Transcript: Mental Health: There's an App For That | Jan 28, 2019

Steve sits in the studio. He's slim, clean-shaven, in his fifties, with short curly brown hair. He's wearing a gray suit, blue shirt, and checked blue tie.

A caption on screen reads "Mental Health: There's an app for that. @spaikin, @theagenda."

Steve says IF YOU USE A SMART
PHONE AND A COMPUTER, YOU
GENERATE A TONNE OF DATA.
COULD SOME OF IT BE CRUNCHED TO
HELP DIAGNOSE, PREDICT, AND EVEN
TREAT DEPRESSION AND OTHER
MENTAL ILLNESSES?
THESE ARE EARLY DAYS, BUT LET'S
FIND OUT ABOUT THE PROMISE OF
ARTIFICIAL INTELLIGENCE AND
DIGITAL TECHNOLOGY WITH:
ALISON DARCY, FOUNDER AND CEO OF
WOEBOT LABS AND AN ADJUNCT
PROFESSOR AT STANFORD
UNIVERSITY'S SCHOOL OF MEDICINE.
SHE JOINS US ON THE LINE FROM
SAN FRANCISCO, CALIFORNIA.

Alison is in her late thirties, with long straight brown hair. She's wearing a gray sweater.

Steve continues AND HERE IN OUR STUDIO:
DAVID GRATZER, ASSISTANT
PROFESSOR OF PSYCHIATRY AT THE
UNIVERSITY OF TORONTO AND AN
ATTENDING PSYCHIATRIST AT CAMH,
THE CENTRE FOR ADDICTION AND
MENTAL HEALTH...

David is in his fifties, clean-shaven, with short dark hair. He's wearing a gray suit, blue shirt, and black tie.

Steve continues AND ERIN KELLY, CEO OF ADVANCED
SYMBOLICS INC., WHO MEMORABLY
INTRODUCED US TO HER FRIEND
POLLY... AN ARTIFICIAL
INTELLIGENCE FORECASTER...
DURING THE LAST ONTARIO ELECTION.

Erin is in her forties, with long wavy blond hair. She's wearing a gray sweater and a black shirt.

Steve continues IT'S GREAT TO HAVE YOU TWO HERE
IN OUR STUDIO, AND ALISON,
THANKS FOR BEING THERE FOR US IN
CALIFORNIA.
I WANT TO START JUST WITH A
QUOTE FROM THE HARVARD BUSINESS
REVIEW TO GET US OFF TO THE
RACES HERE.
SHELDON IS BACK IN THE CHAIR.
IF YOU WOULDN'T MIND, SIR?

A quote appears on screen, under the title "Digital solutions." The quote reads "In a crisis that has become progressively dire over the past decade, digital solutions -many with artificial intelligence (AI) at their core- offer hope for reversing the decline in our mental wellness. New tools are being developed by tech companies and universities with potent diagnostic and treatment capabilities that can be used to serve large populations at reasonable costs."
Quoted from Parie Garg and Sam Glick, The Harvard Business Review. October 22, 2018.

Steve says LET'S GO FROM THERE.
DR. GRATZER, TO YOU FIRST.
HOW WOULD YOU CHARACTERIZE THE
PROMISE THAT ARTIFICIAL
INTELLIGENCE AND DIGITAL
SOLUTIONS HOLD FOR MENTAL HEALTH CARE?

The caption changes to "David Gratzer. University of Toronto."

David says I'M HOPEFUL.
I'M HOPEFUL WE'LL BE ABLE TO DO BETTER.
FOR EVERY TWO PEOPLE WE TREAT
FOR DEPRESSION IN CANADA, ONE
GETS ADEQUATE CARE.
PEOPLE FACE BARRIERS.
BARRIERS IN TERMS OF DIAGNOSIS.
BARRIERS IN TERMS OF TREATMENT.
AND THERE'S HOPE THAT TAPPING
INTO THE DATA THAT WE'RE
GENERATING WITH OUR PHONES, WITH
OUR COMPUTERS AND OUR
INTERACTIONS, MAYBE WE'LL BE
ABLE TO GET PEOPLE HELP FASTER
AND BETTER.

Steve says ALISON DARCY, WHAT'S
YOUR VIEW?

The caption changes to "Alison Darcy. Stanford University."

Alison says I AGREE WITH THAT.
FOR US AND FOR ME, IT'S REALLY
ABOUT ACCESS, YOU KNOW, ABOUT 5S
POPULATION RIGHT NOW DOES NOT
HAVE ADEQUATE ACCESS TO BASIC
HEALTH CARE, AND MENTAL HEALTH
CARE IS JUST NOT SOMETHING THAT
IS AVAILABLE FOR HUGE
POPULATIONS.
AND ABOUT TWO-THIRDS OF PEOPLE
WHO HAVE REAL SYMPTOMS AND COULD
REALLY USE A SENIOR CLINICIAN
COULD NEVER GET IN FRONT OF
THEM.
IT'S BEEN INCREASING ACCESS AND
LOWERING THE BARRIERS TO GETTING
ADEQUATE CARE.

Steve says ERIN KELLY, LAST
TIME YOU WERE HERE, YOU USED
POLLY, YOUR ARTIFICIAL
INTELLIGENCE, TO SORT OF HELP US
UNDERSTAND WHAT WAS GOING ON AS
IT RELATES TO POLITICS AND
ELECTION OUTCOMES AND THAT KIND
OF THING.
SO WHAT IS YOUR CONNECTION TO
MENTAL HEALTH CARE THROUGH
ARTIFICIAL INTELLIGENCE THAT YOU EMPLOY?

The caption changes to "Erin Kelly. Advanced Symbolics Inc."

Erin says WELL, WE WERE
APPROACHED BY THE UNIVERSITY OF
OTTAWA BACK IN 2016 BY A
RESEARCHER THERE WHO WAS LOOKING
TO SEE IF WE COULD USE A.I. TO
PREDICT WAVES OF SUICIDES IN THE
POPULATION, AND SO WE STARTED
THAT RESEARCH PROJECT WITH HER
TWO YEARS AGO, TWO AND A HALF
YEARS AGO NOW, AND SO I AM VERY
HOPEFUL BECAUSE WE'VE SEEN NOW,
AFTER DOING THIS RESEARCH, THAT
IN FACT WE CAN USE POLLY FOR
THIS PURPOSE.
BECAUSE POLLY IS THERE TO DETECT
AND PREDICT AND PRESCRIBE
CHANGES IN HUMAN BEHAVIOUR, SO
WHETHER THAT'S POLITICS OR
MENTAL HEALTH AND SUICIDE
IDEATION, IT'S THE SAME TO HER.

The caption changes to "Connect with us: Twitter: @theagenda; Facebook, agendaconnect@tvo.org, Instagram."

Steve says GOTCHA.
ALL RIGHT.
LET'S PICK UP WITH, IN ONE
STUDY, USING THE FACEBOOK DATA
OF 638 PARTICIPANTS, APPARENTLY
THEY WERE ABLE TO PREDICT FUTURE
DEPRESSION DIAGNOSES AS MUCH AS
THREE MONTHS BEFORE IT APPEARED
IN THEIR MEDICAL RECORDS AS A
DIAGNOSED CONDITION.
AGAIN, DAVID, HOW CAN SOCIAL
MEDIA BE USED TO PREDICT
DEPRESSION?

The caption changes to "David Gratzer. CAMH."

David says THIS IS
EARLY DATA SO WE'RE GOING TO
TAKE IT WITH A GRAIN OF SALT.
BUT, LOOK, YOU AND I INTERACT
WITH THIS WORLD SO DIFFERENTLY
THAN WE WOULD HAVE 10 YEARS AGO,
20 YEARS AGO.
AND THAT GENERATES DATA.
SO MY POSTS ON TWITTER... PARDON
ME, MY POSTS ON FACEBOOK, MY
TWEETS ON TWITTER, MAYBE THAT'S
SHOWING EARLY SIGNS OF AN
EMERGING ILLNESS AND MAYBE WE
CAN GATHER THAT INFORMATION,
INFORMATION BY THE WAY THAT
WOULD BE PRETTY INVISIBLE TO
PROVIDERS LIKE MYSELF, SO OUR
DOCTORS AND NURSES WORK HARD,
BUT WE DON'T GO THROUGH EVERY
SINGLE TWEET OF EVERY PATIENT.
MAYBE WE COULD FIND THAT
INFORMATION AND HARNESS IT WITH
A.I. AND COME BACK WITH A
DIAGNOSIS LONG BEFORE A PERSON
HAD EVEN...

Steve says LET ME DO A QUICK
FOLLOW-UP.
GPS.
CAN YOU USE GPS FOR THE SAME WAY?

David says I HAD A PATIENT A FEW YEARS
AGO AND A SIGN THAT HE WAS
GETTING ILL WAS THAT HIS
EVENINGS WALKS CHANGED.
HE HAD BIPOLAR AND HE WAS
ENTERING INTO A MANIC EPISODE.
HE WALKED LATE AT NIGHT, WALKED
INTO UNFAMILIAR AREAS.
SO HIS DAUGHTER, WHO IS REALLY
ON THE BALL, WOULD GIVE ME A
PHONE CALL.
I WOULD HAVE NO OTHER WAY OF
KNOWING.
BUT HIS SMARTPHONE KNEW THERE
WAS A CHANGE IN HIS PATTERN
BECAUSE HIS SMARTPHONE HAS A
GPS.
MAYBE IN THE FUTURE WE CAN USE
THAT IN OUR EXPERIMENTS LIKE IN
MICHIGAN WITH PRIORI, WHICH IS
AN APP THAT LOOKS AT YOUR GPS
AND LOOKS FOR SIGNS OF EMERGING
ILLNESS.
IT'S EXCITING.
BUT EARLY.

Steve says EARLY DAYS.
ERIN, WHAT SORT OF DATA OR
LINGUISTIC CLUES IS POLLY
LOOKING AT TO HELP IN
DISCOVERING MORE ON MENTAL HEALTH?

The caption changes to "Detecting depression."

Erin says THE REASON
THERE'S SO MUCH PROMISE FOR
MENTAL HEALTH IN SOCIAL DATA IS
BECAUSE THERE'S SO MUCH OF IT,
AND STATISTICS WORKS BEST WHEN
YOU HAVE LOTS AND LOTS OF DATA.
IF YOU'RE LOOKING AT ONE
PATIENT, TEN PATIENTS, IT'S HARD
TO FIND PATTERNS AND TO FIND
WAYS OF DETECTING THESE THINGS.
BUT ONLINE, WE HAVE TRILLIONS OF
DATA POINTS.
THAT'S THE FIRST THING I WANT TO
SAY.
WHAT POLLY DOES, SHE'S LOOKING
NOT JUST AT THE INDIVIDUAL.
HERE IS THE OTHER GREAT THING
ABOUT SOCIAL MEDIA.
WE CAN GO BACK AND LOOK AT THE
HISTORY OF WHAT PEOPLE HAVE
SAID.
SO IF WE START TO DETECT IN 2019
THAT SOMEBODY IS HAVING MENTAL
HEALTH PROBLEMS, MAYBE BECAUSE
THEIR PATTERNS HAVE CHANGED.
NOW, PATTERNS CHANGING COULD BE
A SIGN.
BUT IT COULD ALSO BE... I MEAN,
ANYBODY WHO HAS HAD AN
ADOLESCENT KNOWS THEIR SLEEPING
PATTERNS CHANGE.
THAT COULD BE A SIGN OF
PROBLEMS, IT COULD ALSO BE THE
SIGN OF A HEALTHY ADOLESCENT.
TO GO BACK AND SAY ARE THERE
MORE COMORBID SIGNS?
DO WE SEE SIGNS OF SUBSTANCE
ABUSE?
IS THERE A HISTORY OF CHILD
ABUSE IN THIS PERSON'S HISTORY?
HAS A PARENT BEEN LAID OFF WORK
OR, IN THE CASE OF AN OLDER
PERSON, HAS THERE BEEN A FACTORY
CLOSING DOWN?
THESE TYPES OF THINGS, TOGETHER
WITH THE SIGNS THAT WE SEE AT
THE INDIVIDUAL LEVEL, BECAUSE IF
WE WERE JUST TO CHASE AROUND
INDIVIDUALS, WE'D HAVE A HIGH
FALSE POSITIVE RATE.
BUT WHEN WE CAN GO BACK AND SEE
THAT HISTORY AS WELL, THAT'S
WHAT GIVES US THE ABILITY TO
REALLY PINPOINT THINGS IN A WAY
THAT WE WEREN'T ABLE TO BEFORE.

Steve says ALISON, HELP US WITH
THIS: I PRESUME MANY PEOPLE WHO
SUFFER FROM DEPRESSION KNOW IT.
THEY ALREADY KNOW THEY SUFFER
FROM DEPRESSION.
THEY DON'T NEED AN APP TO TELL
THEM THAT.
SO IS THERE AN IMPORTANT LIMIT
TO HOW USEFUL THIS TECHNOLOGY
REALLY CAN BE?

The caption changes to "Alison Darcy. Woebot Labs."

Alison says WELL, I THINK
IT'S USEFUL TO BEAR IN MIND THAT
ACTUALLY IN A META ANALYSIS THAT
WAS CONDUCTED, ONE OF THE MOST
PREDICTIVE ITEM OF DEPRESSION
TURNS OUT TO BE THE QUESTION:
HOW DEPRESSED HAVE YOU BEEN IN
THE LAST TWO WEEKS?
SO I DO THINK IT IS WHERE YOU
CAN ENGAGE SOMEBODY'S OWN
SELF-REFLECTION, AND IT'S
WORTHWHILE DOING THAT BECAUSE
IT'S ABOUT THE PERSON REALIZING
THEMSELVES.
HOWEVER, NOT EVERYBODY HAS THAT
KIND OF INSIGHT, AND I THINK THE
PROMISE OF WHAT WE ARE SPEAKING
ABOUT TODAY IS THAT ACROSS
PROBLEM AREAS, EARLY DETECTION
IS ONE OF THE MOST USEFUL
METHOD, YOU KNOW, AND IS
PREDICTIVE OF PEOPLE GETTING
WELL EARLIER.
SO THERE IS SOME VALUE IN EARLY
DETECTION.
ABSOLUTELY, PROCEEDING WITH
CAUTION, OF COURSE, IS PRUDENT.

Steve says DAVID, LET'S MAKE
THE COMPARISON, AS YOU MIGHT TO
A CARDIOLOGIST, WHO HAS ALL
SORTS OF TESTS AND DATA THAT HE
OR SHE CAN PUT A PATIENT THROUGH
TO UNDERSTAND HOW TO APPROACH
THINGS.
I GUESS UP UNTIL NOW, YOU
HAVEN'T HAD THAT KIND OF
COROLLARY DATA, BUT NOW MAYBE YOU DO?

David says PSYCHIATRY
IS A WORLD DIFFERENT THAN WHEN I
ENTERED MED SCHOOL 23 YEARS AGO.
WE USE DIFFERENT MEDICATIONS.
I THINK WE USE THEM BETTER.
WE'RE MUCH MORE FOCUSED ON
SHORT-TERM EFFECTIVE THERAPIES.
BUT WHEN I MEET WITH A NEW
PATIENT, WHEN HE OR SHE SITS IN
MY OFFICE, THE INTERVIEW IS VERY
SIMILAR.
I ASK QUESTIONS AND HE OR SHE
ANSWERS.
LET'S CONTRAST IT WITH HOW A
CARDIOLOGIST MIGHT WORK.
GOD FORBID YOU HAVE CHEST PAIN,
YOU GO TO THE E.D.

Steve says E.D.?

David says EMERGENCY DEPARTMENT.
US DOCTORS HAVE TO SPEAK IN VERY
COMPLICATED WAYS.

Steve says IN TV WE ALL CALL IT
THE E.R. ANYWAY...

David says YOU GO TO THE E.R. AND THEY
DO BLOODWORK AND THEY DO AN EKG,
THEY DECIDE YOU SHOULD SEE A
CARDIOLOGIST, DO STRESS TESTS
AND IMAGING.
BY THE TIME YOU ARRIVE IN YOUR
CARDIOLOGIST'S OFFICE, THERE'S A
WEALTH OF DATA AVAILABLE.
THAT DATA TO THIS POINT HASN'T
BEEN AVAILABLE IN MENTAL HEALTH.
SO WE DON'T KNOW ABOUT HOW YOUR
WALKING PATTERNS HAVE CHANGED,
HOW YOUR SPEECH PATTERNS HAVE
CHANGED, HOW YOUR SLEEP PATTERNS
HAVE CHANGED.
ON THE DIAGNOSTIC END I THINK
THERE'S A LOT OF OPPORTUNITY TO
GATHER DATA.
AND ALSO IN TREATMENT.
COULD WE THEN TAKE THAT
INFORMATION AND DEPLOY
TREATMENTS IN REAL TIME, LONG
BEFORE YOU THOUGHT MAYBE YOU
EVEN NEEDED TREATMENT.

Steve says FOLLOW-UP.
SPEECH PATTERNS CHANGING.
WHAT DOES THAT MEAN?

David says SOMEBODY COMING INTO A MANIC
EPISODE IN TERMS OF THEIR
BIPOLAR EFFECTIVE DISORDER, THEY
MIGHT SPEAK MORE QUICKLY.
SOMEONE WITH A DEPRESSIVE
EPISODE MIGHT SPEAK MORE SLOWLY
OR NOT REACHING OUT TO FRIENDS
AND FAMILY THE SAME WAY THEY
USED TO.
AGAIN, IF YOU'RE A CLINICIAN AND
CAN REACH FAMILY MEMBERS, YOU
COULD GET THAT SORT OF
INFORMATION.
BUT THAT INFORMATION ALREADY
EXISTS.
IT'S JUST THAT WE HAVEN'T UNTIL
THIS POINT BEEN ABLE TO TAP
SMARTPHONES AND OTHER
TECHNOLOGIES.

Steve says ERIN, CAN YOU
IMAGINE A TIME IN THE
NOT-TOO-DISTANT FUTURE WHERE YOU
ARE ABLE TO CRUNCH THIS DATA AND
ACTUALLY PREDICT WHEN SOMEBODY
MIGHT BE LOST TO SUICIDE?

Erin says YES.
AND SO WHAT WE'RE DOING RIGHT
NOW IN OUR RESEARCH IS... SEE,
WHAT HAPPENS TODAY IS WE HAVE,
SAY, A WAVE OF SUICIDES IN
NORTHERN QUEBEC.
THE GOVERNMENT THEN SENDS
RESOURCES INTO NORTHERN QUEBEC,
AND WHAT HAPPENS, THEY HAVE TO
TAKE THOSE RESOURCES FROM
SOMEWHERE ELSE.
SO THEN WE HAVE... YOU TAKE
THOSE RESOURCES FROM, SAY,
SASKATCHEWAN, AND THEN WE HAVE A
WAVE OF SUICIDES OVER THERE.
WHAT WE WANT TO DO IS GET THE
RESOURCES TO THE PLACE THEY NEED
TO BE BEFORE THE SUICIDES
HAPPEN.
SO THAT'S THE WORK WE'RE DOING
NOW.
CAN WE PREDICT AND HOW FAR IN
ADVANCE CAN WE PREDICT WHEN
THERE WILL BE... AGAIN, IT'S A
WAVE OF SUICIDES.
POPULATION LEVEL IN COMMUNITIES
THAT ARE AT RISK.
CAN WE DETECT THAT AND CAN WE
GET THE RESOURCES THERE BEFORE
THE SUICIDES HAPPEN SO WE
PREVENT THEM?

The caption changes to "On the robot's couch."

Steve says OKAY.
ALISON, WE'RE GOING TO PUT THE
SPOTLIGHT ON YOU FOR A BIT HERE
BECAUSE, WE ALL KNOW WHAT A
ROBOT IS, BUT YOU'VE DEVELOPED
SOMETHING CALLED A WOEBOT.
WE HAVE A CLIP WE WANT TO SHOW
TO INTRODUCE EVERYBODY TO A
WOEBOT, W-O-E-B-O-T.
STAND BY, EVERYBODY.
SHELDON, IF YOU WOULD?
LET'S ROLL IT.

A clip plays on screen. In the clip, a chat screen appears. A user with a robot avatar types "How are you feeling today?"

A male voice narrates
EVERYBODY HE ASKS HOW YOUR
DAY IS GOING, HOW YOU'RE
FEELING, AND WHAT YOU'RE UP TO.
HE BUILDS AN EMOTIONAL MODEL OF
YOU OVER TIME AND COULD HELP SEE
PATTERNS IN YOUR MOOD.
AS HE LEARNS ABOUT YOU, HE'LL
TEACH YOU THINGS.
LIKE USEFUL STRATEGIES AND
PRACTICAL TOOLS THAT HAVE BEEN
SHOWN TO WORK.

The clip ends.

Steve says OKAY, ALISON, WHERE
DID THE IDEA OF A WOEBOT COME FROM?

Alison says ACTUALLY, IT
JUST CAME FROM... I WAS AT
STANFORD FOR MANY YEARS, FOR
ALMOST TEN YEARS, AND WE WERE
DESIGNING TREATMENTS AND, YOU
KNOW, EVALUATING THEM AND
TREATMENT TRIALS AND
CONVENTIONAL KIND OF RESEARCH.
AND JUST A REALIZATION THAT
ACTUALLY THE MENTAL HEALTH
PROBLEM ALL AROUND US WAS
GETTING OUT OF CONTROL VERY
QUICKLY, AND THE NEED TO MAKE
SOMETHING THAT WAS EFFICACIOUS
BUT ALSO HIGHLY ENGAGING IN THE
POPULATION, TO ENCOURAGE PEOPLE
TO REALLY ENGAGE WITH MENTAL
HEALTH PRACTICES EVERY SINGLE
DAY.
SO WOEBOT CAME FROM THAT
ENDEAVOUR.

Steve says HOW EFFECTIVE IS IT
AT DOING WHAT IT PURPORTS TO DO?

Alison says WELL, WE ACTUALLY CONDUCTED A RANDOMIZED CONTROL TRIAL AT
STANFORD AND DEMONSTRATED THAT
WOEBOT COULD AND SIGNIFICANTLY
AND MEANINGFULLY REDUCE SYMPTOMS
OF DEPRESSION IN JUST TWO WEEKS,
ACTUALLY, WHICH IS FAIRLY QUICK.
WE'VE ALSO REPLICATED THAT SINCE
WITH A LARGER SAMPLE SIZE OF 400
AND SHOWED WE CAN ACHIEVE ABOUT
A 28 PERCENT REDUCTION IN
SYMPTOMS OF DEPRESSION AND A
38 PERCENT REDUCTION IN SYMPTOMS
OF ANXIETY OVER THE COURSE OF 4
WEEKS.
AND SO IT'S FAIRLY EFFECTIVE,
ACTUALLY.

Steve says DO YOU KNOW WHY IT WORKS?

The caption changes to "Alison Darcy, @alisonmdarcy."

Alison says I THINK IT'S BECAUSE IT'S
INTEGRATED INTO EVERYBODY'S
DAILY LIVES, AND THAT IS
SOMETHING THAT HITHERTO HAS NOT
REALLY BEEN POSSIBLE.
AND IT'S ALSO SOMETHING THAT YOU
CAN REACH OUT TO IN A MOMENT OF
DISTRESS OR A MOMENT OF NEED.
NOT EXTREME OR ACUTE DISTRESS,
OF COURSE.
BUT, YOU KNOW, YOU'VE JUST HAD A
FIGHT, YOU KNOW, WITH YOUR
HUSBAND OR YOU'RE ABOUT TO GIVE
A PRESENTATION AT WORK AND
FEELING REALLY ANXIOUS ABOUT IT,
YOU CAN TURN TO WOEBOT, AND THEN
WOEBOT CAN TEACH THESE
EVIDENCE-BASED TECHNIQUES BASED
ON COGNITIVE BEHAVIOURAL
THERAPY, BUT IN THE ACTUAL
MOMENT THAT IT'S USEFUL, AND WE
SUSPECT THAT SOME OF THE GREAT
OUTCOMES WE'RE SEEING IS BECAUSE
OF THAT KIND OF IN-THE-MOMENT IMPACT.

Steve says DAVID, I WOULD HAVE
THOUGHT IF THERE'S ANYTHING IN
THIS WORLD YOU CAN'T REPLACE THE
HUMAN TOUCH WITH, WITH A
COMPUTER OR CHATBOT OR WHATEVER
YOU WANT TO SAY, THIS WOULD BE IT.
SO I'M INTERESTED IN YOUR VIEW
ON WHETHER YOU THINK THIS IS THE
REAL DEAL?

The caption changes to "David Gratzer, @DavidGratzer."

David says I THINK
THERE'S SOMETHING HERE.
I THINK IT'S EARLY DAYS... THERE
IS SOME DATA.
WHAT WE'RE TAPPING IS THAT THERE
ARE EVIDENCE-BASED TREATMENTS
LIKE COGNITIVE BEHAVIOURAL
THERAPY WHICH LOOKS AT HOW YOUR
THOUGHTS INFLUENCE YOUR MOOD,
INFLUENCE YOUR BEHAVIOUR, BUT
WE'RE DOING IT WITHOUT A
THERAPIST.
SO I THINK FOR SOME PEOPLE, THAT
WILL BE APPROPRIATE AND
REASONABLE.
A LOT OF US DO THINGS THROUGH
THE PHONE.
WE DON'T NECESSARILY NEED THAT
HUMAN TOUCH.
BUT MAYBE SOME PEOPLE DO.
SO I DON'T THINK IT'S A PANACEA.
I ALSO THINK CERTAIN POPULATIONS
MIGHT WORK BETTER WITH THIS.
FOR INSTANCE, SOME PEOPLE WHO
ARE AUTISM SPECTRUM AS AN
EXAMPLE MIGHT FEEL BETTER ABOUT
THE FACT THERE'S NO HUMAN
CONNECTION.
SO, YOU KNOW, WHEN I LOOK AT
THESE EARLY DAYS AND THE
EXPERIMENT GOING ON, I THINK
THESE ARE DIFFERENT TOOLS BEING
PUT IN THE TOOL KIT.
AS EVERYONE KNOWS, YOU DON'T
NECESSARILY USE EVERY TOOL IN
THE TOOL KIT TO SOLVE THE PROBLEM.

Steve says BUT THEY'RE GOOD TO HAVE.

David says THEY'RE GOOD TO HAVE.

Steve says ALISON, IN DESIGNING
WOEBOT, DID YOU WANT USERS TO
BELIEVE THAT THE WOEBOT WAS
SENTIENT AND HUMAN-LIKE, OR WAS
THERE SORT OF NEVER ANY QUESTION
THAT, HEY, THIS IS A ROBOT, BUT
IT CAN STILL BE VERY HELPFUL FOR YOU?

Alison says ABSOLUTELY.
WE COMPLETELY AGREE THAT THERE'S
NO SUBSTITUTION FOR HUMAN
CONNECTION, AND SO WOEBOT IS
INTENTIONALLY A ROBOT CHARACTER
AND HE HAS A ROBOT BACKSTORY
WITH ROBOT FRIENDS AND WHAT HAVE
YOU.
IT'S A FICTION THAT PEOPLE SORT
OF BUY INTO.
HE HAS A CHARACTER WITH HIS OWN
VALUES, SO THAT PEOPLE SORT OF
UNDERSTAND THE APPROACH HE'S
GOING TO BE TAKING TO CERTAIN
PROBLEMS.
WE FELT THAT WAS REALLY
IMPORTANT, AND MOST CHATBOT
DEVELOPERS AT THIS POINT DON'T
PRETEND TO BE HUMAN.
I THINK IF YOU LOOK AT SORT OF
THE REALLY UBIQUITOUS ONES LIKE
AMAZON ALEXA AND GOOGLE HOME,
YOU WILL FIND THEY ARE JUST
SOFTWARE AND WE FEEL THAT'S THE
MOST TRANSPARENT AND HONEST WAY
TO PROCEED.

Steve says IF, HOWEVER,
ALISON... I WAS GOING TO SAY,
MAYBE ONE OF THE BIGGEST SIGNS
OF THE TIMES IN WHICH WE LIVE IS
THAT THE UNITED KINGDOM
GOVERNMENT HAS CREATED A
MINISTRY RESPONSIBLE FOR
LONELINESS.

Alison says CORRECT.

Steve says OH, MY GOODNESS.
CAN YOU IMAGINE THAT EVEN 10
YEARS AGO HAVING HAPPENED?
SO LONELINESS AND ISOLATION IS
CLEARLY BEING RECOGNIZED AS A
POTENTIAL PUBLIC HEALTH ISSUE.
HOW DEEP IS YOUR CONFIDENCE THAT
CHATBOTS CAN HELP WITH THIS AT
THE END OF THE DAY?

Alison says I THINK IT'S TRUE THAT THERE
IS NOT ONE APPROACH THAT WORKS
FOR EVERYBODY, AND I THINK IT'S
NOT NECESSARILY, YOU KNOW, THE
CHATBOT ITSELF THAT CONVEYS, YOU
KNOW, USEFUL THERAPEUTIC VALUE.
AND THE FACT THAT CBT AND HOW
WELL THE CHATBOT IS DESIGNED IS
REALLY CRUCIAL.
SO WE HAVE ALMOST 20 YEARS NOW
OF DATA THAT SHOWS THAT YOU CAN
DELIVER CBT WITHOUT A THERAPIST.
IN MANY CASES, WITHOUT
COMPROMISING OUTCOMES.
SO IN OTHER WORDS THE PATIENT
WILL DO JUST AS WELL WITH A
SELF-DIRECTED VERSION OF CBT, AS
WELL AS THEY WILL WITH A PERSON
DELIVERED ONE.
THE INTERESTING THING ABOUT A
CHATBOT THAT IS, YOU KNOW, OUT
AMONGST THE POPULATION IS THAT
WE ARE DISCOVERING THE KINDS OF
PROBLEMS THAT PEOPLE ARE DEALING
WITH IN THEIR EVERYDAY LIVES,
AND ONE OF THE THINGS THAT WAS
SO SURPRISING TO US WAS HOW MANY
PEOPLE DO ACTUALLY REPORT
LONELINESS, EVEN WHEN THEY'RE IN
THE CONTEXT OF A SOCIAL
SITUATION.
AND I DO THINK CONVERSATIONAL
AGENTS, THE BEAUTY ABOUT THEM IS
JUST THAT THERE'S NO LEARNING
INVOLVED, THERE'S NO APP THAT
YOU HAVE TO SORT OF LEARN
THROUGH.
IT TURNS OUT WHEN PEOPLE ARE
FEELING BADLY, THEY TALK ABOUT
THEIR PROBLEMS, THEY DON'T CLICK
THROUGH THEM OR SWIPE THROUGH
THEM.
SO IT'S A VERY NATURAL WAY TO
INTERFACE WITH TECHNOLOGY.
BUT THE ACTUAL PRINCIPLES BEHIND
IT, THE DESIGN PRINCIPLES,
REMAIN THE SAME, AND THE
EFFICACIOUS VALUE OF THE
TECHNIQUES THEMSELVES, THEY
STILL HAVE TO BE REALLY
WELL-DESIGNED AND DELIVERED.

Steve says NOW, IN THE MIDST OF
THAT ANSWER, YOU COULD NOT SEE
THIS, ALISON, DAVID GRATZER WAS
KIND OF... SHELDON, PUT THE
CAMERA ON ME FOR A SECOND HERE.
HE WAS GOING LIKE THIS.

He makes a face gesture and says
WHICH I INTERPRETED TO
MEAN, "YEAH, I KIND OF SEE WHAT
SHE'S SAYING BUT I'M NOT
100 PERCENT WITH HER."
AM I RIGHT ON THAT?

David says I THINK
YOU'VE INTERPRETED MY BODY
LANGUAGE BETTER THAN PERHAPS AN
APP.
AGAIN, I THINK THESE ARE EARLY
DAYS AND THERE'S A LOT OF
POTENTIAL HERE.
IT'S IMPORTANT IN A CANADIAN
CONTEXT TO NOTICE HOW FAR WE
FALL SHORT IN OKAY.
FOR EVERY 100 PEOPLE IN
CANADA... IT'S A STATISTIC FROM
THE CANADIAN JOURNAL OF
PSYCHIATRY.
IF WE CAN MAKE THERAPIES MORE
ACCESSIBLE, PARTICULARLY GOING
PAST THAT BRICKS AND MORTAR
APPROACH, I'M ONSIDE.
CAN WE SIMPLY DO AWAY WITH
THERAPISTS AND GET THE SAME
RESULT?
YES AND NO.
I THINK FOR SOME PEOPLE YES.
I ACTUALLY DID THE REVIEW PAPER
FOR THE CANADIAN MEDICAL
ASSOCIATION JOURNAL THREE YEARS
AGO, AND WE FIND IN FACT A VERY
HIGH DROP-OUT RATE WHEN WE JUST
TELL PEOPLE TO GO TO A WEBSITE.
AGAIN, I SEE THIS AS A TOOL IN
THE TOOL KIT.
I THINK THIS IS A WELL HIGH
PRESSURE DESIGNED APP THAT...
WELL-DESIGNED APP THEY'VE
PRODUCED.
THERE ARE OTHER CONVERSATIONAL
AGENTS THAT ARE WELL-DESIGNED.
I DON'T THINK THIS IS A SOLUTION
FOR THERAPY NEEDS ACROSS THIS
GREAT COUNTRY.

Steve says GOTCHA.
I THINK IT'S TIME FOR ANOTHER
CLIP HERE.
TECH SOCIOLOGIST, NEW YORK TIMES
OP-ED COLUMNIST ON THE POTENTIAL
DOWNSIDE TO PREDICTING HOW WE
FEEL.
SHELDON, THE CLIP, PLEASE.

A clip plays on screen. In the clip, a woman in her forties with long wavy hair stands on a stage and addresses an unseen audience.

She says WHAT IF THE SYSTEM THAT WE DO
NOT UNDERSTAND WAS PICKING UP
THAT IT'S EASIER TO SELL VEGAS
TICKETS TO PEOPLE WHO ARE
BIPOLAR AND ABOUT TO ENTER THE
MANIC PHASE?
SUCH PEOPLE TEND TO BECOME
OVERSPENDERS, COMPULSIVE
GAMBLERS.
THEY COULD DO THIS, AND YOU'D
HAVE NO CLUE THAT'S WHAT THEY
WERE PICKING UP ON.
I GAVE THIS EXAMPLE TO A BUNCH
OF COMPUTER SCIENTISTS.
AND AFTERWARDS, ONE OF THEM CAME
UP TO ME.
HE WAS TROUBLED.
HE SAID THAT'S WHY I COULDN'T
PUBLISH IT.
AND I WAS LIKE, COULDN'T PUBLISH
WHAT?
HE HAD TRIED TO SEE WHETHER YOU
CAN INDEED FIGURE OUT THE ONSET
OF MANIA FROM SOCIAL MEDIA POSTS
BEFORE CLINICAL SYMPTOMS, AND IT
HAD WORKED, AND IT HAD WORKED
VERY WELL.

Steve says OKAY, ERIN.
IS THERE POTENTIAL DANGER TO
HAVING OUR MENTAL STATES
PREDICTED BY ARTIFICIAL
INTELLIGENCE?

The caption changes to "Erin Kelly, @ErinOttawa."
Then, it changes again to "Hacking our vulnerability?"

Erin says I THINK THERE'S
ALWAYS POTENTIAL DANGER AND A
POTENTIAL UPSIDE TO ANYTHING
THAT YOU USE, BUT I THINK THE
UPSIDE IN THIS CASE OFFSETS THE
NEGATIVE SIDE, AND YOU CAN
ALWAYS DEAL WITH THE NEGATIVE
SIDE THROUGH POLICIES AND THINGS
LIKE THAT.
I THINK THERE'S JUST SO MUCH
TREMENDOUS UPSIDE TO THE WORK
WE'RE DOING IN REACHING OUT TO
PEOPLE WHO ARE SUFFERING FROM
MENTAL HEALTH PROBLEMS.
I THINK THE CASES THAT SHE WAS
TALKING ABOUT ARE GOING TO BE
REMOTE BUT YES, OF COURSE, THEY
WERE ALWAYS THERE.
PEOPLE CAN TAKE ADVANTAGE OF
VULNERABLE PEOPLE.
SO THE BEST WAY FOR ME TO GUARD
AGAINST THAT IS TO BE EDUCATED,
TO UNDERSTAND THAT THERE ARE
PEOPLE DOING THAT OUT THERE, BUT
I THINK THE POTENTIAL UPSIDE IS
JUST SO STRONG FROM WHAT WE'VE
SEEN THAT I WOULDN'T THROW OUT
THE BABY WITH THE BATHWATER,
THAT'S WHAT I WOULD SAY.

Steve says ALISON, WHAT'S YOUR
VIEW ON THE UPSIDE VERSUS THE
DOWNSIDE HERE?

Alison says I AGREE.
I THINK WE HAVE TO... THE
DIFFERENCE HERE IS BETWEEN
PASSIVE DETECTION AND ACTIVE
DETECTION OF MOOD STATES.
ACTIVE DETECTION IS, FOR
EXAMPLE, WHAT WE DO, WHERE WE
SIMPLY ASK PEOPLE: HOW DO YOU
FEEL?
AND WOEBOT IS VERY CLEAR ABOUT
THIS.
EVEN IF WOEBOT COULD DETECT YOUR
EMOTIONAL STATE, HE WON'T.
HE'LL ASK YOU, BECAUSE THAT
MOMENT OF SELF-REFLECTION IS THE
KEY FACTOR.
PASSIVE DETECTION, ON THE OTHER
HAND, IS WHERE PEOPLE ARE
DETECTING MOOD STATES, YOU KNOW,
WITHOUT THE PERSON ACTIVELY
PROVIDING ANY INPUT, AND JUST,
YOU KNOW, PROCESSING SOMEBODY'S
DATA.
I THINK THAT IS AN AREA IN WHICH
WE NEED TO BE REALLY CAUTIOUS
AND THOUGHTFUL, BUT I ERR ON THE
SIDE OF TRANSPARENCY.
YOU KNOW, IF PEOPLE KNOW ABOUT
IT AND THEY ACTIVELY OPT IN,
THEN I THINK THAT THERE'S... YOU
KNOW, WE MITIGATE A LOT OF THAT
DANGER.
AND THE DANGER, OF COURSE, WHEN
YOU'RE PASSIVELY DETECTING MOOD
STATES, IS THAT YOU ARE
DETECTING POTENTIALLY A PERSON
IN, YOU KNOW, A STATE OF
EMOTIONAL VULNERABILITY, WHICH
CAN THEN BE EXPLOITED.
YOU KNOW, THIS IS SORT OF AN
ADVERTISER'S DREAM.
SO I THINK, YEAH, WE SHOULD
PROCEED WITH CAUTION.
WE SHOULD HAVE THE CONVERSATION.
AND WE SHOULD JUST BE LETTING
PEOPLE KNOW AND MAKING SURE,
THROUGH REGULATION, THAT PEOPLE
MUST OPT IN, MUCH LIKE EUROPE
HAS DONE WITH THE GDPR
REGULATION, TO ALLOW THEIR DATA
TO BE USED AND PROCESSED IN
EVERY SPECIFIC WAY.

The caption changes to "Connect with us: Twitter: @theagenda; Facebook, agendaconnect@tvo.org, Instagram."

Steve says DAVID, WHERE ARE YOU
ON THAT?

David says I THINK WE
NEED TO BE VERY CAREFULLY.
I MEAN, THERE ARE TREMENDOUS
ETHICAL ISSUES.
WE WANT PEOPLE TO GET HELP IN A
TIMELY WAY.
WE DON'T WANT PEOPLE TO BE
EXPLOITED.
AND WE DON'T WANT PEOPLE TO BE
EXPOSED.
SO, FOR INSTANCE, PEOPLE ARE
GOING THROUGH SOME TROUBLE AND
THEY'RE PUTTING INFORMATION INTO
AN APP THAT SHOULDN'T BE
AVAILABLE TO THEIR EMPLOYER,
THAT SHOULDN'T BE AVAILABLE TO
AN INSURANCE COMPANY.
WE REALLY HAVE THAT PROBLEM.
IF YOU LOOK AT DEPRESSION APPS
TODAY, THERE ARE 1200 APPS IN
THE APP STORE IF YOU TYPE IN
DEPRESSION.
MORE THAN THREE-QUARTERS FAIL A
BASIC STANDARD LIKE A CLEAR
PRIVACY POLICY.
OKAY?

Steve says SORRY, YOU SAID
THREE-QUARTERS?

David says THREE IN EVERY FOUR FALL
SHORT.
OKAY?
SO ARE THERE APPS THAT ARE
EVIDENCE HIGH PRESSURE BASED
THAT ARE PROVIDING TIMELY
INFORMATION?
ABSOLUTELY.
ARE THERE APPS THAT TOUCH ON,
LIKE HER APP, COGNITIVE
BEHAVIOURAL THERAPY THAT IS
EVIDENCE-BASED?
ABSOLUTELY.
ARE THERE APPS FILLED WITH SNAKE
OILS AND BROMIDES?
YUP.
WE HAVE TO BE CLEAR THAT OUR
AGENDA HERE AND I SPEAK AS A
PROVIDER, AS A DOCTOR, IS GOING
TO BE DIFFERENT THAN WALL
STREET'S AGENDA OR BAY STREET'S
AGENDA.
HOW MANY CLICKS THEY GET, HOW
MANY MINUTES THEY SPENT ON AN
APP.
I DON'T CARE ABOUT THAT.
THAT'S WHY YOU SEE THE MOST
SUCCESSFUL MINDFULNESS APPS
DON'T NECESSARILY TEACH
MINDFULNESS THAT WELL.
SO HOLD ON.
LOTS OF OPPORTUNITIES HERE.
BUT LOTS OF PROBLEMS HERE AS WELL.

Steve says ERIN?

Erin says I'D LIKE TO SAY
WE'VE ALREADY GOT A REALLY GOOD
METHODOLOGY I THINK IN CANADA
FOR HOW TO COLLECT HEALTH
INFORMATION.
STATISTICS CANADA SAYS WITHOUT
IDENTIFYING INDIVIDUALS.
SO WE WOULDN'T WANT TO STOP
COLLECTING INFORMATION FROM THE
CENSUS.
WE TRIED THAT AND IT DIDN'T GO
OVER VERY WELL.
YOU NEED THAT INFORMATION FOR
POPULATION HEALTH.
SO WE HAVE TO BE ABLE TO COLLECT
IT IN A WAY THAT IS NOT
PERSONALLY IDENTIFIABLE, IT IS
NOT... IT IS LOOKING AT
REGIONS... FOR EXAMPLE, HOW
WE'RE DOING... WE'RE LOOKING AT
POSTAL CODE REGIONS AND SAYING
THERE'S AN ISSUE IN THIS REGION,
LET'S GET HELP THERE.
NOT INDIVIDUALLY.
IN THAT CASE YOU'RE ACTUALLY
GOING TO BE MORE ACCURATE
BECAUSE YOU'RE LOOKING AT
POPULATION LEVEL AS OPPOSED TO
ONE INDIVIDUAL WHO SAYS I'VE HAD
A BAD DAY AND YOU'RE DRIVING TO
HIS HOUSE.
WHEN YOU SEE THAT THERE'S A
NUMBER OF PEOPLE IN THE
COMMUNITY ARE HAVING PROBLEMS,
THAT INDICATES THAT THAT COULD
BE REAL.
THERE MIGHT HAVE BEEN... OR IT
COULD BE CYBER BULLYING IN THE
SCHOOL.
YOU WANT TO BE DOING IT AT THAT
POPULATION LEVEL BECAUSE YOU'LL
BE MUCH MORE ACCURATE.
YOU DON'T WANT TO BE TARGETING
INDIVIDUALS.
WE DO THAT TODAY WITH STATISTICS
DATA ALL OVER THE WORLD AND IT
WORKS WELL.

Steve says ALISON, WHEN YOU
WERE DESIGNING WOEBOT, DID YOU
TAKE PRIVACY CONCERNS INTO
CONCERN?

Alison says ABSOLUTELY.
WE ARE ACTUALLY A BUNCH OF
CLINICIANS AND FORMER
RESEARCHERS FROM STANFORD, SO WE
KIND OF LIVE AND BREATHE THIS
STUFF AS WELL.
BUT CRUCIALLY, A PRODUCT LIKE
THIS DEPENDS ON TRUST, AND SO IF
YOU... YOU KNOW, SO WE HAVE
HOSPITAL-LEVEL ENCRYPTION, WE
ARE GDPR CLIENT, WHICH IS
ACTUALLY A VERY HIGH STANDARD.
AND PART OF THE REASON FOR
ACTUALLY ESTABLISHING WOEBOT WAS
BECAUSE WE REALLY BELIEVED THAT
CLINICIANS LIKE US REALLY NEED
TO START PARTNERING WITH
TECHNOLOGIES TO MAKE THESE KINDS
OF THERAPIES THAT HAVE
DEMONSTRATED EFFICACY WITHOUT A
THERAPIST MORE ENGAGING.
BECAUSE AS STEVE POINTS OUT, TO
DATE WHEN PEOPLE WHO HAVE REAL
SYMPTOMS ARE OFFERED A WEBSITE,
THEY DON'T REALLY STICK TO IT,
AND IT'S BECAUSE THEY HAVEN'T
BEEN BUILT TO BE DESIGNED TO BE
VERY EFFECTIVE OR ENGAGING, AND
SO THERE'S NO REASON WHY THESE
TWO WORLDS CAN'T COME TOGETHER,
BUT OF COURSE IT HAS TO REST ON
THE FOUNDATION OF RESPECT FOR
PEOPLE'S PRIVACY AND
CONFIDENTIALITY.

Steve says ALISON, I HAVE 30
SECONDS LEFT.
LET ME ASK IT TO YOU.
I THINK IT COULD BE ARGUED THAT
ONE OF THE REASONS SOME PEOPLE
ARE SUFFERING MENTAL HEALTH
PROBLEMS IN SOCIETY TODAY IS
THAT THEY'RE ON THEIR DEVICES
TOO DAMN MUCH AND THEY HAVE
COMPLETELY FORGOT ABOUT THE REAL
WORLD AND ARE LIVING IN THEIR
DEVICES, AND THERE IS AN IRONY
THAT ONE OF THE SOLUTIONS TO
THIS COMES FROM A DEVICE, AS YOU
HAVE DEVELOPED IT.
HOW DO YOU GET AROUND THAT IRONY?

Alison says TOTALLY AGREE.
AGAIN, THE DEVIL IS IN THE
DETAILS.
WE DON'T DO PERSUASIVE TECH.
WE DON'T TRY AND OPTIMIZE FOR
KEEPING PEOPLE IN THE
CONVERSATION AS LONG AS
POSSIBLE.
THESE ARE BRIEF EXPOSURES, FIVE
TO TEN MINUTES A DAY, AND I
THINK, YOU KNOW, REALLY
RESPONSIBLE APP DEVELOPERS
SHOULD BE DESIGNING THINGS LIKE
THIS.
BECAUSE AT THE END OF THE DAY
PROCESSING YOUR NEGATIVE
THINKING IS ACTUALLY KIND OF
CHALLENGING TO DO.
IT IS NOT CANDY CRUSH.
IT'S NOT INHERENTLY GOING TO BE
DOPAMINERGIC IN THAT WAY.
YOU KNOW, AGAIN, IT JUST COMES
DOWN TO REALLY GOOD DESIGN.
WHO IS THE TEAM BUILDING THIS
APP?
WHAT ARE THEIR CORE BELIEFS?
AND, YOU KNOW, WHAT IS THE
STRUCTURE THAT THEY HAVE IN
PLACE TO REALLY BUILD A GREAT
EXPERIENCE?

Steve says TERRIFIC. THAT'S
ALISON DARCY, CEO AND FOUNDER
OF WOEBOT LABS. SHE'S A
PSYCHOLOGIST AT STANFORD
UNIVERSITY. ALISON, WE THANK YOU
FOR BEING THERE FOR US ON THE
LINE FROM SAN FRANCISCO,
CALIFORNIA.

Alison says THANK YOU.

The caption changes to "Producer: Eric Bombicino, @ebombicino."

Steve says ERIN KELLY IS HERE,
THE CEO OF ADVANCED SYMBOLICS,
AND DAVID GRATZER, PROFESSOR
OF PSYCHIATRY AT U OF T,
ATTENDING PSYCHIATRIST AT CAMH.
GREAT TO HAVE YOU TWO ALONGSIDE AS WELL.

Erin and David say THANK YOU.

Watch: Mental Health: There's an App For That