Getting the dose right

Each year, more than four million Australians are administered anaesthetic.¹  

And while anaesthesia-related mortality is very low (3.29 per million or roughly 13 patients per year), up to 25 per cent of patients will experience some side-effects. Most are mild, generally nausea or vomiting, however, some patients can experience more significant issues, such as post-operative delirium or cognitive dysfunction. For clinical providers, even simple complications such as slow recovery from unconsciousness can impose a cumulative strain on resources, impacting on costs and the availability of care.

Research coming out of the ANU School of Engineering hopes to give doctors greater accuracy in predicting how a patient is likely to respond to anaesthetic, and therefore reduce the potential for complications during and post–surgery.

The research being led by Biomedical Engineer Dr Nicolo Malagutti draws on biological information taken from patients in real-time and maps it to a set of known pharmacological models to identify where the patient sits on a spectrum of possible responses. From this, it is possible to more accurately predict how a specific patient will respond following a particular choice of anaesthetic administration by clinicians.

“Doctors use this generic model which gives them a ballpark for all patients, but sometimes an individual sits a little bit too far on one side or the other of that ballpark,” Dr Malagutti said.

“Therefore, doctors must continually monitor evolving conditions and reactively make adjustments to manage the situation.

“If they administer too little, obviously the patient could gain awareness of a procedure which is really undesirable,” Dr Malagutti said.

“But on the other hand, a too-high dose could increase the risk of complications, from minor illness and extended recovery times, all the way to nervous system damage and other nasties. So, for doctors to strike that balance without actually having precise individualised knowledge of the patient is really tricky.”

“So the idea was, can we process some patient response information in real time, while the anaesthetic response is unfolding, and actually help the doctor pinpoint what the trajectory of that patient is going to be in a way that’s more accurate than using that ballpark estimate that is currently used?”

Dr Malagutti’s research uses brain activity data that is measured via an electrode strip that is placed on the forehead of the patient while they are under anaesthetic. The level of brain activity indicates whether the patient is sufficiently asleep, or heading towards neurological harm.

Another metric that Dr Malagutti is exploring is a direct measure of concentration of the drug within the body over time to more accurately predict how a patient will respond.

The behaviour of the drug, and its effect on the brain, depends on a number of factors, including things like an individual’s kidney and liver function. Currently, it’s difficult to monitor exactly how the drug is behaving as it circulates through the body.  

“You can’t give people a continuous blood test to know what’s happening in the bloodstream,” said Dr Malagutti. “But what is being explored internationally, and what we’ve started to look at, is finding a proxy measure of drug concentration based on breath analysis.”

The idea is that by measuring how much of the drug is exhaled, you can then determine the concentration of the drug in the body. The challenge, however, is that blood-carried anaesthetic diffuses into the air in very small quantities.

“You need extremely sensitive sensors in order to pick them up in a meaningful way,” said Dr Malagutti. “Last year I had a student of mine who was working on testing some novel materials as candidates to be highly sensitive receptors of the anaesthetic drug propofol. We’ve got a couple of promising leads, but we need to experiment a bit more.”

While Dr Malagutti’s theoretical models show that using a combination of brain activity and blood concentration can substantially increase the accuracy of the prediction, the question remains as to whether it provides enough of an advantage over current models to be adopted by clinicians.

“We’ve got some demonstrated accuracy in predictions from the brain index work, but at this point we don’t know if that’s enough to make a difference to clinical decision-making,” said Dr Malagutti. “We will be studying that with clinicians now going forward.”

Malagutti’s presentation of the initial phase of his research recently won the Radiation Oncology Private Practice Trust Fund Award for the Best Pre-Clinical Poster at the recent Canberra Health Annual Research Meeting. Malagutti sees this recognition as a good opportunity to now explore the next step of clinical use.

Future predictions

What sets Dr Malagutti’s research apart from other modern computer-based research is that it draws on existing pharmacological modelling. He believes that by using the same baseline models familiar to practitioners, it is more likely that computer-based technologies will be adopted into decision making.

“The reason why a lot of computer-based applications are failing to break through in the clinical space is a lack of interpretability, a lack of a common understanding between the doctor’s mental model and the machine’s model,” said Malagutti.

“If you go to a clinical application and you say, ‘my model tells me that the patient is about to fall off the cliff’. The doctor will ask, ‘Why do you say that? If you were to respond with, ‘the computer told me so’, it is really not something that’s going to fly.”

Dr Malagutti sees the future of computer-based technologies in medical settings as one in which “people and technology work together, feeding off each other, which means that they need to rely on the same baseline model of the problem.”

And while he believes that we will get to a point where some phenomena are just too complex for doctors to manage using their own mental models, and we will need to turn to intricate computer modelling to make decisions on our behalf, “we first need to take the time—and this is where we are now as a multidisciplinary community of doctors and engineers—to build up the field experience, the cross-disciplinary training and the ethical decision-making frameworks.

“Only then I think we will be ready to take that additional step of entrusting autonomous machines with human lives.”