AI May Be Assessing Your Patient Before You Arrive
For decades, the first organized assessment of a 911 patient typically began when EMS arrived on scene. That is starting to change.
Artificial intelligence is now accessible from the same smartphones people use to call 911, and some bystanders are beginning to reach for it during emergencies while waiting for an ambulance.
That raises a question worth thinking through: What happens when your patient's assessment starts before you get there?
A New Kind of Bystander
Picture an elderly patient who falls in a public place and cannot get back up. Someone calls 911. The ambulance is still several minutes away. A bystander wants to help but has almost no medical training, so instead of simply standing there or searching online, they open an AI assistant and ask what they should do.
The AI may start asking questions:
- Is the patient awake?
- Are they breathing normally?
- Did they hit their head?
- Are they bleeding?
- Do they take blood thinners?
- Did they feel dizzy or weak before they fell?
- Can they move their arms and legs?
- Did they lose consciousness?
Those questions probably sound familiar because many of them are the same questions an EMT or paramedic might ask.
The difference is timing.
That information may be collected several minutes before your unit pulls up.
The Minutes Before Arrival Are Becoming Clinical Time
Response time has traditionally been viewed largely as a logistics measurement. Technology is beginning to turn those minutes into part of the patient's medical timeline.
A bystander working through an AI assistant's questions may notice changes in mental status, breathing, pain, weakness, speech, or responsiveness before anyone in uniform reaches the scene.
Some of those observations could be clinically useful.
Consider the patient who appears relatively normal when you arrive but reportedly had slurred speech five minutes earlier.
Or the fall patient who initially complained of dizziness but no longer remembers feeling dizzy.
Or the patient who briefly lost consciousness before responders walked through the door.
Details like those can influence your differential diagnosis and clinical decision-making.
The challenge is making sure that information actually reaches the responding crew.
This Does Not Replace 911 or Emergency Medical Dispatch
It is important to be clear about one thing: consumer AI is not emergency medical dispatch.
Emergency Medical Dispatch operates within structured protocols, medical oversight, quality assurance processes, and established standards. A general-purpose AI assistant does not.
AI can also get things wrong. It can misunderstand the information it receives, provide an incorrect answer with confidence, or miss context that would be obvious to someone physically standing next to the patient.
Anyone experiencing a medical emergency should activate 911 first rather than relying on an AI application for diagnosis or emergency medical guidance.
So the useful question is not whether AI should replace dispatchers, EMTs, or paramedics.
The real question is what EMS should do about the fact that patients and bystanders are already using it.
The Bystander May Be Holding Information You Want
EMS crews already ask family members and witnesses what happened before arrival. AI simply introduces another possible source of information, and sometimes a surprisingly structured one.
Someone who spent several minutes answering an AI assistant's questions may have created a more detailed timeline than they realize.
Questions like these may eventually become more common during an EMS assessment:
- Did you use an AI assistant before we arrived?
- What did you tell it about the patient?
- Did the patient's condition change while you were waiting?
- What did you notice earlier that you are not seeing now?
None of that information should automatically be considered accurate simply because software was involved.
But there is an important distinction between accepting an AI-generated diagnosis and documenting that a witness reports the patient was confused, dizzy, weak, or briefly unconscious before EMS arrival.
The first is a machine's interpretation.
The second is a witness observation, and EMS providers have always considered witness observations when building a patient history.
EMS Documentation Has Some Catching Up to Do
Today's electronic patient care reports are primarily designed around the EMS encounter.
Increasingly, meaningful patient information may exist before that encounter officially begins.
That leaves the profession with several open questions:
- How should crews document AI-assisted bystander observations?
- Should ePCR systems include a structured way to capture significant pre-arrival findings?
- Could dispatch systems eventually incorporate information collected through approved digital tools?
- How should providers distinguish between a witness observation and an AI-generated interpretation?
There are no clean answers yet.
EMS agencies, medical directors, software vendors, regulators, and clinicians will likely have to work through them as these tools become more common.
Rural EMS Could Feel the Impact Even More
A few minutes of pre-arrival information can matter in any EMS system.
In a rural county where the closest ambulance may be twenty minutes away, that window becomes much larger and potentially more consequential.
Family members and bystanders in those systems may already be using AI tools to determine what changes to watch for, what information to collect, or whether a patient's condition appears to be getting worse while EMS is responding.
Done safely, that could give the responding crew a clearer picture of what happened before arrival.
Done poorly, it could introduce inaccurate information or even delay appropriate emergency care.
Either way, the technology is becoming increasingly accessible, which is exactly why EMS cannot simply ignore it.
EMS Is Not Competing With AI
Artificial intelligence is not a replacement for the judgment of a trained EMT or paramedic standing next to a patient.
It cannot palpate a pulse, evaluate skin signs firsthand, physically examine a patient, interpret the entire scene, or weigh dozens of subtle findings against years of clinical experience.
But AI does not have to replace EMS to change EMS.
It can change what happens before you arrive. It can influence how patients describe their symptoms, what questions family members think to ask, what observations are made, and what information may already be waiting when you walk through the door.
That alone could have an impact on prehospital medicine.
The Takeaway
The first assessment of your patient may no longer begin when you arrive.
A family member, witness, smartwatch, smartphone, or AI assistant may have already collected something useful during the minutes before EMS reached the scene.
You still verify it. You still perform your own assessment. You still use clinical judgment.
But dismissing pre-arrival information purely because consumer technology played a role in collecting it could mean overlooking useful pieces of the patient's story.
AI is already entering healthcare.
EMS now has to decide where it fits.
This discussion was prompted in part by a recent JEMS article examining the use of consumer artificial intelligence by a bystander before EMS arrival. Read the original JEMS article here.
AI tools are not a substitute for calling 911, following emergency medical dispatch instructions, or receiving evaluation and treatment from qualified healthcare professionals.
