnosed, or poorly understood. Additionally, AI-powered diagnostic tools are less effective at detecting skin cancers and other lesions in patients from ethnic minorities, as they are primarily trained on data from white patients.
Algorithms are trained on thousands of existing scans, so it is crucial that the training data includes scans from a diverse range of ethnic groups to ensure the technology is effective for all patients.
AI Training
Additionally, a challenge for healthcare buyers is the fact that training in AI takes time, money and resources. It is well known that the NHS has a lot of demand placed on its finances and every financial decision has to be carefully considered and prioritised. This can lead to healthcare buyers being very cautious about where money is spent and wanting certainty about the usefulness and effectiveness of new AI systems.
Healthcare buyers must assess the impact of spending money on AI rather than in other areas. There can also be a reluctance to change from existing systems which staff understand how to use and which seem to be working well. As AI is constantly evolving, training will need to be regularly updated, and this is another problem for healthcare buyers that are trying to keep costs down.
The time needed for training is another issue as many members of staff in healthcare settings do not have much, if any, time to spare. When AI conducts opportunistic testing, inaccuracies or excessive sensitivity could lead to a significant amount of unnecessary testing. However, despite these challenges, AI can be so transformative that it can hopefully save time, money and resources in the long-term.
Opportunities For Healthcare Suppliers
Greater Insights
Healthcare suppliers consist of organisations that provide services to healthcare settings such as hospitals and doctors’ surgeries. There are opportunities for these suppliers when it comes to the deployment of AI. First, these organisations can gain greater insights into various processes by using AI, such as demand forecasting, product development and more efficient decision making.
AI can process and analyse data much more quickly than humans can, which is particularly helpful when there is a large amount of data to analyse. An example of an organisation using AI to deliver healthcare services is Komodo Health, which provides software that enhances patient care and alleviates the burden of disease through data-driven insights.
Cost Processing
Another opportunity for healthcare suppliers is the automation of cost processing. This includes order processing, customer support and inventory management. This automation helps save the company time, resources and money to complete these essential but time-consuming tasks.
Challenges For Healthcare Suppliers
Lack of AI Expertise
Despite the opportunities, there are also some challenges for healthcare suppliers when it comes to AI. Firstly, smaller organisations will likely struggle with a lack of AI expertise, more so than larger organisations will. The knowledge regarding which AI systems to use and how to use them will have more of an impact on the company’s time and resources than it would for a larger company. This is because if one employee must take time out of their usual tasks to learn about AI then this will have a greater impact on a company with fewer staff.
Financial Issues
Additionally, the cost of implementation of AI systems is more challenging for smaller companies than it is for larger ones. Larger organisations may not feel the financial impact of purchasing AI systems as much as smaller organisations will. The cost of implementation may make some smaller organisations more reluctant to implement AI as they may feel they can’t afford it or can’t take the time to increase their confidence in its benefits.
AI Integration
Another issue is integrating AI systems with existing ones. Different organisations have different systems and processes in place, presenting a challenge for people within those organisations to understand how they can transfer data to new AI systems or to at least integrate it successfully. Bringing in new solutions can be very daunting and may make some organisations reluctant, especially if they feel that their current systems are working well.
AI in healthcare should be seen as a tool to help save time and resources to improve the health outcomes for patients. Human input will always be needed. As with the rapid growth of any technology, there will be unforeseen issues that need consideration and need to be overcome.
Bill Gates stated that while AI could be helpful in many areas, it “won’t be perfect and will make mistakes“. The important thing is to adapt and mitigate these mistakes. We must adapt to a changing technological world, which includes changes to the healthcare landscape.
