August 5, 2025 • 31 mins
This week on How I AI, I am joined by Dr. Akash Parvatikar, a computational pathologist and medical imaging scientist who works at the intersection of AI and healthcare.
Akash earned his Ph.D. through the joint Carnegie Mellon–University of Pittsburgh School of Medicine in Computational Biology and has been developing advanced AI workflows that support doctors in analyzing medical images and improving diagnostic accuracy.
We explore how large language models (LLMs), Python, and agentic workflows are being used in modern medical imaging labs, the move toward digital pathology, and the new possibilities that come with remote collaboration and telepathology. This episode offers a behind-the-scenes look at how scientists are approaching AI adoption in ways that are practical, regulated, and patient‑focused.
🔥 Topics We Cover
- How computational pathology is shaping the future of medical imaging
- The shift from physical slides to digital workflows in pathology
- How AI tools assist specialists by streamlining repetitive image analysis
- Agentic workflows and why they are not just for large language models
- The role of telepathology in connecting specialists globally
🛠️ AI Tools and Workflow Akash Mentions
- Python for building AI scripts
- NumPy for numerical computing
- Scikit-learn for machine learning tasks
- LLMs and agentic workflows for diagnostics
- Digital pathology platforms for slide digitization and quality review
📲 Connect with Dr. Akash Parvatikar
- LinkedIn: Akash Parvatikar
🎧 If you enjoyed this episode, please leave a review and share it with a friend working in healthcare, AI, or life sciences. How I AI is here to spotlight the quiet builders and brilliant minds using artificial intelligence to elevate real human impact.
Want to be featured?
Have a compelling AI use case or experimental workflow? Apply to be a guest at howiaipodcast.com.
For listeners ready to dive deeper into AI, The Collective offers a 7-Day Free Trial that includes the AI Mastery Challenge, a vault of advanced trainings, and recordings from past webinars.
More About Brooke:
Website: brookex.com
LinkedIn: Brooke Gramer
More About the Podcast:
Instagram: howiai.podcast
Website: howiaipodcast.com
"How I AI" is a concept and podcast series created and produced by Brooke Gramer of EmpowerFlow Strategies LLC. All rights reserved.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Brooke (00:03):
Welcome to How I AI the podcast featuring real people,
real stories, and real AI inaction.
I'm Brooke Gramer your host andguide on this journey into the
real world impact of artificialintelligence.
For over 15 years, I've workedin creative marketing, events,
and business strategy wearingall the hats.
(00:23):
I know the struggle of trying toscale and manage all things
without burning out, but here'sthe game changer, AI.
This isn't just a podcast, How IAI is a community.
A space where curious minds likeyou can come together, share
ideas, and I'll also be bringingyou exclusive discounts, free
trials and insider resources soyou can test drive the latest
(00:46):
tools and tech yourself.
Because AI isn't just a trend,it's a shift.
The sooner we embrace it, themore freedom, creativity, and
opportunities we'll unlock.
How I AI is brought to you inpartnership with The Collective
designed to accelerate yourlearning and AI adoption.
I joined the collective and it'scompletely catapulted my
(01:08):
learning, expanded my network,and show me what's possible with
ai.
Whether you're just starting outor looking to refine your AI
strategy, The Collective givesyou the resources to grow.
Stay tuned to learn more at the end of this episode, or
check the show notes for myexclusive invite link..
Today on How I AI, we aregetting an insider look at
(01:29):
artificial intelligence fromsomeone who sits right at the
crossroads of medicine, science,and technology.
Dr.
Akash and his work at HistoWizin New York is changing the way
we diagnose cancer.
Akash has a front row seat tohow AI is being adopted across
industries and has a superpowerwith helping teams figure out
(01:52):
which technologies will actuallymove the needle for them.
If you are curious about thereal advantages of how
scientists are adopting AI andwhat that means for making
smarter, more intentional techdecisions, this episode is for
you.
Welcome to another episode ofHow I AI I'm your host, Brooke
(02:14):
Gramer.
Today I have a very excitingguest.
His name is Dr.
Akash Parvatikar.
He's a scientist who works withcomputers and medicine to help
doctors understand diseases likecancer.
Dr.
Akash.
I'm so happy to have you.
Akash Parvatikar (02:32):
Thank you so much, Brooke, for inviting me
here and I'm sure we'll behaving a lot of interesting
conversation.
Right now I'm working at theintersection of AI cancer
diagnosis to help clinicians andresearchers to accurately
diagnose and also help themdiagnose more quicker with the
help of AI.
Once again, thank you so muchfor having me here,
Brooke (02:55):
yes.
I was so excited when youreached out expressing interest
to be on the podcast becausesomething I actually didn't
share with you in our intro callyet I've had three very close
family members be affected bycancer.
So when you shared a bit aboutyour background, I was so
interested to dive deeper ofthis.
Intersection of AI and medicineand AI for good because there
(03:19):
are a lot of negativeconnotations that are happening
right now with people adaptingAI, and I'm so happy to shine
more positive light onto thistopic.
And so let's kick it off here,
Akash Parvatikar (03:31):
Sure.
Brooke (03:31):
I'd love for you to just expand on your background and
how you got into medicine andwhere you're at now with, with
AI?
Akash Parvatikar (03:40):
Oh yeah absolutely.
So I did my bachelor's inelectrical engineering then got
my master's in informationscience.
And during my master's I wasdoing a research project.
In the com bio department atUniversity of Pittsburgh, where
I was the first time exposed tothe biological dataset for drug
discovery.
(04:00):
And then after doing theresearch work at the University
of Pittsburgh, I joined the PhDprogram at the Joint Carnegie
Mellon and University ofPittsburgh Medical School.
In computational biology whereinthe problem at hand was to
understand why is it difficultto diagnose preinvasive breast
(04:20):
cancer, and why is there so muchof disagreement among the
doctors in the diseasediagnosis?
So for me, I, A wanted tounderstand, at the primary level
why is it so difficult?
And B, being a computerscientist, how I can build
solutions for the domain expertwho are pathologists to help and
(04:42):
solve that problem.
So for me, it was more problemdriven, which is the cancer
diagnosis and then, buildingstate of the art solutions to
help both understand the problemand then build solutions on top
of that.
Brooke (04:54):
Wow.
A lot to unpack there.
I have a standard set ofquestions I ask everybody, but
it's different every time based
Akash Parvatikar (05:02):
Yeah.
Brooke (05:02):
level of integration.
And with you, you're on such ahigh level.
My first question is, and maybeyou can break it down to an
easily digestible way so we canunderstand it.
What kind of tools and machinelearning do you use with your
role?
What is your quote unquote techstack look What kind of
technology are you using to doyour job?
Akash Parvatikar (05:23):
Right.
So while I was doing my PhD, andthis is a little over four years
back, so I had to build my ownframeworks from scratch.
So all that I used was thePython programming language and
few packages NumPy and PsychicLearn and all that to load the
data.
But I built my own framework, myown objective function and all
(05:45):
that.
And that gave me a very inepunderstanding of what it takes
to build AI.
For medicine and also buildexplainable and trustworthy AI,
which is so important because,yes you could use the solutions
out there, which is alreadyavailable.
But when it comes to regulatoryapprovals and having trust in
(06:06):
the AI system the clinicianswant to know, if you're using
any open source software, whatit was trained on, where did the
data come from, who did theannotations and all that.
So far, my phD work.
It was from scratch.
Did not use any open sourcesoftware.
But right now where I'mcurrently working at HistoWiz we
do use some open source softwarefor disease detection and all
(06:28):
that.
And it's we, we use thetransformer networks, we use
unit for segmentation.
And we have been sort ofventuring into these LLM models
as well.
How can we build agent AI tosolve very specific tasks.
So it is quite diverse withinthe company that I work for.
But I think the, tech stack it'sproblem driven.
(06:49):
It depends on what you're tryingto solve.
Brooke (06:52):
Maybe you can dive deeper on what is changing right
now in the medical space when itcomes to implementing AI and
specifically in your field withcancer diagnosis.
Akash Parvatikar (07:02):
Right.
So what's changing?
There's a big change happeningactually.
So up until recently, and evennow, as you might know, that the
pathologist look at the physicalglass lights under a microscope.
So these are non digitized.
These are physical raw slidesthat they use microscope to look
at the tissue and do thediagnosis.
(07:25):
But then what changed in 2017 isthat US FDA approved using
digital scanners for primarydiagnosis.
So what that means is now youcan take these large resolution
images.
Of your physical glass slide anda clinician could look at the
digital image at a highresolution.
(07:46):
So that opened up a Pandora oflike..
Good opportunities to share thedata quite easily implement AI
and do your further analysis.
So so that's what's changing,which is going from manual
anatomic pathology to digitalanatomic pathology.
So that's the switch that'shappening within my field.
(08:09):
So say the slides are alreadydigitized.
There have been a lot ofsolutions out there to run AI
for diagnosis, but I thinkthat's where there are some
trust issues between the AIdevelopers and the pathologist.
So if the AI is directlydiagnosing a particular case,
because a cancer diagnosis is socomplex, that is, there are so
(08:32):
many things happening withinyour biopsy tissue and no AI
system is is able to accuratelydetect cancer for you.
But what's happening right nowis AI is doing a really good job
in those tasks, which takes alot of time.
For the doctors to detect, forexample, if they're looking at,
some very specific regionswithin the entire image, now AI
(08:54):
can do a very quick job in doingthat and also quantifying things
for you.
So right now all the AI tools inmy field or in what's coming up
is these assistive tools thandecision making tools.
That's coming into play.
For cancer diagnosis, again, I'mmore than happy to expand upon
what I said, but this is wherethe field is going right now.
Brooke (09:17):
So you are expressing that we're all moving to
digital, which is saving a lotof time, and you're able to find
the issues in the tissues moreaccurately and So my next
question, because you chatted alittle bit about the trust
factor when it comes into thisuse of AI in medicine.
Can you maybe expand on that?
Akash Parvatikar (09:39):
Sure.
I think not just in medicineBrooke.
And I the way you said the issuein the tissue, using ai, I think
that that should be, I don'tknow, a tagline for one of my
posters or something.
But so talking about the trustfactor, I think not just in the
field of medicine, any AIadoption.
There is this hard approach.
Where AI takes the decision foryou.
(09:59):
And there is a soft approachwherein, you could still use the
AI in your routine tasks, butyou are the final decision
maker.
And, you could use it as a toolto better help in your own
workflow, then create a wholenew workflow where it's
completely done by AI.
So in the field of pathology,what's happening is that say
that AI gives you a 90% accuracyof detecting, say, breast
cancer.
(10:21):
But then for FDA to approve thatAI system, we should absolutely
be sure of those 10% cases wherethe AI does not do a good job
because if we do not know that,then we have no idea when a
patient will be wronglydiagnosed.
So it's okay if the AI does agood job, say 80% of the time,
(10:43):
but we should be very sure whatare the failure modes of this AI
system when the AI does not do agood job?
It could be the race of a womanundergoing a breast biopsy.
It could be the region wherethey came from.
So the AI was trained on thepopulation of US and Europe and
say the patient belongs to WestAsia.
Will the AI do a good job?
(11:04):
So I think in facilitating aregulatory approval, all of this
has to be accounted for.
Because it's good to know, okay,fine.
This is where the AI does a goodjob, so we might just implement
over there, but we cannot usethe validation report for a
certain use case and expand itworldwide and make it a general
solution.
So, especially in the field ofmedicine, there are very,
specific AI use cases that'scoming into product uh, mode,
(11:29):
and even the regulatoryapprovals are given for that
specific case.
Up so much that, if some AI istrained on, say, NYU data and
they get an FDA approval, sothey could treat only NYU
patient with that AI system andthey cannot go to say Columbia
or, university of Miami Hospitaland all that.
So I think it's highly specific.
Brooke (11:49):
And you touched on a use case just now.
Could you maybe share a coupleuse cases, maybe one or two of
success you've had integratingAI into your research and into
your work.
Akash Parvatikar (12:04):
Sure.
I think, uh, the first largesuccess that we observed is
integrating an AI to detectquality issues.
Brooke (12:11):
Hmm mm.
Akash Parvatikar (12:12):
In the data so we, at HistoWiz, we produce
tissue data from multiple organsand species.
And for us it takes a lot oftime to detect quality issues
within the image.
So say a particular image couldbe bloody, or it might have some
artifacts on there.
Or it could have some folds andbreaks and all that.
(12:34):
Now, having a person trying tolook at thousands of images,
identifying these, I don't know,pen marks, folds, blurs, and all
that, takes a lot of time.
Because we are dealing with avery large sized image.
So we have developed andintegrated a novel AI solution.
To detect a very simple thing,which is quality issues in this
(12:55):
imaging data because thedevelopment of AI cancer
detection and all that comesnext because if you don't have a
good quality data.
It's garbage in, garbage out.
So you cannot trust what the AIis doing, further on.
So now that's one big successthat we, we have got.
And we are also having veryinteresting discussions with not
just other AI companies who wantto use our AI tool for Quality
(13:19):
control, but also from Collegeof American pathologist, which
is CAP.
And even at the federal levelthey're interested in using our
quality AI tool because we havebeen using it every single day
within our company.
And we have a large clienteleand, to produce good quality
data.
So that's a big win.
And the second big win with theintegration of AI that we have
(13:41):
had within my company is the AImarketplace.
So what we have offered isespecially in the field of
pathology, we cannot force aparticular AI tool to be used by
doctors.
We should give them theflexibility as to the kind of AI
tool that they want to run.
And since there are so many AItools coming up in the field of
(14:03):
cancer diagnosis that, we shouldmake our platform flexible to
accommodate those multiple AItools.
So what we have created on ourplatform is this AI marketplace
wherein.
Multiple AI companies all overthe world could come onto our
platform as buttons, and we makeit quickly accessible to the
(14:24):
doctor and research communityand say they want to run tumor
detection on breast from one AIcompany, and they want to run,
tumor detection on lung fromanother AI company.
And they want to compare theresult.
They want to understand how atumor looks on breast versus a
lung.
Now they could easily do that onthe platform.
So that's another workflowadoption that we have seen.
(14:45):
And the, the pharmaceuticalcompanies and biotech companies
are loving that idea.
So giving, giving theflexibility in their hands.
Brooke (14:52):
That's fantastic.
So it sounds as a scientist it'svery beneficial, but what about
when it comes down to actuallytangibly saving lives and
preventing cancer?
How is AI supporting that?
Akash Parvatikar (15:06):
Yeah.
Well, I think that's a billiondollar question that you just
asked Brooke.
But then Again, simplifyingthings here.
So I am in the field of accuratecancer diagnosis but not in the
treatment side of things.
So all, my efforts and energygoes into the solving the
problem that say a patient comesto a clinic, gets their biopsy
(15:27):
done we want to able to at leastdiagnose that particular case
correctly because then, thetreatment sort of gets affected
by the diagnosis that happens.
So yes AI is playing a largerole in drug discovery and,
getting these new therapeuticdrugs and all that into the
market.
But there is a very, very bigproblem at hand here, which is.
(15:49):
Even for breast cancer, close toa million women in US undergo
breast biopsy every year.
Brooke (15:55):
Mm-hmm.
Akash Parvatikar (15:56):
It so happens that majority of the cases are
pre-invasive.
That means they, they haven'thad got a breast cancer yet, but
it has a high recurrence ofbecoming a breast cancer in the
next 10 years or 15 years.
So now this pre-invasive is anotorious stage of breast cancer
wherein there is a lot ofdisagreement among the doctors.
(16:16):
Whether to call it a stage oneor a stage two should they go
for a surgery or, just lifestylechanges and all that.
So, so now all of my work isfocused on using AI to bring
doctors into consensus, havingconcordance among the doctors to
call out for a diagnosis.
And AI is definitely helpingthat.
And one more thing is that notjust US, but all over the world,
(16:39):
we are facing a shortage ofpathologists.
So what that means is nowpathologists are signing out
more cases than they can handle.
So that increases their burdenand that could also cause
burnout and internally tomisdiagnoses.
So AI is definitely helping inautomating some of the routine
tasks, which for them might havetaken hours to do.
So that's where we are gettingat.
Brooke (17:01):
Thank you for breaking that down further for
Akash Parvatikar (17:03):
yeah.
Brooke (17:04):
Getting more into the mindset of the future.
What do you hope to see AI makespossible in the field of
medicine in the next five to 10years?
Akash Parvatikar (17:17):
Couple things here, Brooke.
So one thing is we'll be seeinga much quicker and accurate
diagnosis.
Not just for cancer, but itcould be other disease types
which involves x-ray scanning orCT scanning, being able to
detect features within a CTscan, x-ray scan and all that,
that will all be automated,digitized, and, it'll be much
more quicker.
(17:38):
And that's one thing which is anachievable win with the help of
AI.
Second thing is in the field ofdrug discovery, where they trial
out so many combinations ofdrugs before it comes into the
market.
And some of it was taking a lotof time just because the
computation power was not there.
And with the help of LLMs rightnow, they can pass through
(17:59):
millions of data points and, tryto find that one compound much
more quicker.
So again.
It's about speed that I'mtalking about for diagnosis and
treatment.
But third very interesting thingis that what AI could help is
making us understand the fieldof medicine much better than we
(18:19):
already did.
For example, if it is cancerdiagnosis, we with our naked
eyes, might not have seen thesefeatures in the past, which the
computer can now easily pick up.
So this will lead to updatingthe medical records and medical
books with the help of AI that,I think looking at this feature
makes more sense than what wepaid attention to early on.
(18:41):
So, that's more of a questionmark,
Brooke (18:43):
Mm-hmm.
Akash Parvatikar (18:43):
exciting, open-ended but very exciting
time to be in.
So.
Brooke (18:48):
Very exciting indeed.
My next question for you is toshare about any challenges
you're facing or have faced withusing AI.
Have there been any momentswhere things didn't go to plan
or
Akash Parvatikar (19:04):
Yeah.
Brooke (19:06):
I think that's important to touch on because
Akash Parvatikar (19:09):
Yeah.
Brooke (19:09):
think that AI is just the light switch, green go
answer for them, but you knowit's a process.
So if you could share maybeabout those, any initial
lessons?
Akash Parvatikar (19:22):
Yeah.
Again a great question that youasked Brooke.
I wanna keep it simple andwhenever we talk about AI, I
think we need to talk aboutdata.
Because there is no data, thereis no AI.
So, I think to reframe yourquestion whether I have faced
the challenges with AI in myexperience I have faced
(19:43):
challenges with the data that Ideal with.
These are pathology dataset.
And to give a context of thekind of data that I deal with
whenever a biopsy is done by aphysician and when they take a
digital image of that tissueslide at 20 x or 40 x
resolution.
So if there was a giant printerthat you might have and you had
(20:04):
to print out this entire imageon that printer, every single
image occupies anywhere betweentwo to three tennis courts.
So that is the volume of that'sone image and every single
image, the size of it isanywhere between 500 megabytes
to several gigabytes.
(20:24):
So compared to an image that youmight take on your phone, it
might be, I don't know, say 10megabytes or something.
Every single image in on withpathology is few hundred,
megabytes towards two gigabytes.
So I had to face a lot ofchallenge in trying to get the
data in the right format to notjust build AI, but run any AI
(20:45):
system on top of that, having tochop up the images into multiple
square patches and then cookstitch it back, and making sure
that, your system is quitepowerful to handle this large
amounts of data.
So I think that was onechallenge that I faced on, which
took me a lot of time.
Even more so than building theAI system because, and this is
(21:05):
an open problem, not just withmy research Google is trying to
solve it.
Microsoft is trying to solve it.
Now Google has their own digitalpathology wing.
Again, a lot of funding.
And Nvidia is getting into thisfield as well, wherein they're
building special systems tohandle this large volumes of
data.
So, different people might be indifferent field, but I think
having this fair understandingof data and the challenges that
(21:28):
come with it is very crucial toknow very early on.
And the second challenge for meis any AI that I build I sort of
talk to domain experts veryearly on.
I don't delay that because whatI feel is that at the end of the
day, they are the ones who willbe using this AI system.
So you cannot build your AI,have a product in place, and
(21:49):
then do these customers for betatesting and all that.
I feel that, if you have to betatest a product in the space of
AI you have to recruit domainexperts early on.
For example, if you're buildinga AI, I don't know, virtual
clothing app, right?
You have to talk to fashiondesigners very early on.
How do they think of fashion?
How do they think of thesedifferent style so you have to
(22:11):
sort of have this communicationbetween the AI and the domain
expert.
So that's one thing that Ilearned during my experience.
Brooke (22:18):
One quick question because
Akash Parvatikar (22:20):
Yeah,
Brooke (22:20):
sparked in me as
Akash Parvatikar (22:21):
sure.
Brooke (22:22):
How do you prevent hallucination because you shared
that you are dealing.
With tennis courts of data, howare you preventing
hallucination?
Akash Parvatikar (22:33):
So, one thing is, I do not deal with LLMs for
pathology images, but there arecompanies out there, very
interesting companies who havebuilt LLMs for pathology images
and I'm more than happy to leavea link to that to this episode
in the description.
So mine is more computer visionalgorithms.
For me using AI to identifyfeatures within the image and
(22:56):
trying to understand theimportance of those features in
for a particular diagnosis.
Brooke (23:01):
Thank you for
Akash Parvatikar (23:02):
Yeah.
Brooke (23:03):
In your personal experience, are physicians
responding to the growth andevolution of AI?
It's starting to creep in, andcompletely rework the way that
we experience healthcare.
Maybe you can, at a very highlevel, what you've seen amongst
(23:24):
your peers.
Is it excitement?
Is it fear?
I'm just curious to hear what isthe general response to AI and
its advancements in the medicalindustry?
Akash Parvatikar (23:38):
Narrowing down to the field of pathology
because medical industry is so,so wide.
Brooke (23:44):
Yes.
Akash Parvatikar (23:44):
so talking about pathology, Brooke there
are current practitioners whohave just graduated from med
school and some earlypathologists who are quickly
adopting digital solutions.
But there still are experiencedpathologists with over 35 years
of experience wherein they arestill hesitating to go digital
(24:04):
because they're completely wiredin to look at the slides under
the microscope, zoom in, zoomout pan and all that, and now
they're a bit hesitant becausefor them they are set on that
and they trust their decisionmaking skills.
So for them it's okay, whyshould I look at a digital file?
I have been doing this fordecades.
So there I think we are having avery hard time going digital
(24:28):
because there's a learning curvefor them, but they are at the
stage of their career whereinthey are they have mastered the
field and so that comes back tothe hospital where if some of
the pathologists are not willingto go digital then the hospital
might take a back seat.
We are trying to convince to godigital to make these AI systems
(24:48):
work.
It is two different challenges.
One challenge is first producethe data from physical to
digital, and then convincingwhat AI solutions will actually
help them.
The industry is trying to figureout baby steps into this.
But to give our viewers a goodinformation on this aspect, Mayo
Clinic have scanned all theirphysical slides.
(25:09):
This is over 12 million slides.
NYU is scanning all theirphysical slides now, so they're
going entire digital.
The pathology department, UMiami is doing a great job and
we have been talking to U Miamias well.
And they're going very big indigitizing their physical
repository as well.
Hospitals are seeing value.
(25:29):
Mount Sinai is doing it.
Sloan Kettering is doing it.
But these are all largehospitals who are doing it and I
guess once they set a trend,maybe other hospitals and
biotechs will follow.
So.
Brooke (25:41):
Have you already seen the misdiagnosis rate go down
quite dramatically?
Akash Parvatikar (25:48):
Dramatically, no,
Brooke (25:49):
No.
Akash Parvatikar (25:49):
But, but we have conducted experiments
wherein, the rate ofdisagreement is going down.
So again, what I mean by that isgoing digital.
What's happening is say I'm inTexas.
As a patient, but I have thispathologist who is in New York
City and say, Miami.
Now by going digital, I caneasily transfer my data to these
(26:10):
two doctors in, in seconds andthey can look at the same image
at the same time and give theiropinion second opinions.
So that's game changing.
And what's game changing is thatnow the rural patients are
getting access to the best care.
From the best in classpathology.
So they do not have to travel tothe rural areas.
(26:31):
They can sit in the comforts oftheir house and still diagnose
the patients.
All that they need is a scannerto digitize the flights and, get
it to them.
So it's called telepathology,which is, being able to remotely
diagnose this patient.
So that's happening.
And there there hasn't been aclean report out talking about
how much a discordance has comedown or, misdiagnosis has come
(26:52):
down.
But if there's something thatcomes up, I'm more than happy to
share that with you.
Brooke (26:56):
Thank you for that insight.
That already sounds like such aadvantage and I love that you
highlighted that as well.
In addition to time saved, it'sthe access to this care which I
think must be very rewarding tobe able to be a doctor and be
able to support patients andmake a difference in someone's
(27:18):
life every day.
What's one key takeaway that youwant listeners to have when it
comes to integrating AI?
Akash Parvatikar (27:27):
So I have two answers to that.
Brooke.
One is if you are interested todevelop your own AI.
The one key takeaway is try tounderstand the data, spend some
time, spend some quality time inunderstanding the nuances of
data that you are dealing with.
It could be in any field thatyou are in even if you are not
(27:47):
aware of the domain sit withKOLs, key opinion leaders and
domain experts.
What are the challenges thatthey are facing because think of
AI as a solution to a problem.
And I think problem comes first,solution comes next.
So don't use AI to create newproblems and then find
solutions.
Try to find existing problems,and then use AI to find quicker
(28:08):
solutions.
So my key takeaway I would saymeditate over the problem for
some time before you develop asolution.
And again also spend some timein the kind of AI you are
building with, because right nowwith agentic AI, LLMs and all
that, that things happen soquickly wherein you might miss
(28:29):
the train on some very keyaspects and you never know, why
the AI might be hallucinating insome cases.
It's finding the right solutionfor everything uh, LLMs might
not be the answer.
if it's low hanging fruit, youcan build a different solutions
very easily.
And that's something what I do Ialso act as an AI advisor to
some of these startup companieswherein I just advise them how
do you think about an AIsolution?
(28:51):
For the problem that you haveand actually evaluate for
yourself, using the ai, is itsort of improving your work
efficiency?
Brooke (28:58):
Thank you for that final point there, and to your point,
I will absolutely link whateverit is that you wanna share in
the show notes, but my finalclosing question for you is.
How can listeners reach out toyou and connect and learn more
about your work?
Akash Parvatikar (29:15):
They can connect me on LinkedIn.
They can reach out to me viaemail as well.
I'm easily accessible.
I am in New York, so if any ofthe viewers are in New York, New
Jersey happy to grab a coffee aswell.
And if any of you are in thestage wherein you have an idea
and you want to develop an AIbased product, and if you want a
(29:36):
quick maybe 30 minute chat on,what it makes the best sense
what kind of AI to implement andall that.
Happy to chat about that aswell.
Because I have spent years withmy research in building the AI
system that you don't have tospend those many hours.
And I, I'm more than happy tocollaborate with, with our other
products as well.
(29:56):
So.
Brooke (29:57):
That's so generous.
Thank you, and I reallyappreciate you taking the time
to chat about a topic a lot ofpeople are very interested in
learning about AI in thisintersection and specifically
within your niche of pathology,so appreciate the time that you
took to speak to me today.
Thank you.
Akash Parvatikar (30:16):
Well, thank you.
Thank you so much, Brooke.
It was a pleasure talking toyou.
And a lot of interestingquestions.
It gives me a high levelperspective of where I am in the
field that I'm in and, becauseagain, whatever AI we build we
are answerable to the generalpublic, not just the expert, so
the kind of question that youask, thank you so much for that.
And I'm sure many of them outthere who might not be aware
(30:38):
that, there is a lot of goodpromise out there, especially in
the field of cancer diagnosisand treatment.
And there is definitely a bighope.
Brooke (30:46):
Wonderful.
So exciting to hear.
Akash Parvatikar (30:48):
Yeah.
Brooke (30:48):
you.
Akash Parvatikar (30:49):
Thank you so much.
Brooke (30:50):
Wow I hope today's episode opened your mind to
what's possible with AI.
Do you have a cool use case onhow you're using AI and wanna
share it?
DM me.
I'd love to hear more andfeature you on my next podcast.
Until next time, here's toworking smarter, not harder.
See you on the next episode ofHow I AI.
This episode was made possiblein partnership with the
(31:12):
Collective AI, a communitydesigned to help entrepreneurs,
creators, and professionalsseamlessly integrate AI into
their workflows.
One of the biggest game changersin my own AI journey was joining
this space.
It's where I learned, connectedand truly enhanced my
understanding of what's possiblewith ai.
(31:33):
And the best part, they offermultiple membership levels to
meet you where you are.
Whether you want to DIY, your AIlearning or work with a
personalized AI consultant foryour business, The Collective
has you covered.
Learn more and sign up using myexclusive link in the show
notes.
Welcome to How I AI the podcast featuring real people,
real stories, and real AI inaction.
I'm Brooke Gramer your host andguide on this journey into the
real world impact of artificialintelligence.
For over 15 years, I've workedin creative marketing, events,
and business strategy wearingall the hats.
(00:23):
I know the struggle of trying toscale and manage all things
without burning out, but here'sthe game changer, AI.
This isn't just a podcast, How IAI is a community.
A space where curious minds likeyou can come together, share
ideas, and I'll also be bringingyou exclusive discounts, free
trials and insider resources soyou can test drive the latest
(00:46):
tools and tech yourself.
Because AI isn't just a trend,it's a shift.
The sooner we embrace it, themore freedom, creativity, and
opportunities we'll unlock.
How I AI is brought to you inpartnership with The Collective
designed to accelerate yourlearning and AI adoption.
I joined the collective and it'scompletely catapulted my
(01:08):
learning, expanded my network,and show me what's possible with
ai.
Whether you're just starting outor looking to refine your AI
strategy, The Collective givesyou the resources to grow.
Stay tuned to learn more at the end of this episode, or
check the show notes for myexclusive invite link..
Today on How I AI, we aregetting an insider look at
(01:29):
artificial intelligence fromsomeone who sits right at the
crossroads of medicine, science,and technology.
Dr.
Akash and his work at HistoWizin New York is changing the way
we diagnose cancer.
Akash has a front row seat tohow AI is being adopted across
industries and has a superpowerwith helping teams figure out
(01:52):
which technologies will actuallymove the needle for them.
If you are curious about thereal advantages of how
scientists are adopting AI andwhat that means for making
smarter, more intentional techdecisions, this episode is for
you.
Welcome to another episode ofHow I AI I'm your host, Brooke
(02:14):
Gramer.
Today I have a very excitingguest.
His name is Dr.
Akash Parvatikar.
He's a scientist who works withcomputers and medicine to help
doctors understand diseases likecancer.
Dr.
Akash.
I'm so happy to have you.
Akash Parvatikar (02:32):
Thank you so much, Brooke, for inviting me
here and I'm sure we'll behaving a lot of interesting
conversation.
Right now I'm working at theintersection of AI cancer
diagnosis to help clinicians andresearchers to accurately
diagnose and also help themdiagnose more quicker with the
help of AI.
Once again, thank you so muchfor having me here,
Brooke (02:55):
yes.
I was so excited when youreached out expressing interest
to be on the podcast becausesomething I actually didn't
share with you in our intro callyet I've had three very close
family members be affected bycancer.
So when you shared a bit aboutyour background, I was so
interested to dive deeper ofthis.
Intersection of AI and medicineand AI for good because there
(03:19):
are a lot of negativeconnotations that are happening
right now with people adaptingAI, and I'm so happy to shine
more positive light onto thistopic.
And so let's kick it off here,
Akash Parvatikar (03:31):
Sure.
Brooke (03:31):
I'd love for you to just expand on your background and
how you got into medicine andwhere you're at now with, with
AI?
Akash Parvatikar (03:40):
Oh yeah absolutely.
So I did my bachelor's inelectrical engineering then got
my master's in informationscience.
And during my master's I wasdoing a research project.
In the com bio department atUniversity of Pittsburgh, where
I was the first time exposed tothe biological dataset for drug
discovery.
(04:00):
And then after doing theresearch work at the University
of Pittsburgh, I joined the PhDprogram at the Joint Carnegie
Mellon and University ofPittsburgh Medical School.
In computational biology whereinthe problem at hand was to
understand why is it difficultto diagnose preinvasive breast
(04:20):
cancer, and why is there so muchof disagreement among the
doctors in the diseasediagnosis?
So for me, I, A wanted tounderstand, at the primary level
why is it so difficult?
And B, being a computerscientist, how I can build
solutions for the domain expertwho are pathologists to help and
(04:42):
solve that problem.
So for me, it was more problemdriven, which is the cancer
diagnosis and then, buildingstate of the art solutions to
help both understand the problemand then build solutions on top
of that.
Brooke (04:54):
Wow.
A lot to unpack there.
I have a standard set ofquestions I ask everybody, but
it's different every time based
Akash Parvatikar (05:02):
Yeah.
Brooke (05:02):
level of integration.
And with you, you're on such ahigh level.
My first question is, and maybeyou can break it down to an
easily digestible way so we canunderstand it.
What kind of tools and machinelearning do you use with your
role?
What is your quote unquote techstack look What kind of
technology are you using to doyour job?
Akash Parvatikar (05:23):
Right.
So while I was doing my PhD, andthis is a little over four years
back, so I had to build my ownframeworks from scratch.
So all that I used was thePython programming language and
few packages NumPy and PsychicLearn and all that to load the
data.
But I built my own framework, myown objective function and all
(05:45):
that.
And that gave me a very inepunderstanding of what it takes
to build AI.
For medicine and also buildexplainable and trustworthy AI,
which is so important because,yes you could use the solutions
out there, which is alreadyavailable.
But when it comes to regulatoryapprovals and having trust in
(06:06):
the AI system the clinicianswant to know, if you're using
any open source software, whatit was trained on, where did the
data come from, who did theannotations and all that.
So far, my phD work.
It was from scratch.
Did not use any open sourcesoftware.
But right now where I'mcurrently working at HistoWiz we
do use some open source softwarefor disease detection and all
(06:28):
that.
And it's we, we use thetransformer networks, we use
unit for segmentation.
And we have been sort ofventuring into these LLM models
as well.
How can we build agent AI tosolve very specific tasks.
So it is quite diverse withinthe company that I work for.
But I think the, tech stack it'sproblem driven.
(06:49):
It depends on what you're tryingto solve.
Brooke (06:52):
Maybe you can dive deeper on what is changing right
now in the medical space when itcomes to implementing AI and
specifically in your field withcancer diagnosis.
Akash Parvatikar (07:02):
Right.
So what's changing?
There's a big change happeningactually.
So up until recently, and evennow, as you might know, that the
pathologist look at the physicalglass lights under a microscope.
So these are non digitized.
These are physical raw slidesthat they use microscope to look
at the tissue and do thediagnosis.
(07:25):
But then what changed in 2017 isthat US FDA approved using
digital scanners for primarydiagnosis.
So what that means is now youcan take these large resolution
images.
Of your physical glass slide anda clinician could look at the
digital image at a highresolution.
(07:46):
So that opened up a Pandora oflike..
Good opportunities to share thedata quite easily implement AI
and do your further analysis.
So so that's what's changing,which is going from manual
anatomic pathology to digitalanatomic pathology.
So that's the switch that'shappening within my field.
(08:09):
So say the slides are alreadydigitized.
There have been a lot ofsolutions out there to run AI
for diagnosis, but I thinkthat's where there are some
trust issues between the AIdevelopers and the pathologist.
So if the AI is directlydiagnosing a particular case,
because a cancer diagnosis is socomplex, that is, there are so
(08:32):
many things happening withinyour biopsy tissue and no AI
system is is able to accuratelydetect cancer for you.
But what's happening right nowis AI is doing a really good job
in those tasks, which takes alot of time.
For the doctors to detect, forexample, if they're looking at,
some very specific regionswithin the entire image, now AI
(08:54):
can do a very quick job in doingthat and also quantifying things
for you.
So right now all the AI tools inmy field or in what's coming up
is these assistive tools thandecision making tools.
That's coming into play.
For cancer diagnosis, again, I'mmore than happy to expand upon
what I said, but this is wherethe field is going right now.
Brooke (09:17):
So you are expressing that we're all moving to
digital, which is saving a lotof time, and you're able to find
the issues in the tissues moreaccurately and So my next
question, because you chatted alittle bit about the trust
factor when it comes into thisuse of AI in medicine.
Can you maybe expand on that?
Akash Parvatikar (09:39):
Sure.
I think not just in medicineBrooke.
And I the way you said the issuein the tissue, using ai, I think
that that should be, I don'tknow, a tagline for one of my
posters or something.
But so talking about the trustfactor, I think not just in the
field of medicine, any AIadoption.
There is this hard approach.
Where AI takes the decision foryou.
(09:59):
And there is a soft approachwherein, you could still use the
AI in your routine tasks, butyou are the final decision
maker.
And, you could use it as a toolto better help in your own
workflow, then create a wholenew workflow where it's
completely done by AI.
So in the field of pathology,what's happening is that say
that AI gives you a 90% accuracyof detecting, say, breast
cancer.
(10:21):
But then for FDA to approve thatAI system, we should absolutely
be sure of those 10% cases wherethe AI does not do a good job
because if we do not know that,then we have no idea when a
patient will be wronglydiagnosed.
So it's okay if the AI does agood job, say 80% of the time,
(10:43):
but we should be very sure whatare the failure modes of this AI
system when the AI does not do agood job?
It could be the race of a womanundergoing a breast biopsy.
It could be the region wherethey came from.
So the AI was trained on thepopulation of US and Europe and
say the patient belongs to WestAsia.
Will the AI do a good job?
(11:04):
So I think in facilitating aregulatory approval, all of this
has to be accounted for.
Because it's good to know, okay,fine.
This is where the AI does a goodjob, so we might just implement
over there, but we cannot usethe validation report for a
certain use case and expand itworldwide and make it a general
solution.
So, especially in the field ofmedicine, there are very,
specific AI use cases that'scoming into product uh, mode,
(11:29):
and even the regulatoryapprovals are given for that
specific case.
Up so much that, if some AI istrained on, say, NYU data and
they get an FDA approval, sothey could treat only NYU
patient with that AI system andthey cannot go to say Columbia
or, university of Miami Hospitaland all that.
So I think it's highly specific.
Brooke (11:49):
And you touched on a use case just now.
Could you maybe share a coupleuse cases, maybe one or two of
success you've had integratingAI into your research and into
your work.
Akash Parvatikar (12:04):
Sure.
I think, uh, the first largesuccess that we observed is
integrating an AI to detectquality issues.
Brooke (12:11):
Hmm mm.
Akash Parvatikar (12:12):
In the data so we, at HistoWiz, we produce
tissue data from multiple organsand species.
And for us it takes a lot oftime to detect quality issues
within the image.
So say a particular image couldbe bloody, or it might have some
artifacts on there.
Or it could have some folds andbreaks and all that.
(12:34):
Now, having a person trying tolook at thousands of images,
identifying these, I don't know,pen marks, folds, blurs, and all
that, takes a lot of time.
Because we are dealing with avery large sized image.
So we have developed andintegrated a novel AI solution.
To detect a very simple thing,which is quality issues in this
(12:55):
imaging data because thedevelopment of AI cancer
detection and all that comesnext because if you don't have a
good quality data.
It's garbage in, garbage out.
So you cannot trust what the AIis doing, further on.
So now that's one big successthat we, we have got.
And we are also having veryinteresting discussions with not
just other AI companies who wantto use our AI tool for Quality
(13:19):
control, but also from Collegeof American pathologist, which
is CAP.
And even at the federal levelthey're interested in using our
quality AI tool because we havebeen using it every single day
within our company.
And we have a large clienteleand, to produce good quality
data.
So that's a big win.
And the second big win with theintegration of AI that we have
(13:41):
had within my company is the AImarketplace.
So what we have offered isespecially in the field of
pathology, we cannot force aparticular AI tool to be used by
doctors.
We should give them theflexibility as to the kind of AI
tool that they want to run.
And since there are so many AItools coming up in the field of
(14:03):
cancer diagnosis that, we shouldmake our platform flexible to
accommodate those multiple AItools.
So what we have created on ourplatform is this AI marketplace
wherein.
Multiple AI companies all overthe world could come onto our
platform as buttons, and we makeit quickly accessible to the
(14:24):
doctor and research communityand say they want to run tumor
detection on breast from one AIcompany, and they want to run,
tumor detection on lung fromanother AI company.
And they want to compare theresult.
They want to understand how atumor looks on breast versus a
lung.
Now they could easily do that onthe platform.
So that's another workflowadoption that we have seen.
(14:45):
And the, the pharmaceuticalcompanies and biotech companies
are loving that idea.
So giving, giving theflexibility in their hands.
Brooke (14:52):
That's fantastic.
So it sounds as a scientist it'svery beneficial, but what about
when it comes down to actuallytangibly saving lives and
preventing cancer?
How is AI supporting that?
Akash Parvatikar (15:06):
Yeah.
Well, I think that's a billiondollar question that you just
asked Brooke.
But then Again, simplifyingthings here.
So I am in the field of accuratecancer diagnosis but not in the
treatment side of things.
So all, my efforts and energygoes into the solving the
problem that say a patient comesto a clinic, gets their biopsy
(15:27):
done we want to able to at leastdiagnose that particular case
correctly because then, thetreatment sort of gets affected
by the diagnosis that happens.
So yes AI is playing a largerole in drug discovery and,
getting these new therapeuticdrugs and all that into the
market.
But there is a very, very bigproblem at hand here, which is.
(15:49):
Even for breast cancer, close toa million women in US undergo
breast biopsy every year.
Brooke (15:55):
Mm-hmm.
Akash Parvatikar (15:56):
It so happens that majority of the cases are
pre-invasive.
That means they, they haven'thad got a breast cancer yet, but
it has a high recurrence ofbecoming a breast cancer in the
next 10 years or 15 years.
So now this pre-invasive is anotorious stage of breast cancer
wherein there is a lot ofdisagreement among the doctors.
(16:16):
Whether to call it a stage oneor a stage two should they go
for a surgery or, just lifestylechanges and all that.
So, so now all of my work isfocused on using AI to bring
doctors into consensus, havingconcordance among the doctors to
call out for a diagnosis.
And AI is definitely helpingthat.
And one more thing is that notjust US, but all over the world,
(16:39):
we are facing a shortage ofpathologists.
So what that means is nowpathologists are signing out
more cases than they can handle.
So that increases their burdenand that could also cause
burnout and internally tomisdiagnoses.
So AI is definitely helping inautomating some of the routine
tasks, which for them might havetaken hours to do.
So that's where we are gettingat.
Brooke (17:01):
Thank you for breaking that down further for
Akash Parvatikar (17:03):
yeah.
Brooke (17:04):
Getting more into the mindset of the future.
What do you hope to see AI makespossible in the field of
medicine in the next five to 10years?
Akash Parvatikar (17:17):
Couple things here, Brooke.
So one thing is we'll be seeinga much quicker and accurate
diagnosis.
Not just for cancer, but itcould be other disease types
which involves x-ray scanning orCT scanning, being able to
detect features within a CTscan, x-ray scan and all that,
that will all be automated,digitized, and, it'll be much
more quicker.
(17:38):
And that's one thing which is anachievable win with the help of
AI.
Second thing is in the field ofdrug discovery, where they trial
out so many combinations ofdrugs before it comes into the
market.
And some of it was taking a lotof time just because the
computation power was not there.
And with the help of LLMs rightnow, they can pass through
(17:59):
millions of data points and, tryto find that one compound much
more quicker.
So again.
It's about speed that I'mtalking about for diagnosis and
treatment.
But third very interesting thingis that what AI could help is
making us understand the fieldof medicine much better than we
(18:19):
already did.
For example, if it is cancerdiagnosis, we with our naked
eyes, might not have seen thesefeatures in the past, which the
computer can now easily pick up.
So this will lead to updatingthe medical records and medical
books with the help of AI that,I think looking at this feature
makes more sense than what wepaid attention to early on.
(18:41):
So, that's more of a questionmark,
Brooke (18:43):
Mm-hmm.
Akash Parvatikar (18:43):
exciting, open-ended but very exciting
time to be in.
So.
Brooke (18:48):
Very exciting indeed.
My next question for you is toshare about any challenges
you're facing or have faced withusing AI.
Have there been any momentswhere things didn't go to plan
or
Akash Parvatikar (19:04):
Yeah.
Brooke (19:06):
I think that's important to touch on because
Akash Parvatikar (19:09):
Yeah.
Brooke (19:09):
think that AI is just the light switch, green go
answer for them, but you knowit's a process.
So if you could share maybeabout those, any initial
lessons?
Akash Parvatikar (19:22):
Yeah.
Again a great question that youasked Brooke.
I wanna keep it simple andwhenever we talk about AI, I
think we need to talk aboutdata.
Because there is no data, thereis no AI.
So, I think to reframe yourquestion whether I have faced
the challenges with AI in myexperience I have faced
(19:43):
challenges with the data that Ideal with.
These are pathology dataset.
And to give a context of thekind of data that I deal with
whenever a biopsy is done by aphysician and when they take a
digital image of that tissueslide at 20 x or 40 x
resolution.
So if there was a giant printerthat you might have and you had
(20:04):
to print out this entire imageon that printer, every single
image occupies anywhere betweentwo to three tennis courts.
So that is the volume of that'sone image and every single
image, the size of it isanywhere between 500 megabytes
to several gigabytes.
(20:24):
So compared to an image that youmight take on your phone, it
might be, I don't know, say 10megabytes or something.
Every single image in on withpathology is few hundred,
megabytes towards two gigabytes.
So I had to face a lot ofchallenge in trying to get the
data in the right format to notjust build AI, but run any AI
(20:45):
system on top of that, having tochop up the images into multiple
square patches and then cookstitch it back, and making sure
that, your system is quitepowerful to handle this large
amounts of data.
So I think that was onechallenge that I faced on, which
took me a lot of time.
Even more so than building theAI system because, and this is
(21:05):
an open problem, not just withmy research Google is trying to
solve it.
Microsoft is trying to solve it.
Now Google has their own digitalpathology wing.
Again, a lot of funding.
And Nvidia is getting into thisfield as well, wherein they're
building special systems tohandle this large volumes of
data.
So, different people might be indifferent field, but I think
having this fair understandingof data and the challenges that
(21:28):
come with it is very crucial toknow very early on.
And the second challenge for meis any AI that I build I sort of
talk to domain experts veryearly on.
I don't delay that because whatI feel is that at the end of the
day, they are the ones who willbe using this AI system.
So you cannot build your AI,have a product in place, and
(21:49):
then do these customers for betatesting and all that.
I feel that, if you have to betatest a product in the space of
AI you have to recruit domainexperts early on.
For example, if you're buildinga AI, I don't know, virtual
clothing app, right?
You have to talk to fashiondesigners very early on.
How do they think of fashion?
How do they think of thesedifferent style so you have to
(22:11):
sort of have this communicationbetween the AI and the domain
expert.
So that's one thing that Ilearned during my experience.
Brooke (22:18):
One quick question because
Akash Parvatikar (22:20):
Yeah,
Brooke (22:20):
sparked in me as
Akash Parvatikar (22:21):
sure.
Brooke (22:22):
How do you prevent hallucination because you shared
that you are dealing.
With tennis courts of data, howare you preventing
hallucination?
Akash Parvatikar (22:33):
So, one thing is, I do not deal with LLMs for
pathology images, but there arecompanies out there, very
interesting companies who havebuilt LLMs for pathology images
and I'm more than happy to leavea link to that to this episode
in the description.
So mine is more computer visionalgorithms.
For me using AI to identifyfeatures within the image and
(22:56):
trying to understand theimportance of those features in
for a particular diagnosis.
Brooke (23:01):
Thank you for
Akash Parvatikar (23:02):
Yeah.
Brooke (23:03):
In your personal experience, are physicians
responding to the growth andevolution of AI?
It's starting to creep in, andcompletely rework the way that
we experience healthcare.
Maybe you can, at a very highlevel, what you've seen amongst
(23:24):
your peers.
Is it excitement?
Is it fear?
I'm just curious to hear what isthe general response to AI and
its advancements in the medicalindustry?
Akash Parvatikar (23:38):
Narrowing down to the field of pathology
because medical industry is so,so wide.
Brooke (23:44):
Yes.
Akash Parvatikar (23:44):
so talking about pathology, Brooke there
are current practitioners whohave just graduated from med
school and some earlypathologists who are quickly
adopting digital solutions.
But there still are experiencedpathologists with over 35 years
of experience wherein they arestill hesitating to go digital
(24:04):
because they're completely wiredin to look at the slides under
the microscope, zoom in, zoomout pan and all that, and now
they're a bit hesitant becausefor them they are set on that
and they trust their decisionmaking skills.
So for them it's okay, whyshould I look at a digital file?
I have been doing this fordecades.
So there I think we are having avery hard time going digital
(24:28):
because there's a learning curvefor them, but they are at the
stage of their career whereinthey are they have mastered the
field and so that comes back tothe hospital where if some of
the pathologists are not willingto go digital then the hospital
might take a back seat.
We are trying to convince to godigital to make these AI systems
(24:48):
work.
It is two different challenges.
One challenge is first producethe data from physical to
digital, and then convincingwhat AI solutions will actually
help them.
The industry is trying to figureout baby steps into this.
But to give our viewers a goodinformation on this aspect, Mayo
Clinic have scanned all theirphysical slides.
(25:09):
This is over 12 million slides.
NYU is scanning all theirphysical slides now, so they're
going entire digital.
The pathology department, UMiami is doing a great job and
we have been talking to U Miamias well.
And they're going very big indigitizing their physical
repository as well.
Hospitals are seeing value.
(25:29):
Mount Sinai is doing it.
Sloan Kettering is doing it.
But these are all largehospitals who are doing it and I
guess once they set a trend,maybe other hospitals and
biotechs will follow.
So.
Brooke (25:41):
Have you already seen the misdiagnosis rate go down
quite dramatically?
Akash Parvatikar (25:48):
Dramatically, no,
Brooke (25:49):
No.
Akash Parvatikar (25:49):
But, but we have conducted experiments
wherein, the rate ofdisagreement is going down.
So again, what I mean by that isgoing digital.
What's happening is say I'm inTexas.
As a patient, but I have thispathologist who is in New York
City and say, Miami.
Now by going digital, I caneasily transfer my data to these
(26:10):
two doctors in, in seconds andthey can look at the same image
at the same time and give theiropinion second opinions.
So that's game changing.
And what's game changing is thatnow the rural patients are
getting access to the best care.
From the best in classpathology.
So they do not have to travel tothe rural areas.
(26:31):
They can sit in the comforts oftheir house and still diagnose
the patients.
All that they need is a scannerto digitize the flights and, get
it to them.
So it's called telepathology,which is, being able to remotely
diagnose this patient.
So that's happening.
And there there hasn't been aclean report out talking about
how much a discordance has comedown or, misdiagnosis has come
(26:52):
down.
But if there's something thatcomes up, I'm more than happy to
share that with you.
Brooke (26:56):
Thank you for that insight.
That already sounds like such aadvantage and I love that you
highlighted that as well.
In addition to time saved, it'sthe access to this care which I
think must be very rewarding tobe able to be a doctor and be
able to support patients andmake a difference in someone's
(27:18):
life every day.
What's one key takeaway that youwant listeners to have when it
comes to integrating AI?
Akash Parvatikar (27:27):
So I have two answers to that.
Brooke.
One is if you are interested todevelop your own AI.
The one key takeaway is try tounderstand the data, spend some
time, spend some quality time inunderstanding the nuances of
data that you are dealing with.
It could be in any field thatyou are in even if you are not
(27:47):
aware of the domain sit withKOLs, key opinion leaders and
domain experts.
What are the challenges thatthey are facing because think of
AI as a solution to a problem.
And I think problem comes first,solution comes next.
So don't use AI to create newproblems and then find
solutions.
Try to find existing problems,and then use AI to find quicker
(28:08):
solutions.
So my key takeaway I would saymeditate over the problem for
some time before you develop asolution.
And again also spend some timein the kind of AI you are
building with, because right nowwith agentic AI, LLMs and all
that, that things happen soquickly wherein you might miss
(28:29):
the train on some very keyaspects and you never know, why
the AI might be hallucinating insome cases.
It's finding the right solutionfor everything uh, LLMs might
not be the answer.
if it's low hanging fruit, youcan build a different solutions
very easily.
And that's something what I do Ialso act as an AI advisor to
some of these startup companieswherein I just advise them how
do you think about an AIsolution?
(28:51):
For the problem that you haveand actually evaluate for
yourself, using the ai, is itsort of improving your work
efficiency?
Brooke (28:58):
Thank you for that final point there, and to your point,
I will absolutely link whateverit is that you wanna share in
the show notes, but my finalclosing question for you is.
How can listeners reach out toyou and connect and learn more
about your work?
Akash Parvatikar (29:15):
They can connect me on LinkedIn.
They can reach out to me viaemail as well.
I'm easily accessible.
I am in New York, so if any ofthe viewers are in New York, New
Jersey happy to grab a coffee aswell.
And if any of you are in thestage wherein you have an idea
and you want to develop an AIbased product, and if you want a
(29:36):
quick maybe 30 minute chat on,what it makes the best sense
what kind of AI to implement andall that.
Happy to chat about that aswell.
Because I have spent years withmy research in building the AI
system that you don't have tospend those many hours.
And I, I'm more than happy tocollaborate with, with our other
products as well.
(29:56):
So.
Brooke (29:57):
That's so generous.
Thank you, and I reallyappreciate you taking the time
to chat about a topic a lot ofpeople are very interested in
learning about AI in thisintersection and specifically
within your niche of pathology,so appreciate the time that you
took to speak to me today.
Thank you.
Akash Parvatikar (30:16):
Well, thank you.
Thank you so much, Brooke.
It was a pleasure talking toyou.
And a lot of interestingquestions.
It gives me a high levelperspective of where I am in the
field that I'm in and, becauseagain, whatever AI we build we
are answerable to the generalpublic, not just the expert, so
the kind of question that youask, thank you so much for that.
And I'm sure many of them outthere who might not be aware
(30:38):
that, there is a lot of goodpromise out there, especially in
the field of cancer diagnosisand treatment.
And there is definitely a bighope.
Brooke (30:46):
Wonderful.
So exciting to hear.
Akash Parvatikar (30:48):
Yeah.
Brooke (30:48):
you.
Akash Parvatikar (30:49):
Thank you so much.
Brooke (30:50):
Wow I hope today's episode opened your mind to
what's possible with AI.
Do you have a cool use case onhow you're using AI and wanna
share it?
DM me.
I'd love to hear more andfeature you on my next podcast.
Until next time, here's toworking smarter, not harder.
See you on the next episode ofHow I AI.
This episode was made possiblein partnership with the
(31:12):
Collective AI, a communitydesigned to help entrepreneurs,
creators, and professionalsseamlessly integrate AI into
their workflows.
One of the biggest game changersin my own AI journey was joining
this space.
It's where I learned, connectedand truly enhanced my
understanding of what's possiblewith ai.
(31:33):
And the best part, they offermultiple membership levels to
meet you where you are.
Whether you want to DIY, your AIlearning or work with a
personalized AI consultant foryour business, The Collective
has you covered.
Learn more and sign up using myexclusive link in the show
notes.
Popular Podcasts
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.
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.