Gene Expression in Neurons Solves a Brain Evo Puzzle (wired): or Brains are Amazing according to new tools

Single cell RNA sequencing is da Bomb

With the rise of Chatbots and AI, we are left to wonder: what remains for paltry humans to do in this ever-evolving world?

A partial answer is here. Allison Whitten writes an evocative piece about the mammalian neocortex and the researchers using Single Cell RNA sequencing to tease apart the difference between ancestral salamander brains, modern reptilian brains and mammalian brains that include a neocortex.

I love this stuff. And it seems that the neocortex (for now, based on best evidence) is purely a mammalian innovation, the ability for higher level reasoning and thinking (and ? the seat of consciousness) different from ancestral or reptilian brains.

It makes me wonder about Wonderful Life, the Burgess Shale, where Gould talks about evolution and the dice-rolling that has occurred and how random it is that WE are here instead of completely different beings.

CMIO’s take: Learn about how we think, in the service of making tools that align and support these thoughts. AI does not supplant human. A human augmented with AI will supplant the human alone.

“I don’t think it should take you 3 days to tell me that my baby is dead” (Information Blocking anecdote, and JAMIA journal article)

Congratulations to Dr. Stephen Rotholz, outstanding colleague and informaticist, who authored this paper describing our organization’s journey to deliver test results immediately, supporting information transparency.

Trouble is, sometimes the results can be life-altering, causing patient distress and anxiety. And of course, this happened on a Friday afternoon.

Our open-access article details what exactly happened, and the request from our OB/Gyn service line to with-hold test results for 3 days, and how we resolved the issue.

It was a big lesson in humility for me, and a major learning about managing change in complex environments.

I’m also hopeful that our experience informs the decisions being made by organizations around the U.S. in the age of Information Blocking / Sharing and the requirement of immediate access to results by patients.

Chatbot perspective from an insider (Rodney Brooks and

What Will Transformers Transform?

Thanks Rodney for a thoughtful discussion of

  • The Hype Cycle (peak of overinflated expectations)
  • The caution needed as our tools grow in skill exponentially
  • The ongoing risk of hallucination and unexpected errors in chatbots
  • The “grounding” problem with AI and robots

I particularly love the following quote:

Roy Amara, who died on the last day of 2007, was the president of a Palo Alto based think tank, the Institute for the future, and is credited with saying what is now known as Amara’s Law:

We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.

I’m feeling an upward acceleration of AI skill with GPT applications, which could open up substantial risks as well as benefits.

For example, in the EHR space, there is discussion that GPT could take a patient and physician’s recorded conversation and automatically write a progress note, qualitatively more accurate and much faster than current commercial tools. Further, it could potentially summarize weeks of progress notes on hospitalized patients and write the discharge summary, a document that, when well-written, can take a human many hours of work. Or even, receive patients’ incoming MyChart messages and clinical questions and “reply to the patient in the voice of their clinician” based on the decade of writing by that clinician in the EHR.

Sure, these seem great. How about the potential deluge of GPT agents writing notes, requests ON BEHALF of patients, or other authors, that could junk up our systems? If 183,000 incoming patient messages is a lot (current monthly patient message volume at UCHealth), what if GPT somehow enabled 10x that number?

How much discussion will be GPT talking to GPT on behalf of employers/ patients/ discussants? I understand science fiction editors now have a 10x increase in sci-fi story submissions, a SUBSTANTIAL FRACTION now being written by GPT based on prior stories?

Too, has written about this:

“I worry that we are very much in a ‘move fast and break things’ phase,” says Holstein, adding that the pace might be too quick for regulators to meaningfully keep up. “I like to think that we, in 2023, collectively, know better than this.”

CMIO’s take? Take a breath, everyone. It’s going to be a bumpy ride.

EHR inbox @ucsf (a talk by me, CT Lin!)

Here are my slides for EHR inbox innovation at UChealth. Including deleting 12 million messages from the inbox. And setting new teamwork standards.

Here are my slides for my talk.

Here is my intro and summary slide from the talk and you can download all the slides from this 10 minute talk. I spoke about Message Appointments, DC the CC to reduce automated message delivery, improve rx renewal standard process, Desktop Medicine to block 1 session per half day for clinicians to do inbox work.

Sorry but you missed the song too: Epic man, if you were not in person or on the hybrid conference call.

The author, trying to sing on-key.
Apologies to the live audience.

Ok, for those of you with FOMO. Here is the original performance.

EHR Inbox @ucsf Hoifing Poon, PhD from Microsoft research

What is Microsoft doing in the research and innovation space with LLM (large language models) matching complex cancer patients to clinical trials.

In research and discovery, what can we look forward to beyond the immediate recent announcements? Beyond DAX express. Beyond the Epic partnership with Microsoft.

Taking masses of unstructured data and building intelligence step by step.

Digesting a complex patient data from a cancer journey. Now able to match a cancer patient to the right trial with lots of criteria almost twice as effectively as prior predictive models. Now with GOT monitoring, real time clinical trial matching. Went from finding 2 patients by manual chart review over weeks. to finding 100+ with GPT assist. Near instant. And drawing from unstructured data. There is a LOT more unstructured than structured, coded data.

Can’t wait to see what Microsoft cooks up next in the AI space.

EHR Inbox @ucsf by Maria Byron MD at UCSF

The brave souls at UCSF. Billing for online patient messages = eVisits.

Dr. Byron discussing UCSF data on billed messages.

Live since summer of 2020. version 1.0. eVisit Vs patient advice request. Patient has to find the options. Really hard to teach them how to do this. ‘I don’t know, CAN you do this online?’ Lot of uncertainty by all. Low use.

11/2021 new experiment. Single way to send. Consent with each message, it may be billed.

Make it super simple for clinician work.

Now experimenting with a protocol for MA to have criteria to sort message to eVisit without bothering clinician. PFAC v supportive of this change. Out of pocket costs comparable to in person visit.

13,000 eVisits. 900,000 threads encompassing about 3M messages. 1.4% conversion rate. (Is that the right number we need to shoot for?) Decline of 3% reduction decrease of messages overall. $470,000 revenue. No increase in phone. No complaints to patient relations. No reports of patients not seeking care. Patient advisory councils report favorable change. Watching for disparities. Not seeing any. Based on race, ethnicity, language, payer, age. still this is a big psychological win for our clinicians.

Thank you to Dr. Byron for such a thoughtful approach and leading the way on this path.

EHR inbox Kris Lee MD Permanente medical group

Dr. Kris Lee showing us development work at KP.

Desktop medicine team is completely separate team with own resources. With Data sciences. Has their own AI. TOP priority for the organization. NLP to group COVID requesting Paxlovid and send to right person. Also sort urgent to the top. Or other sort method with AI bot. 

Task switching is a huge problem. Challenges and balances. Our doctors get 44 patient messages a day.

Don’t be afraid of asking hard questions. Or big decisions. Do it in a transparent way. Try it. It may fail but then learn something to/from/with an advisory board.

Love this framework. Tools are NOT the thing that tell you what to do. Instead define the problem first.

Thanks. Dr. Lee for your inspiring direction and words.

Ha! This is why GPT and AI sort is nowhere near what our docs need to know.

This presenter is hilarious and spot on.
Thanks to Dr. Kris Lee

EHR inbox @ucsf Jane Fogg MD

Use data to redesign inbox work at Atrius. What did they figure out?! Out of box ideas as well as executing on standard work.

Directing messages to the right person right away.

Eliminate automated messages. 98% reduction in media manager by directing them to the right person (specialist who ordered) and not PCP.

Care everywhere notifications. 100% reduction in unnecessary notifications.

Rx renewals. 50% reduction. With embedded automated protocols. Run by non licensed staff. Centralized. Eliminated physician sign off.

Automation lab results. 19% of volume. Removed ‘normal normal’ (normal in all clinical situations)

and send ONLY to patient and not to PCP.

It works. Removes 30% of lab volume today.

Another possibility: nursing directed messages for vitamin D protocol management.

PMAR: allow team members to touch messag first and reconcile a significant volume before it ever gets to physician/APP.

Another idea! Clinical coverage department. Retired clinician who can cover some visits and inbasket to help physicians in distress or extended leave. Wow. Also PCP and APP (advanced practice provider) partnerships.

EHR Inbox @UCSF Julia Adler-Milstein PhD

Transformation of work from paper to electronic inbox. Historical perspective and also ideas about where we might go next.

Concept of 1-touch resolution of an inbox message ?!

A lot of improving this is based on excellent availability and use of EHR audit log data.

Can we now phenotype providers and teams on how they handle messages?!

What happens when we change user interface design? When it is easier to see outside data, do behavior of providers change (turns out, YES).

What happens when we allow for billing for e-Visits?

Both clinicians and patients are affected by this. Small increase in physician billing of messages. Patient message threads dropped once they saw the disclaimer about possible billing.

These are excellent questions we can start to ask and decide on new directions for our organizations.

Hmm pause for creativity. There are more opportunities for health systems to make things better.

I’m going to have to study her words and think what we do next in this space for our organization.

%d bloggers like this: