Letting AI hold the public purse?!

Each year, national and local governments determine the relative priorities of services to allocate funding. How would AI spend the cash? What makes more sense for a vibrant society — spending on economic development or growth, spending on education, international development, social care, libraries? What would ideal balance look like?

If this question sounds familiar, it’s not the first time we’ve tried to apply artificial intelligence (AI) to making this decision. Last year, we asked AI which company should be the UK’s investment banker and what jobs young people should do.

We invited a real-life artificial intelligence to take part in the national debate about the future of the NHS. What should be prioritised? Where should funding be focused? Where should we make cuts? Where can we afford to spend money? In this case, our AI was taught using thousands of programmes from the House of Commons library and oral evidence from experts, patients and experts from several official think tanks.

(What does AI think? Read on…)

Our AI did the same exercise for UK police spending, and did the same for education and school spending.

Feeding AI the data needed to get its ideas was painstaking. But it also highlighted some of the barriers to giving AI access to the data its creators would like.

Our AI often found it impossible to understand how it was being asked to prioritise these areas, and frequently came up with nonsensical answers, for example suggesting that police should spend less on crime prevention.

So how would AI decide what to prioritise for the NHS?

It would take into account government policy, taking into account what our experts say, and the likelihood of political consensus. It would make recommendations based on things like health outcomes — it wouldn’t gloss over preventable causes of disease. It might suggest keeping hospital beds open because patients are happier there. It wouldn’t assume that any one kind of hospital is better at tackling certain problems than another; it would suggest piloting different approaches to see which works best.

Our AI also used social data, including the amount of anxiety a patient had about going to hospital. It would take into account what our experts say. This could be on how doctors handle emergencies, or the quality of care before a condition is diagnosed. It would also weigh up things like the likelihood of patients being treated more fairly by an AI that made decisions based on gender or race, for example.

Our AI fed all of this data into its thinking process. It would look at the policies to see if they were being properly funded, and take into account historical trends.

As our AI would determine all spending categories in government policy — for example, funding for schools or social care — it would take into account when there are to be changes in the law that change the way governments decide on priorities. Our AI might suggest that social care can no longer be overlooked when health is prioritised, for instance.

AI decides on government spending

In the UK, there’s a National Health Service (NHS) budget every year. The budget is shared between all health services, including primary care and social care. The NHS is the largest area of government spending after debt interest payments and it provides a uniform set of services across England, Scotland, Wales and Northern Ireland.

The differences in regions are particularly notable when comparing health data between north and south England. Northern services have been shown to be more cost-effective than those in the south, for example. Northern patients have taken longer to die from some conditions, such as cancer and heart attacks, and others, such as common cancers. Northern patients also have a shorter life expectancy.

In addition to money, the government funds the NHS through a mix of central grants and local government taxation. This includes large amounts of money raised through council tax support for local health services.

The government has the ability to change funding streams and allocations to areas of spending. For example, it can decide that certain areas of services should receive more money than others.

We ran the algorithm on what experts say is important for patients, alongside government policy, not just on social data. It would factor in historical trends — the time it’s taken for certain diseases to be addressed or new treatments to be developed, for example — then make decisions with an eye towards what experts say are the most cost-effective treatments today.

But the NHS is far more than just funding. The government can also shape how it’s used. For example, the Department of Health recently announced plans to place patients at the heart of healthcare in England by putting an emphasis on patient experience with patients’ voices heard in local commissioning. By involving patients in decisions about their own healthcare, researchers say this will help them feel more engaged in their treatment and increase adherence to treatments. This would be part of a wider effort to deliver services at the “right” time, in the “right” place.

A patient’s treatment can also be shaped by local decisions.

For example, following the publication of the Multi-Region Strategic Health Care Review (MRSHC), which aims to improve health care for people of all ages across England, NHS England has announced that it will review how new A&E units are funded and expand access to GP services.

A&E departments are often overstretched, and the MRSHC recognises the need for more A&E closures. However, people can still access GP services via other routes. This could mean that GP visits could be prioritised over urgent A&E care, for example.

The government also has the ability to invest directly in specific areas of health care to improve its effectiveness.

For example, the Department of Health and Social Care has announced a £100m Cancer Drugs Fund to be allocated over the next five years. The government has said it will invest more than £280m in this fund so people can access life-extending drugs across England. This is on top of the £75m a year already being invested through the hospital drugs fund.

In another example, the government has this year launched a new technology accelerator aimed at boosting research and development in areas such as healthcare and manufacturing. The £50 million Innovate UK HealthTech Catalyst scheme will invest in digital innovations that could improve patient care and medical research, including one of the first digital health accelerators in Europe.