view . spring2015
9
Opposite:ComputerScienceprofessorDr.RobinGras in theOdetteSchool
of BusinessFinancialMarkets lab.
What if youcouldpredictwhenanepilepticpersonwasgoing
to sufferhisorhernext seizure?
UWindsor computer scienceprofessorDr. RobinGrashas
developed software that cando just that.
Gras’workhasproven that somecomplexchaotic systemsare
no longerunpredictable.Oneexampleof acomplex system is the
brain.Another is the stockmarket. Essentially, theseare things inwhich
countless small interactionscontribute toafinal outcome.
“Themost common image is that of thebutterflyeffect,” says
Gras. “The small change inacomplex systemcanhavedramatic
consequencesa long time later, onavery large scale.”
Nature is full of complex systems that aremadeupof seemingly
randomevents.Until now, that chaoshas left usbewilderedwhen
trying topredict the futureof these systems, saysGras. But now, he
andhisgraduate student,AbbasGolestani, havedeveloped software
that’scapableof accurately forecasting the futureof chaotic systems.
Gras isanassociateprofessor in theSchool of ComputerScience
andCanadaResearchChair inLearningandSimulation forTheoretical
Biology.HeandGolestani useddataand their software,GenericPred,
to successfullypredict the futureof financialmarkets, climatechange
andepileptic seizures.
Theypublished theirfindings in theOctober 2014editionof the
journal,
ScientificReports
. “Other existingmethodscannot predict
trends.Theirpredictionsdivergevery fast.Now, someonecan
effectivelypredict over a longperiodof time.”
Thefinancial industry isaparticularlycomplex system that varies
drastically fromday todayand is impactedbymany individuals.Gras
says itwouldbe incredible if wecouldbewarned that aneconomic
meltdownwas imminent.
GrasandGolestani plugged in twoyearsof stockprices from
theDowJones IndustrialAverage into theirmodel andwereable
to forecast thenext sixmonths.
GrasandGolestani haveapplied the software toclimatechange
prediction.Theyhave forecasted thenext 30yearsof climatechange
basedon100yearsof previousdata.
Thefindingswerenot positive for theenvironment.Grasplans
tocontinue the research to try forecasting the futureof earthquakes
aswell as suchmedical issuesasheart attacksandParkinson’s.
The teamhasevenmadeamedical breakthrough.Theywereable
topredict if someonewithepilepsywouldhavea seizure in thenext
17minutes.
“Upuntil now, thebest predictionwasa fewminutesaheadand
hadanaccuracyof 60 to70per cent,”Gras said. “Wegot 17minutes
with100per cent accuracy for 21different patients.”
The research teamusedelectroencephalography (EEG) readings
of patients,whichmeasures thebrain’selectrical activityalong the
scalp. Ahalf-hourof EEG readingsplugged into the softwarewas
enoughdata togive someonewithepilepsya17-minuteheads-up that
heor shewasgoing tohavea seizure.
Themedical breakthroughfindingshaveattractedmediaattention
acrossCanada.Gras’ discoverywaswrittenup in
TheWindsorStar
,
OttawaCitizen,VancouverSun
andRadio-Canada’snational French
languagebroadcast andCBCWindsor’sWindsorMorninghave
interviewedhim.Hewasalso featuredonCTV’sflagshipmorning
television show,CanadaAM.
A specialmonitoringdevicewouldbe required to senddata to
the team’sGenericPred software installedonapatient’scellphone.
GrasandGolestani havepatented the softwareandhope toprovide
thepatent toacompany thatwoulddesign suchadevice.
If adevice isdeveloped, itwouldgiveaperson thechance to
prepare for a seizure. “Itwould readyourEEGvalues, ringand say,
‘Youwill have trouble in17minutes. Stopyour car.Go toa safeplace.
Phone someone tohelpyou.’”
n
v
PREDICTING
THEUNPREDICTABLE
RESEARCH
BYSARAELLIOTT
Dr. RobinGras
