Turning Wearable Data Into Personal Care; Medical Language AI at Scale – Oren Nissim & Tim O'Connell
November 27, 2025 • 24 mins
Wearables track thousands of data points daily, but most becomes noise instead of signal. Clinical notes document critical patient information, yet we cannot extract meaning at scale. Two founders solving how we turn data into trusted care.
Oren Nissim is the co-founder and CEO of Brook Health. He has type two diabetes himself, which drove him to build remote care for people with chronic conditions like diabetes, hypertension, CHF, and COPD.
The company works as part of the health system, extending primary care into the home. His mission is simple: people living with multiple chronic conditions at home need agency. The tools are cheap and covered by insurance. Brook collects thousands of data points daily from every patient. AI compares against baselines and identifies anomalies.
But here is what matters: a care team analyzes AI-flagged anomalies first, then brings medical decision recommendations to providers instead of raw data summaries.
Tim O'Connell, MD is a practicing radiologist and CEO of emtelligent, a nine-year-old medical language AI company. The company does large-scale data extraction from clinical notes and AI-assisted chart review.
He started the company in 2016 during the deep learning boom, years before the 2022 LLM explosion. His differentiator is that emtelligent does not use large language models as its core. The company builds custom language models optimized for cost, speed, and accuracy at massive scale.
His vision for healthcare is better data extraction from unstructured notes so we can use the critical information clinicians spend so much time documenting.
Highlights from Oren Nissim at Brook Health:
- His glow up is about use cases, not widgets. The industry is being forced to prove ROI rather than just adding more time and cost.
- The company uses AI to flag anomalies, then care teams validate and present medical decisions to providers. This creates guardrails so providers can trust what they see.
- His spicy take: watch Medicare Advantage closely over the next few months as some players walk away and others walk in.
Highlights from Tim O'Connell at emtelligent:
- His six-month glow up is moving pilots to implementations. After years of experimentation, 2025 is the year of execution.
- When extracting data, the software shows exactly where terms came from in source documents. This builds trust and allows human reviewers to verify accuracy.
- His industry glow up is better healthcare analytics. We need to extract meaning from the documentation clinicians spend so much time creating.
Healthcare gets better when we turn overwhelming data into trusted insights that providers can act on.
A "glow up" signifies a positive transformation, reflecting the journey of becoming a better, more successful version of oneself.
At The Tech Glow Up, we humanize the startup and innovation landscape by focusing on the essential aspects of the entrepreneurial journey. Groundbreaking ideas are often ahead of their time, making resilience and perseverance vital for founders and product leaders.
In our podcast, we engage with innovators to discuss their transformative ideas, the challenges they face, and how they create value for future success.
If you're a founder or product leader seeking your own glow up, or a seasoned entrepreneur with stories to share, we invite you to join our guest list via this link.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Nathan C (00:00):
Hey, it's Nathan.
Welcome to another specialdouble episode from the HLTH
Conference.
Each one of these episodes islaunching Monday and Thursday,
and we'll feature conversationswith two different CEOs so
welcome, listen in
Oren Nissim (00:21):
this is ultimately about something that is very
fundamentally available toeverybody.
It's cheap, it's covered byinsurance.
We can all do it these days.
There's no reason why anybodyliving with a chronic condition
home has to feel that lonelyanymore.
That's Glow Up.
Nathan C (00:39):
There's no reason that a patient at home should be
isolated and lonely.
Like we have the tools toprovide better care.
Fantastic.
Hello and welcome to the HLTHTech Glow Up.
I'm Nathan C, and today I amtalking with Oren Nissim of
Brooke Health.
Oren Thank you so much forjoining me today.
Oren Nissim (00:59):
Thanks so much for having me.
Nathan C (01:00):
Oh my goodness.
So Oren could you, introduceyourself and a little bit of
what you do at Brook Health?
Sure.
We're Brook Health and, we do,remote care for people with
chronic conditions.
We work as part of the, healthsystem, so we primarily work
with providers.
Mostly primary care, taking careof people with, diabetes,
(01:22):
hypertension, CHF, COPD, multichronics, at home, and
extending, primary care to thehome, essentially.
Amazing.
tell me a little bit about yourrole at, Brook Health as well.
Oren Nissim (01:35):
Sure.
So I'm the co-founder and CEO.
Grew a life as a patient.
so I have a complicated versionof, type two diabetes myself,
which I guess got me to be moreintertwined in it, and at some
point decided to, stopcomplaining and do something
about it instead.
Nathan C (01:50):
Oh my gosh.
That is like the classic founderstory, right?
Noticed a problem and just hadto go fix it for yourself.
could you actually dive in, tothat origin story?
I'm curious, how did you makethe decision to take this
entrepreneurial step?
Oren Nissim (02:10):
I was very involved in the worlds of, wearable and,
the ideas behind the fact thatwe today are wearing and
carrying with us a lot ofdevices at all.
Translate a lot of information,very ambiently.
And the idea is that we don'tneed to actually ask a lot of
questions in order to justunderstand what's going on.
So we could very easily comeinto a, if this, then that.
(02:33):
and so that has been thebeginning of the idea and from
that on it evolved.
Nathan C (02:38):
a amazing, I find, a lot of times, especially as the
general public is coming upagainst ideas about artificial
intelligence, right?
What do I even do with it,right?
You see the videos of Sora AImaking quirky, I'm riding a
rhinoceros video", and it's ifthat's what AI does, like how
does it affect me?
I'm a nurse, right?
(02:59):
Like I don't make rhino videos.
can you talk a little bit, a lotof tools are, using ai.
talking about predictive,talking about ways that we can,
get ahead of health outcomesbefore there's a problem.
How do you, approach this andthe use of AI in your work at
Brook?
Oren Nissim (03:19):
Sure.
So we track a lot of informationand we analyze a lot of data in
the background that allows us tounderstand anomaly.
Understand if a person has theseparticular conditions, what are
we actually looking for?
What is the behavior that wewould like to see?
And then what are those gaps?
I think the biggest problem thatcomes up, and I think you see it
in the show a lot.
There's lots of technologyavailable, lots of technology,
(03:41):
and there is lots of people thatwill sell all sorts of things.
And the key question is reallywhat I like to describe is the
accountability gap.
At the end of the day, you are aconsumer who bought X, Y, Z, and
you have it, and now you'refeeling empowered and all of a
sudden you got these signals,but you don't wanna be your own
doctor and you don't necessarilywanna be Google your doctor
(04:02):
either or check GPT.
Now the the new thing.
How do you actually close thatgap, which is where a lot of
this now needs to be broughtback in somehow into the care
system, but not as noise becauseit's just a huge amount of
noise, but actually a signal.
And what is the signal that wecan actually work on?
So I think, AI first use case isin order to decipher noise for
(04:26):
signal.
And then be able to bring itinto, as long as we're not
willing to give up the job oftrusting a provider, which I
don't think we're that close todoing, be able to do it in such
a way that is meaningful to aprovider and actually cut away
on their time as opposed toadding one more thing on a time
We as a company needs a lot ofwork into how to do that better,
(04:49):
because I think that's the mainfailure point with all this
technology available.
Nathan C (04:54):
And I'm super curious about that.
What is your primary way thatyou like to give time back to
doctors?
How do you achieve that?
Oren Nissim (05:02):
You know you're gonna be, we collect probably
thousands of data points a dayfrom every single one of our
patients, and the reality isthat just sifting through that
is a huge amount of pain.
AI is then able to ultimatelycompare this with what is the
baseline, what are we actuallylooking for, and where are those
anomalies?
On those anomalies, we haveessentially a care team that
will do the first level ofanalysis to ultimately bring it
(05:24):
to the provider on the, Hey,here's the next.
Medical decision for you toactually make, as opposed to,
here's just a huge amount ofdata, go read it and here's a
few summary notes, we want toget very quickly into the, given
this, here is the recommendationon that, what would you like to
decide on as the medicalprofessional?
Nathan C (05:44):
And is that, it sounds like there's the AI layer and
then there's also that humanlayer that's doing that
additional validation andresearch, presenting those
opportunities.
So that's happening in thebackground as the provider might
be doing other tasks, reviewingthe records and so can, isn't
just taking things off of theirplate, but potentially working
(06:05):
in the background working whenthey're not necessarily so that
the care is always on.
Yeah.
Regardless of the practitioner.
Oren Nissim (06:12):
Yeah, and I think that ultimately it's about how
do you actually createguardrails?
For this approach.
Because the reality is that weall feel like we're not willing
to hand the keys just yet, maybeover a period of time, and then
who can be in the middle and thesame time, how do you make that
efficient?
You can train health teams to domore, but that means you're
essentially, again, adding upmore time.
(06:33):
As opposed to taking time away.
So how to create it in asupervised fashion a very safe
environment so that people canreally trust what they're
seeing.
And that I think is a lot ofthat inner work that we have to
do these days.
Nathan C (06:45):
Amazing.
Awesome.
so moving on to some of the morebranded questions.
The show is the HTLH Tech GlowUp, and let's start at the
industry level.
I'm curious, what is the healthglow up that you wanna see for
this industry?
Oren Nissim (07:00):
what I wanna see is people not buying the tech for
the tech.
And I'm saying it as a techentrepreneur.
this is not about the widget,it's about the what are you
actually gonna do with it.
So it's about the use case.
And I think glow up here isabout the fact that I see a lot
(07:20):
of people coming in and saying Iget what you can do.
How does that actually make itmore useful to me?
And essentially forcing theindustry to come back and say,
we worked out the use case tothe point that we're not just
adding more time and adding morecost.
We're actually trying to deliveran ROI.
But assuming these new tools arein place, one of the cool
examples that you could seeright now is in the scribe world
(07:42):
whereby, we used to use zoomin.
Now you can use machines.
Okay, but how does that flowactually integrates well such
that we don't need to thenre-review this, we can actually
trust this and this is whereit's the next level.
It's not exactly you can't justuse AI to summarize.
You gotta make sure that it doesit well and to the fact that
it's actually useful.
Nathan C (08:03):
Thank you for that.
so the next question takes theGlow Up a little bit more
personal, a little bit closer tohome.
What's the glow up Oren, thatyou're looking to make and that
you're looking to make for BrookHealth?
Oren Nissim (08:16):
Thanks.
We've been on a mission for along period of time.
I started this with a mission,which is that people living life
with multiple chronic conditionsat home, we need some level of
agency.
ultimately this is about how dowe use these tools that are at
our.
Disposal to basically, getbetter agency in what we do and
how can we bridge that gap witha provider that we trust, but
(08:39):
have something in the middlethat can help us.
for me, this is ultimately aboutsomething that is very
fundamentally available toeverybody.
It's cheap, it's covered byinsurance.
We can all do it these days.
There's no reason why anybodyliving with a chronic condition
home has to feel that lonelyanymore.
That's Glow Up.
Nathan C (08:59):
There's no reason that a patient at home should be
isolated and lonely.
Like we have the tools toprovide better care.
Fantastic.
no pressure on this one, butsocial media loves a good hot
take.
Do you have a healthcare hottake, a spicy opinion, that you
(09:19):
wanna share about the industry?
Oren Nissim (09:24):
I will pay close attention to what's happening
with Medicare Advantage andpeople walking, some people
walking away from it, somepeople walking into it, and I
think that's gonna be aninteresting next few months.
Nathan C (09:36):
Interesting.
I love it.
Oren it's been such a pleasureto chat with you today.
Thank you for sharing, aboutyour AI powered Insights and a
tool that is clearing all of theinformation away so you don't
just have signal, but youactually have data.
thank you so much for visitingwith us on the HLTH Tech Glow
(09:58):
Up.
Yeah, thanks so much.
Appreciate it.
Amazing.
1, 2, 3.
See I'm the worst.
Hello and welcome to the HLTHTech Glow Up.
I am Nathan C and today I'mtalking with Tim O'Connell of
emtelligent Tim, thank you somuch for joining me today on the
(10:18):
Health Tech Glow Up.
Thanks,
Tim O'Connol (10:20):
Nathan.
Nathan C (10:20):
You have a pretty fantastic smile for this late in
the HLTH event.
Thanks for joining me early onday three.
how's your HLTH been?
Tim O'Connol (10:30):
HLTh has been great.
I love HLTh It's one of myfavorite healthcare conferences.
I think it's just got a greatvibe and.
I spend a lot of time in darkrooms and hospitals and, and I
appreciate, you know, we needmore unicorns in hospitals.
Yes.
I love it.
Nathan C (10:44):
Sure.
And the work that you do atemtelligent?.
Tim O'Connol (10:47):
I'm a practicing radiologist.
I work clinically one to twodays a week.
I'm also a CEO of a 9-year-oldstartup called emtelligent we're
a medical language AI company,so we do a couple things,
primarily large scale dataextraction from clinical notes,
and we also do AI assisted chartreview.
Nathan C (11:05):
I stopped by your booth a little bit earlier
today, and I was looking at someof the materials, and it seems
like your approach to AI isn'tmaybe the same approach that you
might see, with, a lot of otherfolks who are talking about AI
and health records.
Can you kind of explain, yourdifferentiator there?
Tim O'Connol (11:25):
Sure.
So as a company that does largescale data extraction, we work
with a lot of people across thehealthcare spectrum.
So it can be producers of noteslike health systems or it can be
consumers and users of thatdata, like payers and life
sciences companies.
And what they often have aproblem is they'll often be
like, we have several billionclinical documents that we need
processed.
(11:46):
And there's obviously a lot ofdata there.
So we've been spending yearsmaking our software so we can
really extract the meaning andcoded data from very large
volumes of data and do it.
Very quickly and efficiently andat a cost that works for our
customers.
So, yeah, I think a lot ofpeople are now using the
ultimate software developmentkit.
They're using a large languagemodel.
(12:06):
Mm-hmm.
Right?
And, there's all kinds ofproblems with that from like a
cost and scaling and accuracyperspective when you really
start dealing with large volumesof data.
Got it.
So we love LLMs, we use them forall kinds of things.
Mm-hmm.
but for the core of what we do,we make our own language models
to be able to do large scaledata extraction.
Nathan C (12:24):
Okay.
And you've been doing it forquite a while, right?
Yeah.
Like it's not just a 2022.
Tim O'Connol (12:30):
We started the company back in 2016, really
during the sort of the deeplearning, machine learning boom
and not knowing what was comingnext.
but you know what?
It's been great.
the market has changed andeveryone wants AI now and
recognizes the value it canbring.
So, yeah.
it's been a great year for us.
Nathan C (12:46):
I've, one of the big themes that I've heard this year
was like, very specifically ai,but assisted with a human,
right?
Like there's, there's some kindof co-pilot, there's some kind
of ride along, there's some kindof, so it.
I'm glad to hear it.
I think it, it's that level of,user experience that people are
(13:06):
and trust, right?
That people are ready for.
One of the questions I am,always interested to ask in this
sort of age of AI applications,it's like trust and identity
start to get really messy.
You know, we, some of the firststuff we're seeing is like deep
fakes and Sora.
you know, there's even likesocial media networks dedicated
to content that's imagined.
(13:28):
and so what's real, what'struth?
Are these actually experts?
Do you have a perspective on, ontrust and safety?
Tim O'Connol (13:38):
The way we've built our software and it's
always been like this has been,if we're extracting something
from a clinical note, let's sayit's a report and it says that
you've got appendicitis.
We're really big on going, thiswas the term we found in the
report.
Here's where it started andended in the report.
Mm-hmm.
When we're doing like, you know,review for humans to look at our
output, we're always like,here's the sentence it came
(14:00):
from.
Here's the part of the, thissection of the report it came
from to really be very sort ofdeterministic and reliable.
And be able to give people allthe data they need.
And then in, in our, our AIassisted, chart, review package,
it really has like a multi-payview where, because you can
always have problems with, forexample, like optical character
recognition of a fax or a tableor colored background.
(14:23):
So we're always like, here's thetext we extracted and here's the
source document side by side.
Yeah.
So that the, the person doingthe adjudication can really
always get back to the source.
Nathan C (14:34):
Do you track those click-throughs into those
additional sources?
Tim O'Connol (14:40):
Not, not yet.
That, that's not reallysomething our customers have
sort of required and, yeah.
Nathan C (14:45):
I find it, it's an interesting data point to follow
on to that.
Like, we have the resources.
Are they actually being used?
An interesting point is like,helps, just document.
I like it that on a user side.
sorry to be pushing on yourproduct.
It's
Tim O'Connol (14:58):
all
Nathan C (14:58):
It's all good.
it's something I ask everyone,just'cause I'm, very curious.
I always like to ask about theorigin story, and you definitely
started with a little bit ofyour professional practice, but
I'm curious, like not allradiologists start a business
startup, tech startup, startworking in ai.
(15:22):
can you talk a little bit aboutwhat inspired that bridge?
Tim O'Connol (15:25):
Yeah, absolutely.
So I did a fellowship inemergency and trauma radiology
and in imaging informatics backin 20 12, 20 13.
And I met, A team of an opensource medical, natural language
processing software package.
called CTAs.
It's a great piece of software.
And I was back at work at thehospital I worked at and in
radiology, particularly inemergency radiology, we don't
(15:46):
get great histories from ourEmerge docs.
We might get a history that justsays, rule out trauma.
And we're sort of like, wellthat's, you know, I'll look at
the x-ray, but that doesn't giveme a whole lot to go on.
Or I'll read the CT scan and Ican assume it's trauma.
So what I needed was I needed adecent patient history.
And so I used this open sourcemedical NLP software.
(16:07):
We processed some radiology.
Ports.
I wrote an app, I integrated itwith our radiology package, so
you could just click on a buttonand see the patient history.
Mm-hmm.
But the problem was that themedical NLP software really
wasn't accurate enough that wecould trust the output.
And so around that time I metone of my co-founders, Dr.
(16:27):
Sarkar, a lifetime machinelearning researcher, and he was
like, oh yeah, that's a hugeproblem.
Like accuracy's really hard.
And so we really hit it offpersonally and professionally
and, and with a couple otherguys.
we ended up starting emtelligentSo really the goal was just if
we can get at what's in thoseclinical notes, we can solve so
many problems in healthcare.
Nathan C (16:49):
The, frustrated Doctor as a founder Is like very much,
a theme from this year'sinterviews.
there's a very interestingconnection between, those who
are working with data andinformatics.
And I was literally digging inthis morning.
I'm like, what is theinformatics to founder pipeline?
'cause there's like a.
There's something about when youcan, when your work is based in
(17:11):
data mm-hmm.
It can feel like a tool in allkinds of ways.
Yeah.
so, awesome.
Let's, let's get, back to someof the theme of the, reason why
we're here.
Sure.
At Health.
as one of the larger, events forhealthcare technology, it's a
great place to kind of connectwith the pulse, and share, ideas
back, to the community.
(17:31):
A glow up is a notabletransformation like a rebirth.
I'm curious, is there a glow upor a, a notable goal that you
have for the healthcare and thehealth tech industry, at large?
Where'd you like to see us go?
Tim O'Connol (17:45):
I would love to see, years ago I watched a
documentary and was talkingabout someone who worked at Ford
Motor Company after World Warii, and he was saying that they
used to pay their invoices byweighing them.
Right.
So we've gotta put a pound ofpaper here from US Steel.
Let's cut them a check for amillion dollars and see if that
covers it.
I feel like in healthcare today,we're sort of at that level of
(18:07):
healthcare analytics.
We need better data and sothat's what I'm really
passionate about, is gettingpeople the data they need.
No matter what their role is inhealthcare from those
unstructured notes, right?
Yeah.
We have clinicians, doctors,nurses, so much ultra to valid
health professionals who spendso much time documenting
important stuff about ourpatients, and yet we can't use
that data.
So that's the transformation Iwant to see.
Nathan C (18:30):
Yeah.
just thinking back to likerecent times when family was in
the hospital, the frequency.
That like, you'd have to givekind of a whole history of like,
why we're here.
Like basically everyprofessional needed to do the
same interview so that they feltthey had the notes they needed.
Right.
And like, which made it great ifthey were on shift for a few
(18:53):
days in a row.
Right.
But it was hard if they wereonly there for a few hours at a
time.
Tim O'Connol (18:57):
So very, very hard.
Right.
that's a great way We could helpimprove the patient experience.
Right.
Nathan C (19:01):
Amazing.
So, thank you for that.
and for, for emtelligent mm-hmm.
Let's take it in a little bit.
what are the, what are thegoals?
Tim O'Connol (19:09):
So the six month go up, I think right now is,
we've got 20 25.
If 2024 and 2023 were years ofexperimentation.
with the healthcare industry,2025 is like the year of
implementation.
So, we've done a really largenumber of pilots in the last,
six to eight weeks.
Even that, pilots always takelike a couple months of, legal
(19:30):
and this sort of stuff to evenstart before someone even sends
you data.
'cause it can always contain PHIand this sort of thing.
So we've done a huge number of.
Pilots and we have a huge numberstill to go.
And really the glow up I want tosee is those pilots moving to
implementations.
Nathan C (19:46):
Amazing.
and congrats.
That's like hard earned.
if you could be even a gengeneral answer to this, it's
amazing.
there's a lot of folks in likethe deep tech world that I work
in that believe that healthcare,that their solution is a, has a
healthcare use case, and oftenthey're looking for those first
pilots.
(20:07):
And so when you're talking abouta lot of work and planning and
time and partnerships goes intoit, is there.
Some broad generalizations.
Like is there a sort of a rulethat you keep for like, well, a
partnership takes four months.
Takes seven months, how muchdoes it take to do a partnership
in healthcare?
Tim O'Connol (20:26):
You know what, it really depends on the customer.
Mm-hmm.
and when you're dealing withlike large mega.
Core kind of customers.
Mm-hmm.
getting through contracting canbe six to eight months
Nathan C (20:35):
just on the contract,
Tim O'Connol (20:36):
contracting.
Nathan C (20:36):
That's the paperwork part.
After you know what you want todo after you
Tim O'Connol (20:39):
Yeah.
So it's not me complaining aboutit.
when you're working with, say,startups, they tend to be far
more agile, and can work a wholelot faster.
we've been really trying tooptimize our processes for that.
like working with legal counselto simplify contracts and doing
all these things oftentimes aspart of the pilot process,
customers will come back and go,oh, like, well, you know, you
made some assumptions here andthey weren't all correct.
(21:00):
Right?
And we're like, yeah, you'reabsolutely right.
We did.
'cause we didn't wanna spend, aweek in planning before we got
you stuff.
We wanted to get you some dataand iterate on that.
Thank you.
Nathan C (21:10):
Oh, that's the way you do it.
Yeah.
Start small.
Yeah, exactly.
Start in one building, not four.
Amazing.
we're still on the thematicquestions.
Okay.
heroes and Legends is the themeof HLTH 2025.
I use this as an opportunity toask people about the heroes,
legends and mentors that haveguided their journey and
(21:32):
entrepreneurism.
How have mentors and legends inthis space, impacted your
journey?
Tim O'Connol (21:39):
I've been so lucky to have so many great mentors,
in healthcare and engineering ifI had to provide a shout out to
any, mentor, one of them wouldbe, a wonderful cardiac
electrophysiologist I workedwith as a medical student.
Great guy named Dr.
Paul Dorian at St.
Michael's Hospital in Toronto.
Amazing.
he was an incredible mentor tome he was trying to solve
complex physiology problems andjust allowed me to use curiosity
(22:04):
and hard work to try and solveproblems with him for a couple
different summers, doingresearch.
And so, yeah, I mean, it wasjust such a positive,
overwhelmingly positiveexperience.
Nathan C (22:13):
One of the reasons why I really like that question is
because.
the power of somebody saying, Ibelieve you.
Keep trying, keep going, keeptrying is like so powerful.
And like to start with someonewho's like, be curious.
Yeah.
Go learn, right?
Is like what a gift.
Right.
Amazing.
last question is totallyoptional, but do you have a
spicy healthcare hot take?
Tim O'Connol (22:36):
Not so much for healthcare.
I'm a little bit worried aboutai.
Right now we're starting to seesomething called vendor
financing, where people who sellstuff are loaning money to
people who want to buy stuff.
But during the internet, boom, Iworked for a company called
Nortel Networks.
I was a network engineer.
And it was an amazing place towork, but that company got into
(22:57):
trouble by doing vendorfinancing.
And that can be a sign thatthere's a bubble going on, and
I'm worried about that.
So I don't think it's gonna, youknow, bubbles happen.
We know that bubbles burst.
We know that.
For what we do.
Nothing has really changed otherthan a lot of customer interest
with the AI ramp up.
(23:18):
So, you know, we're gonna keepdoing what we're doing, but I'm
just worried
Nathan C (23:24):
as a leader who's been through a bubble or so.
Mm-hmm.
do you have a strategy or advicefor how to approach it or how do
you approach, the possibility ofdisruption.
Tim O'Connol (23:35):
you know, the worst part I saw about that the
internet boom, was there were somany wonderful people that I
worked with at Nortel and theyhad their, literally their life
savings in stock.
Work in that company and I thinka lot of people got hurt.
And so I would say no matter howgood your own hype is, always
(23:56):
expect that externalexternalities can happen.
Nathan C (23:58):
Yeah.
Tim O'Connol (23:59):
Right.
Nathan C (23:59):
Oh my gosh, that's fantastic.
Tim, it has been such a greattime chatting with you on the
HTLH Tech Glow Up.
Thank you.
Thank you so much, for joiningus.
We've got one last thing to do.
Alright.
One second, we're gonna clap itout.
Okay.
One.
Two, three.
Awesome.
Thank you.
Hey, it's Nathan.
Welcome to another specialdouble episode from the HLTH
Conference.
Each one of these episodes islaunching Monday and Thursday,
and we'll feature conversationswith two different CEOs so
welcome, listen in
Oren Nissim (00:21):
this is ultimately about something that is very
fundamentally available toeverybody.
It's cheap, it's covered byinsurance.
We can all do it these days.
There's no reason why anybodyliving with a chronic condition
home has to feel that lonelyanymore.
That's Glow Up.
Nathan C (00:39):
There's no reason that a patient at home should be
isolated and lonely.
Like we have the tools toprovide better care.
Fantastic.
Hello and welcome to the HLTHTech Glow Up.
I'm Nathan C, and today I amtalking with Oren Nissim of
Brooke Health.
Oren Thank you so much forjoining me today.
Oren Nissim (00:59):
Thanks so much for having me.
Nathan C (01:00):
Oh my goodness.
So Oren could you, introduceyourself and a little bit of
what you do at Brook Health?
Sure.
We're Brook Health and, we do,remote care for people with
chronic conditions.
We work as part of the, healthsystem, so we primarily work
with providers.
Mostly primary care, taking careof people with, diabetes,
(01:22):
hypertension, CHF, COPD, multichronics, at home, and
extending, primary care to thehome, essentially.
Amazing.
tell me a little bit about yourrole at, Brook Health as well.
Oren Nissim (01:35):
Sure.
So I'm the co-founder and CEO.
Grew a life as a patient.
so I have a complicated versionof, type two diabetes myself,
which I guess got me to be moreintertwined in it, and at some
point decided to, stopcomplaining and do something
about it instead.
Nathan C (01:50):
Oh my gosh.
That is like the classic founderstory, right?
Noticed a problem and just hadto go fix it for yourself.
could you actually dive in, tothat origin story?
I'm curious, how did you makethe decision to take this
entrepreneurial step?
Oren Nissim (02:10):
I was very involved in the worlds of, wearable and,
the ideas behind the fact thatwe today are wearing and
carrying with us a lot ofdevices at all.
Translate a lot of information,very ambiently.
And the idea is that we don'tneed to actually ask a lot of
questions in order to justunderstand what's going on.
So we could very easily comeinto a, if this, then that.
(02:33):
and so that has been thebeginning of the idea and from
that on it evolved.
Nathan C (02:38):
a amazing, I find, a lot of times, especially as the
general public is coming upagainst ideas about artificial
intelligence, right?
What do I even do with it,right?
You see the videos of Sora AImaking quirky, I'm riding a
rhinoceros video", and it's ifthat's what AI does, like how
does it affect me?
I'm a nurse, right?
(02:59):
Like I don't make rhino videos.
can you talk a little bit, a lotof tools are, using ai.
talking about predictive,talking about ways that we can,
get ahead of health outcomesbefore there's a problem.
How do you, approach this andthe use of AI in your work at
Brook?
Oren Nissim (03:19):
Sure.
So we track a lot of informationand we analyze a lot of data in
the background that allows us tounderstand anomaly.
Understand if a person has theseparticular conditions, what are
we actually looking for?
What is the behavior that wewould like to see?
And then what are those gaps?
I think the biggest problem thatcomes up, and I think you see it
in the show a lot.
There's lots of technologyavailable, lots of technology,
(03:41):
and there is lots of people thatwill sell all sorts of things.
And the key question is reallywhat I like to describe is the
accountability gap.
At the end of the day, you are aconsumer who bought X, Y, Z, and
you have it, and now you'refeeling empowered and all of a
sudden you got these signals,but you don't wanna be your own
doctor and you don't necessarilywanna be Google your doctor
(04:02):
either or check GPT.
Now the the new thing.
How do you actually close thatgap, which is where a lot of
this now needs to be broughtback in somehow into the care
system, but not as noise becauseit's just a huge amount of
noise, but actually a signal.
And what is the signal that wecan actually work on?
So I think, AI first use case isin order to decipher noise for
(04:26):
signal.
And then be able to bring itinto, as long as we're not
willing to give up the job oftrusting a provider, which I
don't think we're that close todoing, be able to do it in such
a way that is meaningful to aprovider and actually cut away
on their time as opposed toadding one more thing on a time
We as a company needs a lot ofwork into how to do that better,
(04:49):
because I think that's the mainfailure point with all this
technology available.
Nathan C (04:54):
And I'm super curious about that.
What is your primary way thatyou like to give time back to
doctors?
How do you achieve that?
Oren Nissim (05:02):
You know you're gonna be, we collect probably
thousands of data points a dayfrom every single one of our
patients, and the reality isthat just sifting through that
is a huge amount of pain.
AI is then able to ultimatelycompare this with what is the
baseline, what are we actuallylooking for, and where are those
anomalies?
On those anomalies, we haveessentially a care team that
will do the first level ofanalysis to ultimately bring it
(05:24):
to the provider on the, Hey,here's the next.
Medical decision for you toactually make, as opposed to,
here's just a huge amount ofdata, go read it and here's a
few summary notes, we want toget very quickly into the, given
this, here is the recommendationon that, what would you like to
decide on as the medicalprofessional?
Nathan C (05:44):
And is that, it sounds like there's the AI layer and
then there's also that humanlayer that's doing that
additional validation andresearch, presenting those
opportunities.
So that's happening in thebackground as the provider might
be doing other tasks, reviewingthe records and so can, isn't
just taking things off of theirplate, but potentially working
(06:05):
in the background working whenthey're not necessarily so that
the care is always on.
Yeah.
Regardless of the practitioner.
Oren Nissim (06:12):
Yeah, and I think that ultimately it's about how
do you actually createguardrails?
For this approach.
Because the reality is that weall feel like we're not willing
to hand the keys just yet, maybeover a period of time, and then
who can be in the middle and thesame time, how do you make that
efficient?
You can train health teams to domore, but that means you're
essentially, again, adding upmore time.
(06:33):
As opposed to taking time away.
So how to create it in asupervised fashion a very safe
environment so that people canreally trust what they're
seeing.
And that I think is a lot ofthat inner work that we have to
do these days.
Nathan C (06:45):
Amazing.
Awesome.
so moving on to some of the morebranded questions.
The show is the HTLH Tech GlowUp, and let's start at the
industry level.
I'm curious, what is the healthglow up that you wanna see for
this industry?
Oren Nissim (07:00):
what I wanna see is people not buying the tech for
the tech.
And I'm saying it as a techentrepreneur.
this is not about the widget,it's about the what are you
actually gonna do with it.
So it's about the use case.
And I think glow up here isabout the fact that I see a lot
(07:20):
of people coming in and saying Iget what you can do.
How does that actually make itmore useful to me?
And essentially forcing theindustry to come back and say,
we worked out the use case tothe point that we're not just
adding more time and adding morecost.
We're actually trying to deliveran ROI.
But assuming these new tools arein place, one of the cool
examples that you could seeright now is in the scribe world
(07:42):
whereby, we used to use zoomin.
Now you can use machines.
Okay, but how does that flowactually integrates well such
that we don't need to thenre-review this, we can actually
trust this and this is whereit's the next level.
It's not exactly you can't justuse AI to summarize.
You gotta make sure that it doesit well and to the fact that
it's actually useful.
Nathan C (08:03):
Thank you for that.
so the next question takes theGlow Up a little bit more
personal, a little bit closer tohome.
What's the glow up Oren, thatyou're looking to make and that
you're looking to make for BrookHealth?
Oren Nissim (08:16):
Thanks.
We've been on a mission for along period of time.
I started this with a mission,which is that people living life
with multiple chronic conditionsat home, we need some level of
agency.
ultimately this is about how dowe use these tools that are at
our.
Disposal to basically, getbetter agency in what we do and
how can we bridge that gap witha provider that we trust, but
(08:39):
have something in the middlethat can help us.
for me, this is ultimately aboutsomething that is very
fundamentally available toeverybody.
It's cheap, it's covered byinsurance.
We can all do it these days.
There's no reason why anybodyliving with a chronic condition
home has to feel that lonelyanymore.
That's Glow Up.
Nathan C (08:59):
There's no reason that a patient at home should be
isolated and lonely.
Like we have the tools toprovide better care.
Fantastic.
no pressure on this one, butsocial media loves a good hot
take.
Do you have a healthcare hottake, a spicy opinion, that you
(09:19):
wanna share about the industry?
Oren Nissim (09:24):
I will pay close attention to what's happening
with Medicare Advantage andpeople walking, some people
walking away from it, somepeople walking into it, and I
think that's gonna be aninteresting next few months.
Nathan C (09:36):
Interesting.
I love it.
Oren it's been such a pleasureto chat with you today.
Thank you for sharing, aboutyour AI powered Insights and a
tool that is clearing all of theinformation away so you don't
just have signal, but youactually have data.
thank you so much for visitingwith us on the HLTH Tech Glow
(09:58):
Up.
Yeah, thanks so much.
Appreciate it.
Amazing.
1, 2, 3.
See I'm the worst.
Hello and welcome to the HLTHTech Glow Up.
I am Nathan C and today I'mtalking with Tim O'Connell of
emtelligent Tim, thank you somuch for joining me today on the
(10:18):
Health Tech Glow Up.
Thanks,
Tim O'Connol (10:20):
Nathan.
Nathan C (10:20):
You have a pretty fantastic smile for this late in
the HLTH event.
Thanks for joining me early onday three.
how's your HLTH been?
Tim O'Connol (10:30):
HLTh has been great.
I love HLTh It's one of myfavorite healthcare conferences.
I think it's just got a greatvibe and.
I spend a lot of time in darkrooms and hospitals and, and I
appreciate, you know, we needmore unicorns in hospitals.
Yes.
I love it.
Nathan C (10:44):
Sure.
And the work that you do atemtelligent?.
Tim O'Connol (10:47):
I'm a practicing radiologist.
I work clinically one to twodays a week.
I'm also a CEO of a 9-year-oldstartup called emtelligent we're
a medical language AI company,so we do a couple things,
primarily large scale dataextraction from clinical notes,
and we also do AI assisted chartreview.
Nathan C (11:05):
I stopped by your booth a little bit earlier
today, and I was looking at someof the materials, and it seems
like your approach to AI isn'tmaybe the same approach that you
might see, with, a lot of otherfolks who are talking about AI
and health records.
Can you kind of explain, yourdifferentiator there?
Tim O'Connol (11:25):
Sure.
So as a company that does largescale data extraction, we work
with a lot of people across thehealthcare spectrum.
So it can be producers of noteslike health systems or it can be
consumers and users of thatdata, like payers and life
sciences companies.
And what they often have aproblem is they'll often be
like, we have several billionclinical documents that we need
processed.
(11:46):
And there's obviously a lot ofdata there.
So we've been spending yearsmaking our software so we can
really extract the meaning andcoded data from very large
volumes of data and do it.
Very quickly and efficiently andat a cost that works for our
customers.
So, yeah, I think a lot ofpeople are now using the
ultimate software developmentkit.
They're using a large languagemodel.
(12:06):
Mm-hmm.
Right?
And, there's all kinds ofproblems with that from like a
cost and scaling and accuracyperspective when you really
start dealing with large volumesof data.
Got it.
So we love LLMs, we use them forall kinds of things.
Mm-hmm.
but for the core of what we do,we make our own language models
to be able to do large scaledata extraction.
Nathan C (12:24):
Okay.
And you've been doing it forquite a while, right?
Yeah.
Like it's not just a 2022.
Tim O'Connol (12:30):
We started the company back in 2016, really
during the sort of the deeplearning, machine learning boom
and not knowing what was comingnext.
but you know what?
It's been great.
the market has changed andeveryone wants AI now and
recognizes the value it canbring.
So, yeah.
it's been a great year for us.
Nathan C (12:46):
I've, one of the big themes that I've heard this year
was like, very specifically ai,but assisted with a human,
right?
Like there's, there's some kindof co-pilot, there's some kind
of ride along, there's some kindof, so it.
I'm glad to hear it.
I think it, it's that level of,user experience that people are
(13:06):
and trust, right?
That people are ready for.
One of the questions I am,always interested to ask in this
sort of age of AI applications,it's like trust and identity
start to get really messy.
You know, we, some of the firststuff we're seeing is like deep
fakes and Sora.
you know, there's even likesocial media networks dedicated
to content that's imagined.
(13:28):
and so what's real, what'struth?
Are these actually experts?
Do you have a perspective on, ontrust and safety?
Tim O'Connol (13:38):
The way we've built our software and it's
always been like this has been,if we're extracting something
from a clinical note, let's sayit's a report and it says that
you've got appendicitis.
We're really big on going, thiswas the term we found in the
report.
Here's where it started andended in the report.
Mm-hmm.
When we're doing like, you know,review for humans to look at our
output, we're always like,here's the sentence it came
(14:00):
from.
Here's the part of the, thissection of the report it came
from to really be very sort ofdeterministic and reliable.
And be able to give people allthe data they need.
And then in, in our, our AIassisted, chart, review package,
it really has like a multi-payview where, because you can
always have problems with, forexample, like optical character
recognition of a fax or a tableor colored background.
(14:23):
So we're always like, here's thetext we extracted and here's the
source document side by side.
Yeah.
So that the, the person doingthe adjudication can really
always get back to the source.
Nathan C (14:34):
Do you track those click-throughs into those
additional sources?
Tim O'Connol (14:40):
Not, not yet.
That, that's not reallysomething our customers have
sort of required and, yeah.
Nathan C (14:45):
I find it, it's an interesting data point to follow
on to that.
Like, we have the resources.
Are they actually being used?
An interesting point is like,helps, just document.
I like it that on a user side.
sorry to be pushing on yourproduct.
It's
Tim O'Connol (14:58):
all
Nathan C (14:58):
It's all good.
it's something I ask everyone,just'cause I'm, very curious.
I always like to ask about theorigin story, and you definitely
started with a little bit ofyour professional practice, but
I'm curious, like not allradiologists start a business
startup, tech startup, startworking in ai.
(15:22):
can you talk a little bit aboutwhat inspired that bridge?
Tim O'Connol (15:25):
Yeah, absolutely.
So I did a fellowship inemergency and trauma radiology
and in imaging informatics backin 20 12, 20 13.
And I met, A team of an opensource medical, natural language
processing software package.
called CTAs.
It's a great piece of software.
And I was back at work at thehospital I worked at and in
radiology, particularly inemergency radiology, we don't
(15:46):
get great histories from ourEmerge docs.
We might get a history that justsays, rule out trauma.
And we're sort of like, wellthat's, you know, I'll look at
the x-ray, but that doesn't giveme a whole lot to go on.
Or I'll read the CT scan and Ican assume it's trauma.
So what I needed was I needed adecent patient history.
And so I used this open sourcemedical NLP software.
(16:07):
We processed some radiology.
Ports.
I wrote an app, I integrated itwith our radiology package, so
you could just click on a buttonand see the patient history.
Mm-hmm.
But the problem was that themedical NLP software really
wasn't accurate enough that wecould trust the output.
And so around that time I metone of my co-founders, Dr.
(16:27):
Sarkar, a lifetime machinelearning researcher, and he was
like, oh yeah, that's a hugeproblem.
Like accuracy's really hard.
And so we really hit it offpersonally and professionally
and, and with a couple otherguys.
we ended up starting emtelligentSo really the goal was just if
we can get at what's in thoseclinical notes, we can solve so
many problems in healthcare.
Nathan C (16:49):
The, frustrated Doctor as a founder Is like very much,
a theme from this year'sinterviews.
there's a very interestingconnection between, those who
are working with data andinformatics.
And I was literally digging inthis morning.
I'm like, what is theinformatics to founder pipeline?
'cause there's like a.
There's something about when youcan, when your work is based in
(17:11):
data mm-hmm.
It can feel like a tool in allkinds of ways.
Yeah.
so, awesome.
Let's, let's get, back to someof the theme of the, reason why
we're here.
Sure.
At Health.
as one of the larger, events forhealthcare technology, it's a
great place to kind of connectwith the pulse, and share, ideas
back, to the community.
(17:31):
A glow up is a notabletransformation like a rebirth.
I'm curious, is there a glow upor a, a notable goal that you
have for the healthcare and thehealth tech industry, at large?
Where'd you like to see us go?
Tim O'Connol (17:45):
I would love to see, years ago I watched a
documentary and was talkingabout someone who worked at Ford
Motor Company after World Warii, and he was saying that they
used to pay their invoices byweighing them.
Right.
So we've gotta put a pound ofpaper here from US Steel.
Let's cut them a check for amillion dollars and see if that
covers it.
I feel like in healthcare today,we're sort of at that level of
(18:07):
healthcare analytics.
We need better data and sothat's what I'm really
passionate about, is gettingpeople the data they need.
No matter what their role is inhealthcare from those
unstructured notes, right?
Yeah.
We have clinicians, doctors,nurses, so much ultra to valid
health professionals who spendso much time documenting
important stuff about ourpatients, and yet we can't use
that data.
So that's the transformation Iwant to see.
Nathan C (18:30):
Yeah.
just thinking back to likerecent times when family was in
the hospital, the frequency.
That like, you'd have to givekind of a whole history of like,
why we're here.
Like basically everyprofessional needed to do the
same interview so that they feltthey had the notes they needed.
Right.
And like, which made it great ifthey were on shift for a few
(18:53):
days in a row.
Right.
But it was hard if they wereonly there for a few hours at a
time.
Tim O'Connol (18:57):
So very, very hard.
Right.
that's a great way We could helpimprove the patient experience.
Right.
Nathan C (19:01):
Amazing.
So, thank you for that.
and for, for emtelligent mm-hmm.
Let's take it in a little bit.
what are the, what are thegoals?
Tim O'Connol (19:09):
So the six month go up, I think right now is,
we've got 20 25.
If 2024 and 2023 were years ofexperimentation.
with the healthcare industry,2025 is like the year of
implementation.
So, we've done a really largenumber of pilots in the last,
six to eight weeks.
Even that, pilots always takelike a couple months of, legal
(19:30):
and this sort of stuff to evenstart before someone even sends
you data.
'cause it can always contain PHIand this sort of thing.
So we've done a huge number of.
Pilots and we have a huge numberstill to go.
And really the glow up I want tosee is those pilots moving to
implementations.
Nathan C (19:46):
Amazing.
and congrats.
That's like hard earned.
if you could be even a gengeneral answer to this, it's
amazing.
there's a lot of folks in likethe deep tech world that I work
in that believe that healthcare,that their solution is a, has a
healthcare use case, and oftenthey're looking for those first
pilots.
(20:07):
And so when you're talking abouta lot of work and planning and
time and partnerships goes intoit, is there.
Some broad generalizations.
Like is there a sort of a rulethat you keep for like, well, a
partnership takes four months.
Takes seven months, how muchdoes it take to do a partnership
in healthcare?
Tim O'Connol (20:26):
You know what, it really depends on the customer.
Mm-hmm.
and when you're dealing withlike large mega.
Core kind of customers.
Mm-hmm.
getting through contracting canbe six to eight months
Nathan C (20:35):
just on the contract,
Tim O'Connol (20:36):
contracting.
Nathan C (20:36):
That's the paperwork part.
After you know what you want todo after you
Tim O'Connol (20:39):
Yeah.
So it's not me complaining aboutit.
when you're working with, say,startups, they tend to be far
more agile, and can work a wholelot faster.
we've been really trying tooptimize our processes for that.
like working with legal counselto simplify contracts and doing
all these things oftentimes aspart of the pilot process,
customers will come back and go,oh, like, well, you know, you
made some assumptions here andthey weren't all correct.
(21:00):
Right?
And we're like, yeah, you'reabsolutely right.
We did.
'cause we didn't wanna spend, aweek in planning before we got
you stuff.
We wanted to get you some dataand iterate on that.
Thank you.
Nathan C (21:10):
Oh, that's the way you do it.
Yeah.
Start small.
Yeah, exactly.
Start in one building, not four.
Amazing.
we're still on the thematicquestions.
Okay.
heroes and Legends is the themeof HLTH 2025.
I use this as an opportunity toask people about the heroes,
legends and mentors that haveguided their journey and
(21:32):
entrepreneurism.
How have mentors and legends inthis space, impacted your
journey?
Tim O'Connol (21:39):
I've been so lucky to have so many great mentors,
in healthcare and engineering ifI had to provide a shout out to
any, mentor, one of them wouldbe, a wonderful cardiac
electrophysiologist I workedwith as a medical student.
Great guy named Dr.
Paul Dorian at St.
Michael's Hospital in Toronto.
Amazing.
he was an incredible mentor tome he was trying to solve
complex physiology problems andjust allowed me to use curiosity
(22:04):
and hard work to try and solveproblems with him for a couple
different summers, doingresearch.
And so, yeah, I mean, it wasjust such a positive,
overwhelmingly positiveexperience.
Nathan C (22:13):
One of the reasons why I really like that question is
because.
the power of somebody saying, Ibelieve you.
Keep trying, keep going, keeptrying is like so powerful.
And like to start with someonewho's like, be curious.
Yeah.
Go learn, right?
Is like what a gift.
Right.
Amazing.
last question is totallyoptional, but do you have a
spicy healthcare hot take?
Tim O'Connol (22:36):
Not so much for healthcare.
I'm a little bit worried aboutai.
Right now we're starting to seesomething called vendor
financing, where people who sellstuff are loaning money to
people who want to buy stuff.
But during the internet, boom, Iworked for a company called
Nortel Networks.
I was a network engineer.
And it was an amazing place towork, but that company got into
(22:57):
trouble by doing vendorfinancing.
And that can be a sign thatthere's a bubble going on, and
I'm worried about that.
So I don't think it's gonna, youknow, bubbles happen.
We know that bubbles burst.
We know that.
For what we do.
Nothing has really changed otherthan a lot of customer interest
with the AI ramp up.
(23:18):
So, you know, we're gonna keepdoing what we're doing, but I'm
just worried
Nathan C (23:24):
as a leader who's been through a bubble or so.
Mm-hmm.
do you have a strategy or advicefor how to approach it or how do
you approach, the possibility ofdisruption.
Tim O'Connol (23:35):
you know, the worst part I saw about that the
internet boom, was there were somany wonderful people that I
worked with at Nortel and theyhad their, literally their life
savings in stock.
Work in that company and I thinka lot of people got hurt.
And so I would say no matter howgood your own hype is, always
(23:56):
expect that externalexternalities can happen.
Nathan C (23:58):
Yeah.
Tim O'Connol (23:59):
Right.
Nathan C (23:59):
Oh my gosh, that's fantastic.
Tim, it has been such a greattime chatting with you on the
HTLH Tech Glow Up.
Thank you.
Thank you so much, for joiningus.
We've got one last thing to do.
Alright.
One second, we're gonna clap itout.
Okay.
One.
Two, three.
Awesome.
Thank you.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!