♪ ♪ >> CATHY: ROCHESTER-BASED MAYO CLINIC IS USING ARTIFICIAL INTELLIGENCE WITH THE AIM TO SPEED UP HEALTHCARE.

TO TRY HEALTHCARE.

EARLIER THIS YEAR, MAYO LAUNCHED ITS DIGITAL PATHOLOGY DIVISION WITH THEIR PARTNER NVIDIA.

AND ON MONDAY, THE DEPARTMENT ANNOUNCED THE IMPLEMENTATION OF NVIDIA SYSTEMS TO BOOST THEIR A.I.

CAPABILITIES.

JIM ROGERS IS CEO OF MAYO CLINIC DIGITAL PATHOLOGY.

ALONG WITH HIM IS DR. MATTHEW CALLSTROM, WHO IS THE LEAD OF MAYO'S GENERATIVE A.I.

PROGRAM.

I'M GOING TO START WITH YOU, DOCTOR, BECAUSE YOU'RE A RADIOLOGIST.

SO YOU'RE USED TO SKIING SCANS AND MAKING DIAGNOSES.

HOW WILL THIS A.I.

HELP PATHOLOGISTS MAYBE SPEED UP THEIR DIAGNOSIS AND MAYBE HELP DO BETTER DIAGNOSTICS?

>> YEAH, IT'S A GREAT QUESTION AND RADIOLOGY WAS THE FIRST AREA WHERE WE STARTED TO DEVELOP A.I.

SOLUTIONS.

AND WITH THIS, WHAT WE CAN ACTUALLY DO IS SEE THINGS WITH THESE NEW A.I.

TECHNIQUES.

AND SO WE CAN TAKE AND IDENTIFY, FOR EXAMPLE, EARLY STAGE CANCERS INCLUDING PANCREATIC CANCER, AND WITH THAT, IMPROVE OUTCOMES FOR PATIENTS BECAUSE WE CAN DIAGNOSE THEM EARLIER.

WHERE THEY'RE TYPICALLY DIAGNOSED LATE.

IF YOU TRANSLATE THAT TO PATHOLOGY, WHAT YOU CAN DO IS GO FROM TODAY O LARGELY ANALOG TO NOW GOING DIGITAL AND YOU CAN ACTUALLY START TO CHARACTERIZE AND IDENTIFY CANCER CELLS.

YOU CAN ACTUALLY USE AN UNSTAINED SLIDE AND GENERATE THINGS THAT ARE MORE ESOTERIC IN THEIR DEVELOPMENT AND START TO IDENTIFY WHAT TYPES OF CANCERS AND THE STAGES OF CANCER, ARE THEY GRADES OF CANCER AS YOU DO THAT.

AND WHAT IT DOES IS ALLOW YOU TO TRAIN ON THAT DATA.

>> Cathy: IN TERMS OF TRAINING ON DATA, YOU MUST HAVE MILLIONS OF SLIDES.

>> WE DO.

AND IN PATHOLOGY IN PARTICULAR, WE HAVE OVER 21 MILLION SLIDES RIGHT NOW.

SO THIS IS PART OF OUR ARCHIVE.

WE WENT BACK AND SCANNED OUR ARCHIVE.

THE MAYO BROTHERS ACTUALLY STARTED IT.

SO WE HAVE SLIDES IN OUR ARCHIVE, GLAD ARCHIVE, GL GLASS SLIDES WITH THEIR HANDWRITING ON IT.

THESE ARE SLIDES THAT THEY USED IN THEIR SURGERIES ALL THE AY TO THE PRESENT.

AND MAYO IS A FULLY DIGITIZED PRACTICE RIGHT NOW.

SO IN ADDITION TO OUR ARCHIVE, WE'RE CONTINUOUSLY ADDING TO IT THROUGH OUR CURRENT-DAY DIGITIZATION PROCESS.

>> Eric: DOES THIS HAVE AN IMPACT AT ALL ON HEALTHCARE COSTS FOR THE CONSUMER, FOR THE PATIENT?

>> WHAT IT WILL DO, EVENTUALLY, IF YOU LOOK AT A LOT OF THE WORK THAT'S DONE IN HEALTHCARE, PATHOLOGY IS A CRITICAL TEST.

IF YOU CAN GET TO THE RIGHT DIAGNOSIS AND TREATMENT FASTER, YOU ACTUALLY AVOID WHAT TAKES TYPICALLY A LONG TERM FOR THAT TO HAPPEN FORENSIC FORENFORMOST PATIENTS AND THAT L SAVE COSTS.

>> Cathy: ARE YOU RUNNING INTO ANY ISSUES WITH HALLUCINATIONS AND MAYBE, YOU KNOW, SOME GLITCHES?

BECAUSE YOU'RE STILL TRAINING THE ALGORITHMS, RIGHT?

>> WE'RE STILL BUILDING THEM.

WE'VE WORKED ON A FRONTIER OR FOUNDATION MODEL AND WE'VE DONE A LOT OF WORK TO MAKE SURE THAT THAT'S AN ACCURATE MODEL AND THE BEAUTY OF USING THAT MODEL IS YOU DON'T HAVE TO HAVE SO MUCH DATA TO BE ABLE TO RUN YOUR ALGORITHMS AND BUILD THEM.

BUT WE ALSO HAVE A LOT OF DIFFERENT PROCESSES THAT WE WILL USE TO ENSURE THAT THE ALGORITHMS ARE SAFE, EFFECTIVE, AND ACCURATE, AND MOST IMPORTANTLY, WE KEEP A HUMAN IN THE LOOP.

SO THE A.I.

IS GOING TO RECOMMEND, BUT THE A.I.

IS NOT GOING TO MAKE THE FINAL DECISION.

A HUMAN WILL AKE THE FINAL DECISION.

>> Eric: AND PEOPLE'S PRIVACY IS PROTECTED, THEIR DATA IS SAFE?

>> ABSOLUTELY.

YEAH, ONE OF THE KEY TENETS OF THIS IS WE PROTECT DATA, PATIENT DATA, VERY CAREFULLY.

>> Eric: AND YOU DO THAT OF COURSE ALREADY.

>> WE DO IT ALREADY AND WHAT YOU O IS YOU BUILD A.I.

MODELS, FOR MOST OF THEM, THEIR DATA IS NOT CONTAINED IN THAT.

IF WE DO BUILD GENERATIVE MODELS, THOSE EMBEDDING, SOME OF THAT CONTENT IS THERE BUT IT'S DE-IDENTIFIED AND WITH THAT, NONE OF THAT PATIENT-SPECIFIC INFORMATION WOULD EVER BE RELEASED AS PART OF THAT MODEL.

>> Cathy: IS THIS OPEN INFORMATION?

IF ANOTHER HOSPITAL SYSTEM, CAN THEY USE YOUR INFORMATION?

AND SHARE IT THROUGHOUT THE REST OF THE WORLD PERHAPS?

>> SO WE ARE WORKING TO DO THAT IN MANY DIFFERENT WAYS.

ONE WAY IS THROUGH MAYO CLINIC PLATFORM.

AND THAT IS A DATA NETWORK, IF YOU WILL.

AND THE WAY IT WORKS IS, EACH ORGANIZATION CAN CONTRIBUTE THEIR DATA, BUT THEIR DATA STAYS WITHIN THEIR ORGANIZATION.

AND YOU CAN STILL HAVE INSIGHTS FROM THAT.

THE OTHER WAY YOU CAN HINK ABOUT IT IS THIS FOUNDATION MODEL.

THE FOUNDATION MODEL CAN BE SHARED.

THE DATA ITSELF IS NOT BEING SHARED.

INSIGHTS FROM THE DATA ARE BEING SHARED.

IT'S REALLY ABOUT HOW DO WE UNLOCK THE INSIGHTS THAT CURRENTLY, AS IT RELATES TO PATHOLOGY ARE CONTAINED WITHIN GLASS SLIDES.

IT'S TOTALLY LOCKED UP.

YOU CAN'T USE A.I.

TO UNLOCK THE DATA AND THE INSIGHTS.

NOW YOU AN, AND THAT'S WHAT'S EXCITING TO US BECAUSE WE THINK IT'S ALMOST A ONCE IN A GENERATION OPPORTUNITY TO TRULY TRANSFORM HOW WE DO MEDICINE.

>> Eric: IN THE SCOPE OF ALL THE DEVELOPMENTS IN MEDICINE OVER HOW MANY YEARS, WHERE DOES THIS RANK AS FAR ASKER IMPORTANCE, REVOLUTIONARY ACTIVITY?

WHAT WOULD YOU SAY?

>> WE THINK IT'S A PIVOTAL MOMENT REALLY FOR HEALTHCARE.

IF YOU LOOK AT WHERE WE ARE TODAY AND WHERE WE'RE GOING, IT'S EXTRAORDINARILY MANUAL.

YOU TRY TO LEARN ABOUT PATIENTS.

OUR CLINICIANS ARE SYNTHESIZING PATIENTS IN THEIR HEAD TO MAKE SURE THEY CAN DO THE RIGHT THINGS.

THE GROWTH OF KNOWLEDGE IN MEDICINE IS EXPLODING.

AND THE CAPABILITIES ARE ALSO GOING ALONG WITH THAT, WITH NEW THERAPIES.

SO HOW DO YOU STAY AT THE EDGE IS THE CRITICAL THING.

WITH THIS, LARGE LANGUAGE MODELS, GENERATIVE MODELS, E CAN ACTUALLY GET HEIR FASTER IN TERMS OF UNDERSTANDING THAT PATIENT AND ACTUALLY UNDERSTAND THE LITERATURE AS PART OF THAT, SO WE CAN ACTUALLY DELIVER THE BEST CARE FOR PATIENTS.

>> Eric: I THINK NVIDIA IS THE MOST PROFITABLE COMPANY ON EARTH.

HOW HAVE THEY BEEN TO WORK WITH?

>> THEY'VE BEEN GREAT.

ABSOLUTELY.

THEY'VE BEEN A WONDERFUL PARTNER TO WORK WITH.

WE WORK WITH MANY DIFFERENT PARTNERS, BUT WHEN YOU THINK ABOUT WHAT WE'VE BEEN ABLE TO DO WITH NVIDIA, WE HAVE THE BLACK WELL CLUSTER NOW.

THIS IS REALLY NOW THEIR NEXT GENERATION OF COMPUTES.

THEY ALSO DO SERVICES AROUND THOSE CHIPS AND PRODUCTS AROUND THOSE CHIPS.

AND THE COMPUTE OW IS GOING TO ALLOW US TO GO VERY QUICKLY.

IT'S A SIGNIFICANT ADVANCE OVER THE PREVIOUS COMPUTE THAT WE HAD AVAILABLE TO US.

>> Eric: PEOPLE GOING TO LOSE JOBS AT MAYO?

>> WHAT IT IS DOING IS MAKING PEOPLE THAT ARE DOING THEIR JOBS BETTER.

AND WHAT'S KEY ABOUT HOW WE'RE APPROACHING THIS AT MAYO IS PEOPLE WHO KNOW THE PROBLEM THAT NEEDS TO BE SOLVED, ARE ALL ENABLED TO MAKE THE DEVELOPMENT.

SO WE BELIEVE WITH WHAT WE'RE DOING, IT REALLY JUST MAKES THOSE INDIVIDUALS BETTER AT THEIR ROLE.

>> Cathy: GENTLEMEN, THANK YOU FOR BEING HERE.