How Artificial Intelligence can help medical incident response
Artificial Intelligence (AI) is changing society and our lives. A lot of the “buzz” around AI concerns consumer devices such as Amazon Alexa, self-driving cars, and industrial devices such as robots which are changing the workplace. AI is also having a major impact on clinical medicine, especially in contexts such as medical imaging where complex data needs to be analysed.
AI can also help first responders, that is people on the scene who deal with medical emergencies until paramedics or other qualified medical personnel arrive (such as airline cabin crew when there is an in-flight medical emergency). Recent data from the International Air Transport Association (IATA) indicates 7.8 billion people are expected to take to the sky by 2036. The pressure is therefore on to find innovative ways to support cabin crew first response.
It can be extremely challenging to play a first responder role; you need to make decisions swiftly – with only basic medical training – in what is often a very stressful environment. An AI assistant can help people like cabin crew take care of their passengers, and in particular, make good decisions, and hand over key information to medical professionals. In doing this, AI assistants can help save lives, and can also make the first-responder experience less stressful. In more detail:
Reducing risk: The worst-case scenario for any first responder is that they forget to do something or are unsure what to do, and the patient deteriorates. AI systems can monitor what first responders do, and give reminders or alerts, prompting them to take action appropriately. Monitoring can be based on sensor data and information entered by the first responder; in the longer-term computer vision systems could be used to watch the first responder and the patient. It may sound like “Big Brother” to have an AI system literally watching over the responder’s shoulder, but most would welcome anything which observes their actions and warns them about potential changes in patient status.
Giving advice: Sometimes first responders need to make potentially life-critical decisions. These can be medical interventions, such as giving the patient oxygen. They can also be logistical, such as deciding whether a flight needs to make an emergency diversion (which of course is expensive for the airline and disruptive for all passengers). AI systems can help in such cases by making recommendations based on real-time available data, including contextual data (e.g. how far the plane is from its destination) as well as medical data. These recommendations can be based on explicit “best-practice” rules; they can also be based on analysis (perhaps using machine learning techniques) of previous incidents.
Handover: When the passenger is delivered to ambulance services or other medical personnel, the first responder needs to brief the paramedics on the patient’s status, what happened before the emergency (e.g. observations from people in neighbouring seats), and what happened while the first responder was looking after the patient (both medical problems and interventions carried out). This may sound trivial but it’s not; people often forget to mention important information during handover, especially in stressful situations. Again, AI systems can help in handover, by pulling together relevant data and using natural language generation (NLG) technology to produce an accurate summary of the data for the paramedics. It's like a translator, which converts data into language.
In 2018, AI is just starting to be used in first responder contexts. In five years’ time, I expect that AI will be an essential tool used by all first responders. I certainly think it can play a major role in helping cabin crew do a high-quality job of looking after their passengers.
Written by Professor Ehud Reiter, Department of Computing Science, University of Aberdeen.