Essay, Thinking, Hard and Soft
Leave a Comment

Triage 2.0

Image: Wounded arriving at triage station, Suippes, France

Excerpted from Thinking, Hard and Soft

In healthcare, a triage – from the French verb “trier”, to sift or to sort – is a system to evaluate and categorize patients by the severity of their medical condition to prioritize care. The future of healthcare hinges on new triage technology.

Triage was first developed for battlefield medicine by Dominique Jean Larrey, an army surgeon during the Napoleonic wars. In 1792 France was at war with most of Europe and Larrey was a newly minted twenty-six-year-old regimental surgeon-major frustrated by the state of combat casualty care. Surgeons like him were confined to field hospitals located no less than five kilometres from the battlefield.

If he was “lucky”, a wounded soldier was able to make his own way to the field hospital or comrades might carry him. Otherwise he’d have to wait for a slow moving cart from the field hospital to haul him in. These carts, called Fourgons, were piled high with bodies and provided the most basic transport. If he did make it to the hospital alive, the common soldier would have to wait until the high-ranking wounded had been cared for.  An enemy prisoner would be treated last.

Larrey wanted a more direct route. He had noticed the efficiency and speed with which horse-drawn, “flying” artillery navigated the battlefield to punch holes into the opposition with grape-shot. He adapted the idea and designed a two-wheeled carriage with suspension that could hold an injured man all stretched out. His plans were approved and flying ambulances began to fight their way into battles to reach the wounded.

Bringing patients in was only the start. Under Larrey’s watch, the order of wound management and limb amputations was prioritized by severity. Lighter cases were set aside, as were those soldiers deemed mortally wounded. Soldiers of any rank and nationality whose survival depended on a surgeon were treated first. Faster time to treatment and effective prioritization cut the mortality rate from amputations: from one in two to one in ten, according to some estimates. Morale was boosted as soldiers felt safer. Napoleon was delighted and bestowed honour upon honour on Larrey.

Developed further during the First World War, emergency prioritization such as the Manchester Triage System is now standard in hospitals world-wide. Resources are limited and prioritization must be systematic if it is to be fair. But with all our modern technology and data, the descendants of Napoleonic triage seem primitive. 802.15.6 is the latest IEEE-SA (Institute of Electrical and Electronics Engineers Standards Association) standard for Wireless Body Area Networks (WBAN). In the future, smart watches and powerful mobile devices connected to personalized medical monitors via WBAN will transform triage.

In 2013, the Michael J. Fox foundation sponsored a data analytics competition on Kaggle, inviting participants to predict the incidence and progression of Parkinson’s disease from smartphone data. Parkinson’s disease gradually kills dopamine-generating cells in mid-brain, leading to involuntary movement disorders and dementia. Treatment depends on the stage of the disease. During the first stage, dopamine agonists are prescribed to activate dopamine receptors. Patients are given a diary recording whether the symptoms are under control “on” or not under control, “off”.

Diaries aren’t ideal, they are impractical and somewhat unreliable. Participants in the Kaggle competition showed it was possible to use accelerometer data from smartphones to classify Parkinson patients effectively. Smartphone data is crude and noisy. If good results are possible with smartphones, imagine the improvement possible with carefully calibrated accelerometers and sensors that measure the electrical activity of muscles.

The central challenge in triage lies in assigning patients into categories correctly. Mistakes will always be made: a well-known doctors’ quip is that in healthcare, the only unambiguous diagnosis is death. But the aim is to reach diagnostic certainty while patients are alive. Classifying patients correctly makes it possible to administer the right care at the right time.

The future of medical care may involve self-triage, or self-sorting. Technology makes it possible for patients to monitor themselves. Far from triage for limb amputation, consider the problem of sorting out a voice disorder. Approximately 7% of the U.S population suffers from vocal cord disorders caused by voice misuse. Recent studies show how to classify vocal patterns using data from accelerometers attached to the large dip at the base of the neck above the collar bone (the suprasternal notch). In one particular study, the Java derived Android platform on a Nexus-S smartphone was customized to read data from an accelerometer through the microphone channel. The 9.3 Gb of data generated weekly fit comfortably in the smartphone’s memory. One of the possible applications is to generate biofeedback that a patient can use to modify the use of his voice.

Triage in emergency rooms will always be Napoleonic by necessity: there is no time for vagaries; either a patient has to be resuscitated, or he seems alright for now. But outside the emergency room, technology will increasingly sort patients minute by minute, second by second on a continuous scale.

This will not happen overnight. The technology will radiate outwards from smartphones and smartwatches to disease specific monitors. But sooner or later, the components will come together, linked through a WBAN. The challenges are immense. The 802.15.6 standard is already known to be vulnerable to security breaches. If you are worried about privacy in the era of the Internet of Things, how about privacy in the era of the Internet(s) of Bodies?

Perhaps that’s the least of our worries. How willing are we to trust algorithms that provide us with bio-feedback? Up to what point do you want to be triaged remotely before insisting on a specialist? And when you do enter a hospital, sensor readings in hand and demanding priority status based on solid evidence, how will you be sorted and assigned?

The challenge for triage 2.0, and for the future of healthcare, is to balance the demands of informed patients with expert assessment of their real needs.


Learning to detect vocal hyperfunction from ambulatory neck-surface acceleration features: Initial results for vocal fold nodules, IEEE Trans Biomed Eng. 2014 Jun; 61(6): 1668–1675,

Predicting Parkinson’s Disease Progression with Smartphone Data, Kaggle Competition sponsored by the Michael J. Fox Foundation,

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s