AI in A&E: high-tech care with respect for the patient

What exactly is the ESCORT project?
Ewout De Roock, project manager at the VUB: “The official name of the project is AI-enabled healthcare services during cross-border medical emergencies and regular patient services. That sounds complex, but it comes down to one goal: looking at how artificial intelligence and digital technologies can help healthcare professionals, especially in emergency situations. Think of disasters, pandemics or major incidents where several countries have to work together. The project is funded by the European Commission under the EU HORIZON framework and brings together different partners: universities, hospitals and technology companies. The technical partners develop new tools and systems, while the clinical partners, such as us, test whether they work in practice.”

Prof. dr. Ives Hubloue, head of the Emergency Department at UZ Brussel and professor of emergency and disaster medicine at the Faculty of Medicine and Pharmacy of the VUB: “The core question is actually very simple: how can technology help us make better decisions in favour of the patient? Especially in disaster situations, responders are confronted with an enormous amount of information in a short time. AI can help analyse that information more quickly and thus reduce the risk of human error.”

What are the main action areas of the project?
De Roock: “We are working on three major scenarios. The first is care for chronic patients. Here we try to develop technology that can reduce hospital visits by patients, for example through remote monitoring. A second area is pandemic prediction. We try to use algorithms to detect when a virus may start spreading rapidly. That is of course complex, because such models require a lot of data and those are not always available. The third scenario focuses on major incidents or disasters, such as attacks, natural disasters or industrial accidents. In such situations, several emergency services and hospitals often have to work together, sometimes even across national borders. We are trying to develop technology that makes that cooperation more efficient.”

Hubloue: “Europe is one large community, but in healthcare countries often still operate according to their own systems and procedures. If, for example, you have to deploy ambulances from different countries during a disaster, you notice that those systems do not always connect perfectly. That is what we are trying to align better with this project.”

Can you give some concrete applications?
De Roock: “One example is the so-called connected ambulance. This is an ambulance that can continuously transmit data to the hospital: the patient’s vital parameters, images of the situation and medical data. This allows the hospital to prepare better even before the patient arrives.”

Hubloue: “For that we use, among other things, augmented-reality glasses. The responder in the ambulance can share images with specialists in the hospital. In that way, a surgeon can already observe a patient who is trapped in a wreck after a traffic accident.”

De Roock: “In addition, we are also working on systems that reduce the administrative burden for healthcare professionals. Today, healthcare professionals have to manually register a great deal of data for every patient. With speech recognition and AI we can automate that process, allowing responders to maximise their time with the patient.”

Hubloue: “In disaster medicine we are also looking at wearables: small sensors that measure, for example, blood pressure, heart rate or oxygen levels. If you have many victims at the same time, AI can analyse that data and indicate which patients most urgently need help.”

The research runs until the end of this year. What are the main lessons so far?
De Roock: “One of the biggest challenges is dealing with data. AI systems need a lot of data to work well, but medical data are obviously very sensitive. That sometimes makes it difficult to collect sufficient datasets. In addition, we notice that every country has its own procedures and technologies, which makes cooperation within the EU more complex.”

Hubloue: “But at the same time we also see how valuable that cooperation is. In Europe we have an enormous amount of expertise. By working with partners from, for example, Sweden, Italy and Germany, we learn a lot from each other. What strikes me most is that technology only works if it fits the reality of healthcare professionals. That is why we test everything in simulations and exercises. It has to be practically usable; otherwise it makes no sense.”

What do you still dream of?
Hubloue: “My dream is that AI becomes a kind of extra assistant in the emergency department. Imagine a patient suddenly has a dangerous drop in blood pressure. A system could then immediately warn us: ‘Be careful, something is happening here.’ It does not replace the doctor, but it helps ensure that nothing is missed. This is already happening in medical imaging as well. AI can, for example, detect a small fracture on a scan that a doctor might overlook.”

De Roock: “I think we can also still make a lot of progress in simulation. Today we use computer simulations to recreate disasters. In the future we want to implement AI in those simulations, so that they become more realistic.”

If a major disaster were to happen in Brussels tomorrow, are we better prepared than a few years ago?
Hubloue: “Yes and no. We have learned a lot from recent crises, such as the coronavirus pandemic or the attacks in Europe. As a result, there are better plans and procedures. But disaster medicine remains complex. If there are suddenly many victims, you have to make difficult choices. You cannot always help everyone at the same time. That is why we also invest in training and cooperation, for example with defence. We train doctors and nurses for situations that fortunately do not occur every day, such as war injuries. Technology can help with that, but in the end it is still the human who makes the decisions.”