Surgical site infection
factors and new information – so, can artificial intelligence help to better predict those patients at risk of contracting an SSI? AI can be trained, using large amounts of healthcare data (or ‘big data’), to recognise patterns and can identify new trends that are emerging, rather than relying on what we already know. There are some potential benefits – including earlier identification of key factors of interest, the ability to focus down to the level of individual patients to develop personalised prevention approaches, and to support sustainable approaches by optimising our resources. AI is being used to improve the recognition and diagnosis of SSI. In terms of prediction, most of the studies use binary data (does the patient have diabetes: yes/no) or continuous data (age, systolic bp). “We are now seeing the ability to interrogate
free text in patients’ electronic patient records; to pull out signals and patterns to feed that narrative text into prediction models to increase our ability to identify those patients that are likely to develop a healthcare-associated infection…This really does work,” he commented. However, Dr. Price added that there is a
lack of application to clinical practice. Are we actually stopping infections in our patients, therefore?
He went on to provide an insight into the
development of an AI prediction tool, initiated, with colleagues Ash Myall and Sid Mookerjee, in 2020. With the arrival of the pandemic, the tool was re-directed to help predict the risk of COVID-19. (Myall et al, Lancet Digital Health 2022) He explained that a mathematical model
was built, and patient electronic data was accessed in order to understand key factors, such as comorbidities and the environment. It also included the movement of patients around hospitals, to understand their interaction with one another and staff, which improved accuracy significantly. “We were able to validate this work with
comparable hospitals and we collaborated with an institute in Geneva. We applied these models at different points during the pandemic and found that this AI infection prediction tool was able to accurately predict which patients were more likely to get hospital-onset COVID [to an accuracy of around 90%]. We were really pleased with its performance,” he commented.
“We were able to identify key variables that
were associated with hospital-onset COVID. Unsurprisingly, it was the amount of COVID in the hospital at the time, how close they were to someone with an infection, and if they had any direct contact.” Although this wasn’t an interventional study, the data was used to help target patients that were identified as being at risk of going on to develop COVID with bespoke care bundles. They found a reduction in the onset of infections and the study provided a signal of what might be possible, in the future, by using AI. “We have been working with a start-up
company [NEX.Q] about developing these tools to bring about a decision support to help us in our infection prevention and control,” he revealed. Dr. Price explained that the technology has
the potential to securely utilise data streams within healthcare, including microbiology and environmental data, to support real-time collection of data used for infection prevention, which is normally manual and time consuming. “This is overlaid with mathematical modelling
to recognise patterns of HCAIs – to identify them as early as possible, identify transmissions as
26
www.clinicalservicesjournal.com I May 2024
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68
