科技英语摘要文章.docx
《科技英语摘要文章.docx》由会员分享,可在线阅读,更多相关《科技英语摘要文章.docx(22页珍藏版)》请在冰点文库上搜索。
科技英语摘要文章
I.INTRODUCTION
THEbirthofbigdata,asaconceptifnotasaterm,isusually
associatedwithaMETAGroupreportbyDougLaney
entitled“3-DDataManagement:
ControllingDataVolume,
Velocity,andVariety”publishedin2001[1].Furtherdevelopments
nowsuggestbigdataproblemsareidentifiedbythe
so-called“5V”:
volume(quantityofdata),variety(datafrom
differentcategories),velocity(fastgenerationofnewdata),veracity
(qualityofthedata),andvalue(inthedata)[2].
Foralongtimethedevelopmentofbigdatatechnologieswas
inspiredbybusinessintelligence[3]andbybigscience(such
astheLargeHadronCollideratCERN)[4].Butwhenin2009
GoogleFlu,simplybyanalyzingGooglequeries,predictedflulike
illnessratesasaccuratelyastheCDC’senormouslycomplex
andexpensivemonitoringnetwork[5],someanalystsstartedto
claimthatallproblemsofmodernhealthcarecouldbesolved
bybigdata[6].
In2005,thetermvirtualphysiologicalhuman(VPH)was
introducedtoindicate“aframeworkofmethodsandtechnologies
that,onceestablished,willmakepossiblethecollaborative
ManuscriptreceivedSeptember29,2014;revisedDecember14,2014;accepted
February6,2015.DateofpublicationFebruary24,2015;dateofcurrent
versionJuly23,2015.
M.VicecontiiswiththeVPHInstituteforIntegrativeBiomedicalResearch,
andtheInsigneoInstituteforInSilicoMedicine,UniversityofSheffield,
SheffieldS13JD,U.K.(e-mail:
m.viceconti@sheffield.ac.uk).
P.HunteriswiththeAucklandBioengineeringInstitute,Universityof
Auckland,1010Auckland,NewZealand(e-mail:
p.hunter@auckland.ac.nz).
R.HoseiswiththeInsigneoinstituteforInSilicoMedicine,Universityof
Sheffield,SheffieldS13JD,U.K.(e-mail:
d.r.hose@sheffield.ac.uk).
DigitalObjectIdentifier10.1109/JBHI.2015.2406883
investigationofthehumanbodyasasinglecomplexsystem”
[7],[8].Theideawasquitesimple:
1)Toreducethecomplexityoflivingorganisms,wedecompose
themintoparts(cells,tissues,organs,organsystems)
andinvestigateonepartinisolationfromtheothers.This
approachhasproduced,forexample,themedicalspecialties,
wherethenephrologistlooksonlyatyourkidneys,
andthedermatologistonlyatyourskin;thismakesitvery
difficulttocopewithmultiorganorsystemicdiseases,to
treatmultiplediseases(socommonintheageingpopulation),
andingeneraltounravelsystemicemergencedue
togenotype-phenotypeinteractions.
2)Butifwecanrecomposewithcomputermodelsallthe
dataandalltheknowledgewehaveobtainedabouteach
part,wecanusesimulationstoinvestigatehowtheseparts
interactwithoneanother,acrossspaceandtimeandacross
organsystems.
Thoughthismaybeconceptuallysimple,theVPHvision
containsatremendouschallenge,namely,thedevelopmentof
mathematicalmodelscapableofaccuratelypredictingwhatwill
happentoabiologicalsystem.Totacklethishugechallenge,
multifacetedresearchisnecessary:
aroundmedicalimaging
andsensingtechnologies(toproducequantitativedataabout
thepatient’sanatomyandphysiology)[9]–[11],dataprocessing
toextractfromsuchdatainformationthatinsomecases
isnotimmediatelyavailable[12]–[14],biomedicalmodeling
tocapturetheavailableknowledgeintopredictivesimulations
[15],[16],andcomputationalscienceandengineeringtorun
hugehypermodels(orchestrationsofmultiplemodels)under
theoperationalconditionsimposedbyclinicalusage[17]–[19];
seealsothespecialissueentirelydedicatedtomultiscale
modeling[20].
Buttherealchallengeistheproductionofthatmechanistic
knowledge,quantitative,anddefinedoverspace,timeand
acrossmultiplespace-timescales,capableofbeingpredictive
withsufficientaccuracy.Aftertenyearsofresearchthishasproduced
acompleximpactscenarioinwhichanumberoftarget
applications,wheresuchknowledgewasalreadyavailable,are
nowbeingtestedclinically;someexamplesofVPHapplications
thatreachedtheclinicalassessmentstageare:
1)TheVPHOPconsortiumdevelopedamultiscalemodeling
technologybasedonconventionaldiagnosticimaging
methodsthatmakesitpossible,inaclinicalsetting,to
predictforeachpatientthestrengthoftheirbones,how
thisstrengthislikelytochangeovertime,andtheprobability
thattheywilloverloadtheirbonesduringdailylife.
Withthesethreepredictions,theevaluationoftheabsolute
riskofbonefractureinpatientsaffectedbyosteoporosis
willbemuchmoreaccuratethananypredictionbasedon
externalandindirectdeterminants,asitisincurrentclinical
practice[21].
2)Morethan500000end-stagerenaldiseasepatientsin
Europeliveonchronicintermittenthaemodialysistreatment.
Asuccessfultreatmentcriticallydependsonawellfunctioning
vascularaccess,asurgicallycreatedarteriovenous
shuntusedtoconnectthepatientcirculationto
theartificialkidney.TheARCHprojectaimedtoimprove
theoutcomeofvascularaccesscreationandlong-term
functionwithanimage-based,patient-specificcomputational
modelingapproach.ARCHdevelopedpatientspecific
computationalmodelsforvascularsurgerythat
makespossibletoplansuchsurgeryinadvanceonthe
basisofthepatient’sdata,andobtainapredictionofthe
vascularaccessfunctionoutcome,allowinganoptimization
ofthesurgicalprocedureandareductionofassociated
complicationssuchasnonmaturation.Aprospectivestudy
iscurrentlyrunning,coordinatedbytheMarioNegriInstitute
inItaly.Preliminaryresultson63patientsconfirm
theefficacyofthistechnology[22].
3)Percutaneouscoronaryintervention(PCI)guidedbyfractional
flowreserve(FFR)issuperiortostandardassessment
alonetotreatcoronariesstenosis.FFR-guidedPCI
resultsinimprovedclinicaloutcomes,areductioninthe
numberofstentsimplanted,andreducedcost.However,
currentlyFFRisusedinfewpatients,becauseitisinvasive
anditrequiresspecialinstrumentation.Alessinvasive
FFRwouldbeavaluabletool.TheVirtuHeartproject
developedapatient-specificcomputermodelthataccurately
predictsmyocardialFFRfromangiographicimages
alone,inpatientswithcoronaryarterydisease.Ina
phase1studythemethodsshowedanaccuracyof97%,
whencomparedtostandardFFR[23].Asimilarapproach,
butbasedoncomputedtomographyimaging,isevenata
moreadvancedstage,havingrecentlycompletedaphase2
trial[24].
WhiletheseandsomeotherVPHprojectshavereachedthe
clinicalassessmentstage,quiteafewotherprojectsarestillin
thetechnologicaldevelopment,orpreclinicalassessmentphase.
Butinsomecasesthemechanisticknowledgecurrentlyavailable
simplyturnedouttobeinsufficienttodevelopclinicallyrelevant
models.
Soitisperhapsnotsurprisingthatrecently,especiallyinthe
areaofpersonalizedhealthcare(sopromisingbutsochallenging)
somepeoplehavestartedtoadvocatetheuseofbigdata
technologiesasanalternativeapproach,inordertoreducethe
complexitythatdevelopingareliable,quantitativemechanistic
knowledgeinvolves.
Thistrendisfascinatingfromanepistemologicalpointof
view.TheVPHwasbornaroundtheneedtoovercomethe
limitationsofabiologyfoundedonthecollectionofahuge
amountofobservationaldata,frequentlyaffectedbyconsiderable
noise,andboxedintoaradicalreductionismthatprevented
mostresearchersfromlookingatanythingbiggerthanasingle
cell[25],[26].Suggestingthatwereverttoaphenomenological
approachwhereapredictivemodelissupposedtoemergenot
frommechanistictheoriesbutbyonlydoinghigh-dimensional
bigdataanalysis,maybeperceivedbysomeasasteptoward
thatempiricismtheVPHwascreatedtoovercome.
Inthefollowing,wewillexplainwhytheuseofbigdatamethods
andtechnologiescouldactuallyempowerandstrengthen
currentVPHapproaches,increasingconsiderablyitschancesof
clinicalimpactinmany“difficult”targets.Butinorderforthat
tohappen,itisimportantthatbigdataresearchersareawarethat
whenusedinthecontextofcomputationalbiomedicine,bigdata
methodsneedtocopewithanumberofhurdlesthatarespecific
tothedomain.Onlybydevelopingaresearchagendaforbig
dataincomputationalbiomedicinecanwehopetoachievethis
ambitiousgoal.
II.DOCTORSANDENGINEERS:
JOINEDATTHEHIP
Asengineerswhohaveworkedformanyyearsinresearch
hospitals,werecognizethatclinicalandengineeringresearchers
shareasimilarmind-set.Bothintraditionalengineeringandin
medicine,theresearchdomainisdefinedintermsofproblem
solving,notofknowledgediscovery.Themottocommontoboth
disciplinesis“whateverworks.”
Butthereisafundamentaldifference:
Engineersusuallydeal
withproblemsrelatedtophenomenaonwhichthereisalarge
bodyofreliableknowledgefromphysicsandchemistry.When
agoodreliablemechanistictheoryisnotavailable,engineers
resorttoempiricalmodels,asfarastheycansolvetheproblem
athand.Butwhentheydothis,theyareleftwithasenseof
fragilityandmistrust,andtheytrytoreplacethemassoonas
possiblewiththeory-basedmechanisticmodels,whichareboth
predictiveandexplanatory.
Medicalresearchersdealwithproblemsforwhichthereis
amuchlesswell-establishedbodyofknowledge;inaddition,
thisknowledgeisfrequentlyqualitativeorsemiquantitative,and
obtainedinhighlycontrolledexperimentsquiteremovedfrom
clinicalreality,inordertotamethecomplexityinvolved.Thus,
notsurprisingly,manyclinicalresearchersconsidermechanistic
models“toosimpletobetrusted,”andingeneralthewholeidea
ofamechanisticmodel