Machine learning predicts negative anaesthesia outcomes

Researchers from the University of Adelaide and the North Adelaide Local Health Network (NALHN) are conducting a pilot study to determine if machine learning can predict when patients will have adverse reactions to anaesthesia.

The researchers will draw on an extensive database of more than 100,000 patients collected over the past 25 years at the Lyell McEwin Hospital, in Adelaide’s northern suburbs.

Multiple sets of observational data will be analysed along with anaesthetic pharmacology, laboratory data, biographical and comorbidity data.

“We aim to predict the likelihood of adverse outcomes after patients are discharged from the operating theatre, when there is an unplanned admission to the intensive care unit (ICU), in medical emergency response calls and in the first 48 hours after their surgery, in order to allow early intervention,” said Professor Matthew Roughan, Interim Director of the Teletraffic Research Centre at the University of Adelaide.

“Advanced mathematical and statistical tools have a long history of application to health and medical applications.

“However, not much work has considered the application of machine learning to digitised health records, in particular for information after a patient’s operation.

“We are very excited at the possibilities of this research and the ability to use mathematics to make a real difference in people’s lives.”

Dr Tim Beckingham, Consultant Intensivist at the Lyell McEwen Hospital, said while having anaesthesia was relatively safe in Australia, certain people are at a higher risk of having complications which may be identified early.

“The risk of death from anaesthesia is low in Australia, with a mortality rate of one death for every 57,023 patients,” Dr Beckingham said.

“Complications from surgery that require an unplanned admission to the ICU, or a return to the operating room, are more common.

“We are very excited at the possibilities of this research and the ability to use mathematics to make a real difference in people’s lives.”Professor Matthew Roughan

“We are hoping to develop early warning systems that clinicians can use to predict when patients will deteriorate, reducing the risk of serious illness or death due to surgery.”

The project is underway with results expected in the first half of 2023.

The researchers will use information stored in the NALHN data warehouse, which is maintained by SA Health.

Along with a separate anaesthetic database, the data warehouse contains data of more than 118,000 patients who had anaesthesia at the Lyell McEwin Hospital, in addition to a wide group of demographic, diagnostic, observational and medication data.

/University Release. This material from the originating organization/author(s) may be of a point-in-time nature, edited for clarity, style and length. The views and opinions expressed are those of the author(s).View in full here.