Emergency physician, Charleston. Building Palmetto Medical Defense Consultants. Defense only. Also shipping RefAssess for referee coaches and mentors.
Charleston, SC, USA
Joined January 2009
- Tweets2.5K
- Following5.2K
- Followers1.1K
- Likes2.9K
Pinned Tweet
New episode: I sat down with Shane Peterson on @TheShiftTales to talk about life as an ER doc.
Listen here: [youtube.com/watch?v=Hazp1HKH…]
#MedTwitter
A board complaint never asks what kind of doctor you are.
It asks what you did, why, and whether the note shows it.
In the ER, I write every note assuming someone outside medicine will read it with no context.
Physician by night, founder by day.
The honest part: something always gives. Tonight it's sleep. It can never be the patient.
Doctors building on the side: what's the one thing you refuse to let slip?
RefAssess update: last week I asked for 5 referee assessors to test it. The X post got 0 replies.
It has 10 testers now. None came from X.
Lesson: the people I'm building for aren't scrolling here. The next ask goes where they already are.
When an ER chart gets reviewed in a lawsuit, what's hardest to reconstruct?
100%What we knew, when
0%Why we waited
0%What the patient said
2 votes • Final resultsThe ER Version of scope of practice: the debate is about who can see the patient.
The chart asks something else. A co-signature is one click at the end of a shift. In a deposition, it reads as supervision.
For the defense, who actually saw the patient is the first question.
Busy Monday yesterday. Morning: on set with @WCBD talking about a question I get every shift: ER or urgent care?
Quick version from the ER side: Chest pain, trouble breathing, stroke signs, a bad head injury, or "worst I've ever felt"? Come to the ER.
Sprain, cut, sore throat, UTI symptoms? Urgent care is usually faster and cheaper. When in doubt, come see us.
Evening: back to building…
From the ER side of building: I made RefAssess myself between shifts, no CS degree, no team. Big AI made the code cheap. It didn't make knowing the workflow cheap. A tool nobody opens mid-shift is worth zero. Is domain the last moat?
An $84M chest pain verdict, and most of the takes blame the PA.
From the ER side, I'd look at the schedule first. 20+ patients a day, two visits, and chest pain in a 53-year-old man. The system set that clinician up to miss it.
Defense lawyers, how often does the schedule get as much scrutiny as the clinician?
Ughhhh they had a clinic PA seeing over 20 patients a day and this led to provider missing ACS (acute coronary syndrome) in a 53 year old male patient (who happened to be an exec at Ernst & Young). Patient was evaluated twice for chest pain … and both times the PA appears to have misdiagnosed him with something benign.
In an ER malpractice defense, which record does the most work?
0%Triage Note
100%Physician MDM
0%Nursing Notes
0%EMR Audit Log
1 votes • Final resultsLooking for 5 soccer referee assessors or mentors to test RefAssess this fall.
You log events live, tag each one to the official who owned it, then score the criteria after.
Free. I want blunt feedback more than users. Reply or DM me.
The ER Version: the federation of state medical boards is drafting guidelines for AI that works with little or no physician supervision.
At 3 a.m., the tool suggests the order. I sign it.
When it's wrong, the board complaint has my license number on it. Not the tool's.
I can't enter this one, but I love that it exists.
I'm a full-time ER doc. I built RefAssess, a soccer referee assessment app, myself between shifts. No CS degree, no team, a lot of late nights.
A program like this would have saved me months. Physicians with an idea: submit it!
Calling all physicians attending #HLTH2026.
Have an idea that could improve healthcare? We want to help you build it.
We're teaming up with @OpenAI and Physician Side Gigs for the Physician Builder Challenge.
Submit your idea by October 20 for a chance to build a working prototype with us and present it live at Physicians at HLTH on November 16 at 5 PM.
→ abridge.com/hlth2026?utm_sou…
Short version: a coworker likes to say you don't rise to the occasion, you fall to your training. That's what keeps the ER calm when the world is in crisis.
Full episode: youtube.com/watch?v=Hazp1HKH…
Tomorrow I'm on Tales from the Shift, a show about what jobs actually do to the people who work them.
We talked about working in the ER.
Episode drops Sunday.
@TheShiftTales
RefAssess. Center and assistant referees don't get the same form.
Center: match control, game management, positioning, movement, teamwork, participants.
Assistant: offside, involvement, positioning, movement, teamwork.
Same match, different job. A generic report misses both.
The AMA's agentic AI framework asks the question these cases will turn on: who is responsible for the risks and outcomes?
ER version: an agent orders a head CT from triage before I've seen the patient.
For the defense, the chart has to show who held authority at each step.
Patients deserve the benefits of increasingly capable AI without compromising the safety, accountability, and clinical judgment at the heart of good medical care.
As agentic AI assumes increasingly consequential clinical functions, clinical authority must be earned through evidence and maintained through demonstrated performance. We need a clear framework for determining what these systems should be permitted to do, under what conditions, and who is accountable for their actions and outcomes.
Today, the AMA is setting out that framework and committing to advancing its broad adoption.
This fall, the AMA Center for Digital Health and AI will prioritize work with physicians, policymakers, regulators, and technology developers to translate these principles into practice, with a particular focus on liability, licensure, and regulation. spr.ly/6013BHESoc
LLMs talking patients through their wearable data is coming fast.
ER version: a patient arrives post-seizure and says "the app told me to wait."
Now the timeline includes a chatbot. Who answers for that advice is the question nobody's settled.
Imagine large language models (LLMs) serving as an interface between patients and all the wearable devices they use!
A study just proved that it works effectively.
"A new AI-driven platform could allow people to interact with their medical devices by sending messages using a smartphone application and receiving feedback in real time.
Participants in a proof-of-concept study undergoing monitoring for seizures could ask questions about their brain activity, receive personalized alerts, report symptoms and behaviors, and help train the system to better recognize seizure-related patterns."
You wear a health sensor and can ask questions about the data and the insights derived from that, like talking to ChatGPT.
This is going to boost the already huge population (around 1 billion people) using health wearables.
Source: penntoday.upenn.edu/news/pen…
