With a growing need to improve processes, patient and clinician satisfaction, and patient outcomes, Jon Payne, Technology Strategist and Innovator at InterSystems discusses the opportunity – and increasing requirement – to use AI to reduce the post-covid treatment backlog and innovate across healthcare.
What challenges is the healthcare industry currently facing?
The COVID-19 pandemic has already created a substantial backlog of patients needing to see a doctor or nurse, with 175,000 diagnoses of key conditions estimated to have been missed in 2020, according to an official assessment by the Department of Health and Social Care (DHSC). With UK GP surgeries now being urged to resume face-to-face appointments at pre-Covid levels, having a really efficient booking system has never been more critical to ensuring wait times don’t skyrocket – and GPs don’t burn out.
How can AI help GPs to see more patients without added strain?
Compared to other industries, healthcare has been relatively slow to invest in and adopt new technology and Artificial Intelligence (AI) and Machine Learning (ML) have been no exception. However, the opportunity – and increasing requirement – to use AI to innovate, be that to improve processes, patient and clinician satisfaction, or patient outcomes, has never been greater.
More than 15 million appointments in general practice have been found to go to waste each year in England because patients don’t show up or dial in, according to NHS figures. This puts the cost of “did not attends” (DNAs) at £216 million, which would pay the annual salary for more than 2,300 new, full-time family doctors. To combat this, some healthcare practices are now using ML to intelligently overbook medical appointments, in the same way that airlines overbook flights. In a nutshell, these practices are using patient data to determine who is likely to miss their appointment, and enabling doctors to see someone else, if a patient is a no show.
As we head into Autumn, COVID-19 infections are set to creep up and epidemiologists are warning of a potential winter crisis. AI-enabled overbooking models could prove a crucial tool in optimising appointment booking for patients in the run up to Christmas, while at the same time reducing the strain and backlog on already stretched healthcare personnel and services.
How exactly does the technology work?
Say, for example, that a GP practice offers 30 appointments every day. If the no show rate is calculated at 14%, four patients a day are going to skip their appointments. This leaves GPs waiting around at considerable, cumulative cost to the NHS, while unwell people trying to book in to see them are told there are no free slots.
By overbooking four slots randomly throughout the day or according to their own anecdotal knowledge of patients’ track records, practices might be able to claw back some ad hoc appointment time. More effective, however, would be to use machine learning to predict the probability that a person is not going to show up and overbook the slots with a probability score of over 85%, for example. This probability can be calculated by taking all the surgery’s historical data and training the ML algorithm on that. It could then intelligently figure out which features or variables in that data set impact the outcome, create a rule set and then apply it to current appointment data.
Are there other healthcare processes that can be optimised in the same way?
Patient readmissions are another costly challenge for healthcare providers around the globe. This is because it can be difficult for organisations to know where they should focus their discharge planning efforts, to ensure they minimise the number of people who end up back in hospital.
If you examine the data, looking at four different factors – length of stay, the acuity of the admissions, comorbidities, and number of emergency visits – organisations can make a best guess prediction around who is most likely to come back and then take action.
However, to get an even better result, trials have shown that if you implement ML, healthcare organisations could save even more money because the technology finds more patients who are likely to return. While week on week, the saving might seem modest, in a hospital with 785 discharges every day, and with a readmission rate of 17.5%, the saving could be as much as £4 million a year.
If the results are good, why isn’t everyone doing it?
What is holding healthcare organisations back from AI adoption if the efficiencies it brings are both substantial and much needed? Not enough quality training data, concern over trustworthiness and data bias, as well as limited access to computing resources, were all referenced as key barriers to success in AI by hospitals surveyed as part of IDC’s AI in Healthcare study.
To combat bias, we must find ways to present models visually for those using them to easily understand and interpret, as well as use algorithms that include explainability. In this way, those working in healthcare organisations know exactly how decisions are being made and can track, assess and modify these as necessary. Good, clean well-organised data is also vital to AI and ML’s success.
However, despite these uncertainties, the ways in which AI and ML can be used to make hospital processes more efficient are proving numerous. What’s more, as organisations continue on their path to AI maturity, we will likely see more benefits emerge for patients and clinicians alike.
