such interesting responses here
personalized EMRs are going to pop up soon not just from tech teams but maybe even clinical teams
medplum don’t mess this up lol
personalized EMRs are going to pop up soon not just from tech teams but maybe even clinical teams
medplum don’t mess this up lol
My EHR costs $550 a month. Cloud based. No server, no IT person.
I dictate with Apple's built-in dictation and saved phrases. No expensive dictation system. It keeps overhead low.
Solo docs: what does your EHR cost you a month?
3
0
14
3,127
had an attending in residency tell me that flonase is usually always the right answer
I wrote a guide to clearing your nose.
I generated my advice by having subscribers use different methods so I could figure out what works and what doesn't work.
Give it a read: cremieux.xyz/p/quitafrin
4
2
107
27,944
lol it’s such a medicine thing to comment on this and say 6 am is too late for pre rounds cause ofc you gotta say how your own program is so much worse
tbh i don’t even remember what time i used to pre round i just thought this was funny lol
tbh i don’t even remember what time i used to pre round i just thought this was funny lol
you’ve never met at PGY-1 pre-rounding at 6 am
1
1
41
6,812
has been wonderful to work with Ortho RI team and see how @PatientsApp can help patients get to surgery without all the delays
i’ve done a lot of pre-op visit and it’s the small details that hold up surgeries so catching them early is a great outcome for everyone, especially patients
i’ve done a lot of pre-op visit and it’s the small details that hold up surgeries so catching them early is a great outcome for everyone, especially patients
A cancelled case doesn’t have to mean an empty OR.
At Ortho Rhode Island, earlier visibility into patient records gave the care team more time to act—and more time to refill the schedule.
Here’s what changed. 🧵
0
0
12
750
sooo that nyc geography trivia game i’ve been sharing recently has been a little side project i’ve been working on - it’s called tap NYC (nod to maptap!)
excited to soft release it today with a special theme for @NYCMayor’s birthday haha
if you like nyc trivia give it a try, join the leaderboard
and if you’d like to host a day and make questions for others to play shoot me a message!
Tap NYC #201 🌃
Zohran's Birthday! 🎂
818 / 1,000
🟢🟡🟢🟢🟡
tapnyc.app
excited to soft release it today with a special theme for @NYCMayor’s birthday haha
if you like nyc trivia give it a try, join the leaderboard
and if you’d like to host a day and make questions for others to play shoot me a message!
Tap NYC #201 🌃
Zohran's Birthday! 🎂
818 / 1,000
🟢🟡🟢🟢🟡
tapnyc.app
2
2
5
403
i really like this take
I’m always struck by how much distress “upward feedback” creates for people in organizations.
Managers should be seeking to receive several times more feedback than they give (ratio should be at least 3 to 1 for each person they manage). This is so rarely the case.
Often the people who care deeply and who are willing to speak truth to power are in turn given feedback that they need to massage their feedback to make it more palatable to managers. It is totally fair to ask for feedback to be conveyed more thoughtfully but that should not stop a manager from wanting to listen closely to the actual concerns.
Good managers can look past the imperfections in _how_ the feedback is conveyed in order to hold curiosity toward _what_ is being conveyed such that the key observations and insights are not lost. And after acting on the feedback, they will separately offer guidance on how to improve communication in the future.
The above requires the capacity to repeatedly sit with things that are difficult to hear. Doing that well creates safety & trust. Not doing that well creates resentment & disengagement.
1
0
5
2,772
who are the best people you know working in cardiometabolic health / prevention?
i’m not as interested in longevity medicine but thinking about care delivery for behavior change and access to preventative tools and tests
i feel like the longevity / concierge people get a lot of the marketing here but there are likely people or companies doing interesting work in the space i’d like to learn from
i’m not as interested in longevity medicine but thinking about care delivery for behavior change and access to preventative tools and tests
i feel like the longevity / concierge people get a lot of the marketing here but there are likely people or companies doing interesting work in the space i’d like to learn from
4
1
8
1,198
more evidence that AI demos showing clinical diagnosis are not really use case we see in practice
The users of @OpenEvidence skew heavy toward IM and FM. Radiology is substantially underrepresented. The most common queries are about treatment/management, not diagnosis.
jamanetwork.com/journals/jama/…
1
3
13
1,418
is the use case here that i may need some kind of cell therapy in the future and can use my own younger cells that haven’t mutated yet?
i get the longevity bro take but what else is there here? genuinely curious
i get the longevity bro take but what else is there here? genuinely curious
In 2012 I said machines would do 80% of what doctors do. Many were furious. Now AI out-diagnoses doctors in study after study. Is printing organs on demand next? Having your own stem cell repository is next generation "planning".
@CellinoBio is building autonomous foundries that make personalized stem cells with AI and lasers. Your cells age every day. Preserve your immune cells with @withreservoir. Excited to back @nabihasaklayen, @madridmarinna & team.
1
0
3
434
somehow my knowledge of nyc is not that elite it seems lol
Tap NYC · Oct 15
634 / 1,000
🟢🟢🔴🟡🔴
tapnyc.app
Tap NYC · Oct 15
634 / 1,000
🟢🟢🔴🟡🔴
tapnyc.app
1
0
2
173
oof third party models no longer included - was wondering what the catch would be
Cursor is now part of The X Subscription.
Current Cursor subscribers can keep their plan or switch to The X Subscription.
Paid plans share one usage pool across Cursor, Grok Build, and other Grok products, with twice the coding usage on Grok models as the equivalent Cursor plan. Premium, Super, and Ultra also include Grok Bot.
To make room for more Grok usage, third-party models in Cursor move to API pricing on the new plans. We'll keep working to make your usage go further, with more compute, better inference, and more token-efficient models.
The X Subscription is available at cursor.com/pricing.
0
0
2
316
general vs vertical agents
there has been a lot of debate on whether specialized models are better than general models. in healthcare specifically, companies like @OpenEvidence have claimed that their specialized models can outperform general models from major labs
this debate is now moving to agents and whether general agents that can do everything are going to be better than “vertical agents” - ie agents specific for a given domain
for most day to day things the general agent will be enough. we’ve already seen this through agents like instinct or @Muse that can do almost anything you throw at them
more importantly, they have large social networks to reach consumers and can build integrations where they need in order to fill in the gaps (eg 1PW, Shopify, etc) and in the case of muse access to proprietary data like IG
there is an argument to be made however that having vertical agents for specific domains may still a good idea.
a couple areas that come to mind (in healthcare):
- specific clinical workflows - i’m thinking EMR automation, clinic tasking workflow triage, referral management, even prior auth. this is our thesis on the ASC space at @PatientsApp (the idea here is that a vertical agent that is built to understand the end to end workflow will be much better than a general agent alone. this doesn’t mean building your own model, rather building your own agent harness)
- care coordination for a patient and their family members - this is the @MeetKinHealth playbook. most of what Kin does can be done in chatgpt or claude but having a dedicated space for it (in the places where your family interacts) is very helpful. not to mention a UI that is built around this specific agentic use case
- clinical decision support - we’ve seen this with @OpenEvidence and @doximity but i think the real reason for this isn’t that the models are better. it’s more that docs want a dedicated space that feels separate from a consumer app. i think this is why chatgpt for clinicians is not gaining widespread use among my colleagues
on the consumer side there is also a different argument for vertical agents and that is that most people just need separation of different parts of their life. sure having one agent that can do it all is nice but having a long running thread with one agent about multiple things can sometimes become a state of information overload
agents like @bot solve this by letting the user create multiple agents/bots, a trick out of the openclaw book. my general test for this kind of interface is “the mom test” and “the sister test” - my mom isn’t a big tech user at all and my sister is the average tech user (email, netflix, etc). if a product wouldn’t land for them then i don’t see the average person using it
i don’t think the average user is going to build an army of bots for different domains of their life. maybe they will be open to download bots from a marketplace similar to apps but remembering to go to a thread about your work vs family stuff vs personal tasks doesn’t feel as clean and distinct as different apps and spaces for these. having vertical agents that specialize in a specific domain starts to make more sense when you think about this way. its not a limitation of the technology but more how people tend to think and work.
maybe this is a UI/UX issue and there is an underlying agent connecting all of these things and it’s just a matter of separating the UI for users or even building better ways to split domains inside one app like muse
i still think we have a ways to go with this but as much as the frontier labs want to pitch the general agent, a well scoped and defined vertical agent can carry a lot of weight and value
there has been a lot of debate on whether specialized models are better than general models. in healthcare specifically, companies like @OpenEvidence have claimed that their specialized models can outperform general models from major labs
this debate is now moving to agents and whether general agents that can do everything are going to be better than “vertical agents” - ie agents specific for a given domain
for most day to day things the general agent will be enough. we’ve already seen this through agents like instinct or @Muse that can do almost anything you throw at them
more importantly, they have large social networks to reach consumers and can build integrations where they need in order to fill in the gaps (eg 1PW, Shopify, etc) and in the case of muse access to proprietary data like IG
there is an argument to be made however that having vertical agents for specific domains may still a good idea.
a couple areas that come to mind (in healthcare):
- specific clinical workflows - i’m thinking EMR automation, clinic tasking workflow triage, referral management, even prior auth. this is our thesis on the ASC space at @PatientsApp (the idea here is that a vertical agent that is built to understand the end to end workflow will be much better than a general agent alone. this doesn’t mean building your own model, rather building your own agent harness)
- care coordination for a patient and their family members - this is the @MeetKinHealth playbook. most of what Kin does can be done in chatgpt or claude but having a dedicated space for it (in the places where your family interacts) is very helpful. not to mention a UI that is built around this specific agentic use case
- clinical decision support - we’ve seen this with @OpenEvidence and @doximity but i think the real reason for this isn’t that the models are better. it’s more that docs want a dedicated space that feels separate from a consumer app. i think this is why chatgpt for clinicians is not gaining widespread use among my colleagues
on the consumer side there is also a different argument for vertical agents and that is that most people just need separation of different parts of their life. sure having one agent that can do it all is nice but having a long running thread with one agent about multiple things can sometimes become a state of information overload
agents like @bot solve this by letting the user create multiple agents/bots, a trick out of the openclaw book. my general test for this kind of interface is “the mom test” and “the sister test” - my mom isn’t a big tech user at all and my sister is the average tech user (email, netflix, etc). if a product wouldn’t land for them then i don’t see the average person using it
i don’t think the average user is going to build an army of bots for different domains of their life. maybe they will be open to download bots from a marketplace similar to apps but remembering to go to a thread about your work vs family stuff vs personal tasks doesn’t feel as clean and distinct as different apps and spaces for these. having vertical agents that specialize in a specific domain starts to make more sense when you think about this way. its not a limitation of the technology but more how people tend to think and work.
maybe this is a UI/UX issue and there is an underlying agent connecting all of these things and it’s just a matter of separating the UI for users or even building better ways to split domains inside one app like muse
i still think we have a ways to go with this but as much as the frontier labs want to pitch the general agent, a well scoped and defined vertical agent can carry a lot of weight and value
2
0
11
1,262