Artificial Intelligence in Trauma: The Future of Trauma Systems

The use of artificial intelligence (AI) in maximizing the potential of trauma systems is still in its infancy. Tools that utilise artificial intelligence have yet to move from their pilot phases into real clinical practice. But several machine learning models have been shown in simulation studies to be as effective as existing triage tools in determining trauma center suitability and future iterations may have the ability to learn in real-time from incorrect triage decisions. An inherent flaw in the current triage methodology is that decisions are static and do not incorporate a systematic feedback system enabling triage rules to evolve based on previous wrong decisions. This is the strength of advanced AI which could lead to it revolutionising how we triage and transport the patients of the future.

 Predicting resource utilisation

Imagine an AI system that could use historical data to predict trauma volumes depending on the time of day, day of the week, time of year, major events and even the prevailing weather conditions. It may be able to predict not only trauma volumes but the likely blunt or penetrating mechanisms, anatomical patterns and likely injury severity.

 One could imagine a morning team briefing at the local trauma center going something like “okay team, today’s the day of the state college football finals, based on data from previous years with the weather we have today we can expect anywhere between 4 to 7 alcohol-related traumatic head injuries in males aged 19 to 27 so let’s check our stock of ICP bolts and hypertonic saline. I think it’s worth us running a quick head injury sim to make sure we’re slick. Dr Guerrero, could you liaise early with neuro-ICU…..”

 As the British Army adage of the 7P’s goes: Prior Planning and Preparation Prevents P*ss Poor Performance

 Triage decision-making

 Technology may also have a role to play through the integration of real-time clinical data collected via smart watches and other internet-connected wearable devices.

 In 2018 a Korean team successfully produced a data-driven AI model for prehospital and mass casualty triage which incorporated vital signs from wearable devices. The devices could also detect verbal and motor responses and use that data to generate a modified GCS.

 At some point in the not-too-distant future, ambulance crews may well be able to tap into real-time patient data and accurately determine the need for Level 1 or 2 care.  Based on location and traffic patterns, the optimal means and route of transport could be automatically decided upon prior to dispatch

 The Dutch TESLA trial is currently underway to assess the use of a machine-learning prediction model (the Trauma App) and see how it impacts on triage accuracy, hospital resource use, and a cost-utility. This is something that could soon be incorporated onto the smartphones or devices of ambulance crews.

 Geospatial planning of trauma centre location

There have been several studies looking at the use of geospatial mapping technologies to optimise designation of new trauma centres.

 One such example utilised the American College of Surgeons Needs-Based Assessment of Trauma System Tool (NBATS) which was created by College to help determine the need for trauma centers within a region. In the region surrounding the Elvis Presley Trauma Center (EPTC) in Memphis, the addition of a suburban trauma center resulted in only a 1% increase in population coverage but a steep reduction in patient volumes to EPTC, whereas the addition of 2 rural trauma centers resulted in a big increase in the number of injured people within 45 mins of a trauma center and caused a less significant reduction in patient volumes at EPTC.

 But the capabilities of AI are an enticing addition to tools like NBATS, bringing the ability to account for dynamic population growth, changing traffic patterns, and regional infrastructure. This could dramatically improve trauma system planning and optimize placement of future trauma centers to areas of highest impact.

 The opportunities afforded by increasingly sophisticated AI are limited only by our human imaginations. Time and money invested now could be dividends in future not only in trauma system efficiency but in lives saved.

Obi Nnajiuba is a British surgical resident with a specialist interest in trauma, acute care, prehospital care, triage, mass casualty events and trauma systems. His postgraduate qualifications include an MSc in Trauma Sciences and membership of the Royal College of Surgeons of England. He was recently awarded his PhD for his work on optimising the London Trauma System Triage Tool. He is also a registered Motorsport UK physician, providing trackside advanced trauma care to competitors at world famous motor-racing circuits such as Brands Hatch, Goodwood and Silverstone.