Assessing how serious a person's injury is requires a quick trade-off of many different parameters. More lives could be saved if medical professionals were able to get fast, life-critical decision support from AI tools. This was presented by research institutes at Chalmers University of Technology, the University of Gothenburg and the University of Bolles in Sweden.
If a seriously injured person is taken directly to a general university hospital, the chances of survival are increased because there are resources there to deal with various types of injuries. Therefore, we need to be able to better judge who is seriously injured and who is not, so that everyone receives appropriate care and resources can be used in the best possible way. Anna Bakidu, a PhD student in the Remote and Pre-hospital Digital Health Research Group in the Department of Electrical Engineering at Chalmers University of Technology care@distance said.
In a recently published study, Anna Bakidou and her co-authors developed five different mathematical models based on data from adults exposed to ambulance care from 2013 to 2020.
The data comes from more than 47,000 real-life events in the Swedish Trauma Registry, and it also shows where these people were sent.
By weighing a complex range of variables, such as respiratory rate, type of injury, blood pressure, age, and gender, the results showed that all AI models performed better on clinical outcomes than the transport decisions made by ambulance crews at the time.
Many seriously injured patients were taken to general hospitals
It was found that 40% of seriously injured patients were not taken directly to the polyuniversity hospital.
At the same time, 45% of patients with non-serious injuries were unnecessarily sent to university hospitals, as their injuries could be treated in general hospitals.
Ambulance crews are often faced with difficult and urgent decisions. We hope that a more objective decision support system will act as an 'extra colleague', allowing staff to see more complex connections and think twice when injuries are difficult to perceive or assess." Anna Bakidu said.
As an example, she said, young people (who are often involved in traffic accidents) are often considered more seriously injured than they actually are.
On the other hand, when older people are involved in events such as fall accidents, they are often assessed as minor injuries, although their condition may suddenly become life-threatening due to consequences such as internal bleeding.
There are a few more steps before the technology can be put to use
Although mathematical models suggest that many lives could potentially be saved, there is still a long way to go before ambulance crews can use this technology.
The key step is to find a way to quickly and easily get all the information into the AI tool and that enables the service to interact well with the user.
For example, can we communicate with tools so that both hands are free? How can AI be collaborated within existing workflows and protocols, and how can recommendations for staff be updated when new data is added? We need to test and take these things into account when doing more research and prototype work. Anna Bakidu said.
Large-scale clinical trials are needed before AI services can become part of the day-to-day work of ambulance workers.
The regulations mean that this will take time, and there are some concerns about AI. If something goes wrong, the consequences can be severe. Everything that is introduced into the medical field must be verified. At the same time, we also know that some of the methods used today are not always the best. There isn't much research on AI when it comes to ambulance care, and we hope that our mathematical models can provide adaptable support for the work environment and provide more equitable care in the long term. Co-author Stefan Candeford, associate professor in the Department of Electrical Engineering at Chalmers University of Technology, said.