[PDF] Curvas ROC - idUS - Universidad de Sevilla

bitstream › handlePDF



Previous PDF Next PDF





Curvas ROC - Fisterra

Cité 20 fois — La curva ROC se obtiene representando, para cada posible elección de valor de corte, la FVP en 





El análisis de curvas ROC en estudios - CORE

1998 · Cité 13 fois — Mediante el análisis de curvas ROC se estudia la precisión diagnós- tica del Child Behavior Manual for rhe Child Behavior Checklist/4-18 and 1991 Profile Burlington, VT: Uni-



Sensibilité, spécificité, courbe ROC etc - Cedric-Cnam

2012 · Cité 2 fois — Facile d'avoir une bonne sensibilité: déclarer tout le monde positif mais peu spécifique • Un bon test 



Comparación de pruebas diagnósticas desde la curva ROC

2007 · Cité 28 fois — Palabras clave: curvas ROC, sensibilidad, especificidad, AUC, bootstrap Abstract We study the problem 





Curvas ROC para Datos de Supervivencia Aplicación a Datos

2010 · Cité 2 fois — Esta curva ROC sería, de hecho, aquella que tendría valores de sensibilidad iguales a 1-especificidad en 



Anthropometric indicators as predictors of high blood pressure

2011 · Cité 73 fois — Resultados: Las áreas bajo las curvas ROC con los intervalos de confianza fueron: Jóvenes del sexo 

[PDF] customer orientation pdf

[PDF] cv algerie doc

[PDF] cv algerie pdf

[PDF] cv de travail en algerie

[PDF] cv emi genie informatique

[PDF] cv emistes

[PDF] cv etudiant pdf

[PDF] cv exemple francais pdf

[PDF] cv freelance translator

[PDF] cv gestion de projet informatique

[PDF] cv gestionnaire de projet

[PDF] cv images for engineering ece pdf

[PDF] cv in english example word

[PDF] cv in english examples

[PDF] cv ingenieur industriel exemple

[1] [4]?

DBe(p))D=(

0Cuandoelindividuonopresentaelevento

1Cuandoelindividuos{presentaelevento

prevalencia=p=Pr(0estado0= 1) =Pr(D= 1) =Pr(presentarelevento)

1p=Pr(0estado0= 0) =Pr(D= 0) =Pr(nopresentarelevento)

0

Prueba0=(

Positivoy= 1sixc

Negativoy= 0six < c

8>>>< >>:V +si(D= 1; y= 1) F +si(D= 0; y= 1) F si(D= 1; y= 0) V si(D= 0; y= 0) ??????? ?? ?????c?? ?? ????? ?? ????? ? ????? ??????? ??? ?????? ??? ???? ?????

S=Pr(y= 1jD= 1) =Pr(y= 1\D= 1)P

r(D= 1) S=V+V ++F ++F V +V ++F=FV

E=Pr(y= 0jD= 0) =Pr(y= 0\D= 0)P

r(D= 0) E=VV +F+

1E= 1VV

+F+=V+F+V +F+VV +F+=F+V +F+ P r(D= 1jy= 1) =

Pr(y= 1jD= 1)Pr(D= 1)P

r(y= 1jD= 1)Pr(D= 1) +Pr(y= 1jD= 0)Pr(D= 0)=

SpSp+ (1E)(1p)

V PP=V+V

++F+ P r(D= 0jy= 0) =

Pr(y= 0jD= 0)Pr(D= 0)P

r(y= 0jD= 0)Pr(D= 0) +Pr(y= 0jD= 1)Pr(D= 1)=

E(1p)(1S)p+E(1p)

V PN=VV

+F

V G=V++VV

++F++V+F

FV P+FFN=Pr(y= 1jD= 1) +Pr(y= 0jD= 1) = 1

FFP+FV N=Pr(y= 1jD= 0) +Pr(y= 0jD= 0) = 1

D= 1FV P=V+V

++FFFN=FV ++F1

D= 0FFP=F+V

+F+FV N=VV +F+1

ROC(c) =(

y=S(c) x= 1E(c) ROC p(c) =( y=FV P(c) x=FFP(c) P

ROC(t) = 1FE(F1

S(1t)) 0t1

t= 1E= 1FS(x0) q=S= 1FE(x0) ???? ???? ???????x0????? ?? ????? ?? ??????

E=P(y= 0jD= 0))E(u) =FS(xu) =FS(u))

)(1E)(u) =FS(u > t) = 1FS(u)

S=P(y= 1jD= 1))S(u) =FE(x > u) = 1FE(u)

(1E;S) = (1FS(u);1FE(u))) t= 1FS(u))u=F1

S(1t))

ROC(t) = 1FE(F1

S(1t))

Z=^prprSD(pr)=^prprp^pr^qr=n

SD(S) =v

uutV +(V++F)F(V++F)(V++F)=sV +F(V++F)3

SD(E) =v

uutV (V+F+)F+(V+F+)(V+F+)=sV

F+(V+F+)3

IC(S) =FV PZ=2sV

+F(V++F)3

IC(E) =FV NZ=2sV

F+(V+F+)3

[42]?????? ??? ????

Fn:R![0;1]

x!^Fn(x) =1n n X i=1" i(x) i(x) =( 1xix

0xi> x

FnE(x) =1n

En EX i=1" i(x)

FnS(x) =1n

Sn SX i=1" i(x) fx1;:::;xnSg??? ??? ???

1^FnS(x);1^FnE(x)

ROCn(t) = 1^FnE(1^FnS(x))

?????F+?V+ [30]? ??? ????? ??? ?????[31]? fE(x) =1n Eh1n EX i=1K 1xxih 1 fS(x) =1m Sh2n SX i=1K 2xxih 2

RKi(x)x= 1

RxKi(x)x= 0

Rx2Ki(x)x >0

????i= 1;2?

FE(x) =1n

En EX i=1x 11h

1K1uxiEh

1 u

FS(x) =1m

Sm SX i=1x 11h

2K2uxiSh

2 u

FE(x) =1n

En EX i=1xxiEh 1

FS(x) =1m

Sm SX i=1xxiSh 2

ROC(t) = 1~FE~F1

S(1t) [7]? [1]? ??????? ? ???[33]? [36] [24]? ^a=^E^S^E b=^S^E

ROC(t) = 1(^a+^b1(1t)) 0t1

[7]? ?? ??????? ???

ROC(t) = 1FE(F1

F w;E(t) =FE(T1(t)) F w;S(t) =FS(T1(t)) ROC w(t) = 1Fw;E(F1 w;S(1t)) = = 1Fw;E(T(F1

S(1t))) =

= 1FE(T1(T(F1

S(1t)))) =

= 1FE(F1

S(1t)) =ROC(t)

ROC(t) = 1FE(F1

S(1t))

ROC(t) = 1FETT1F1

S(1t) =

= 1S+S1(1t)E E ???????S?E? = 1(^a+^b1(1t)) ROC H

0: ^a1= ^a2^^b1=^b2

H

1: ^a16= ^a2_^b16=^b2

2=^a12^V(^b12) +^b212^V(^a12)2^a12^b12cov(^a12;^b12)^

V(^a12)^V(^b12)cov(^a12;^b12)

[43]? ?????^a12= ^a1^a2?^b12=^b1^b2? ???

AC=Pr(y= 1jD= 1)Pr(D= 1) +Pr(y= 0jD= 0)Pr(D= 0))

AC=SPr(D= 1) +EPr(D= 0))

AC=Sprevalencia+E(1prevalencia)

AC=V++VV

Si(F++F)!0)^AC!1

PD=^ACFV NFV PFV N

=Pr(y= 1jD= 1) +Pr(y= 1jD= 0) =S+E1 =FV PFFP

LRP=Pr(y= 1jD= 1)P

r(y= 1jD= 0)

LRN=Pr(y= 0jD= 1)P

r(y= 0jD= 0)

LRN=1SE

LRP=FV PFFP

LRN=FFNFV N

SD([LRP) =sFFN

V ++FV NF

SD(\LRN) =sFV P

F +FFPV

IC(LRP) =FV PFFP

exp(Z1=2SD([LRP))

IC(LRN) =FFNFV N

exp(Z1=2SD([LRP))

Odds(A) =Pr(A)1Pr(A)=Pr(A)P

r(A) Odds presencia=Pr(y= 1jD= 1)P r(y= 0jD= 1) Odds Odds ausencia=Pr(y= 1jD= 0)P r(y= 0jD= 0) Odds ausencia=1EE

Oddsratio=OddspresenciaOdds

ausencia

Oddsratio=SE(1S)(1E)

d

OD=V+VF

F+ xjD= 0N(1;) xjD= 1N(2;) =separaciondispersion=jESj 8>< E=x E S=x S =SD(x) AUC=1 0

ROC(t)t

[9]???? ??????? ??????? [9]????? ??? ?? ???? ??? ?????? ??0;7????? ??? ???? ?????

AUC=TX

t=112 (FFPtFFPt1)(FV Pt+FV Pt1) ?????H0:Pr(x > y) = 1=2? ??? ?? ??????? ???? ?????H0:Pr(xS> xE) = 1=2? [10]? SD(AUC) =sAUC(1AUC) + (nE1)(Q1AUC2) + (nS1)(Q2AUC2)n EnS ???????Q Q

1=Pr(yEi> ySj; yEk> ySj)

i;k= 1;:::;nE j= 1;:::;nSQ Q

2=Pr(ySi> yEj; ySk> yEj)

i;k= 1;:::;nS j= 1;:::;nE AUC=1 0

ROC(t)t

ROC(t) = 1~FE~F1

S(1t) AUC=1 0

ROC(t)t=1

0

1~FE~F1

S(1t) t AUC=1 0

ROC(t)t=1

0 1FEF1

S(1t)t

AUC=1 0

ROC(t)t=1

0

1(^a+^b1(1t))t=

^ap1 +b2 ^E^Sp

2E+2S!

AUC (a;b)=b a

ROC(t)t

AUC(a;b)=b

a

ROC(t)t

AUC(a;b)=b1+b2+BX

t=A12 (FFPtFFPt1)(FV Pt+FV Pt1) ??????? ??1E???? ???? ????? ?? ??????B=min(bj)=jb???j= 1;:::;T?

1= (ia)12

(FV Pi+FV Pi1)b

2= (bj)12

(FV Pj+1+FV Pj)

0[AUC(a;b)ba1

AUC(1)

(a;b)[AUC(2) (a;b) [AUC(1) (a;b)?? ???? ??????? ????? ??? ???????(x=a; x=b)?? ??? ?? ??? [AUC(2) ?? ???? ?????? ?? ?? ???(a= 0; b= 1)? H

0:AUC1=AUC2

H

1:AUC16=AUC2

z=([AUC1[AUC2)SD([AUC1[AUC2)

SD([AUC1[AUC2) =

q^

V([AUC1) +^V([AUC2)2rSD([AUC1)SD([AUC2)

r=cov([AUC1;[AUC2)SD([AUC1)SD([AUC2) H

0:AUC= 0;5

H

1:AUC6= 0;5

z=[AUC0;5SD([AUC) [45]????? ????? ???? ? ????? ??? ?????? ??120? x dist(x) =x2+ (y1)2=x2+ [f(x)1]2

2x+ 2[f(x)1]f0(x) = 0 =)f0(x) =x1f(x)

x=f

0(x) =1E1S=1Pr(Y= 0jD= 0)1Pr(Y= 1jD= 1)=Pr(Y= 1jD= 0)P

r(Y= 0jD= 1)= P r(Y= 1TD= 0)P r(D= 0)Pr(D= 1)P r(Y= 0TD= 1)=pPr(falsospositivos)(1p)Pr(falsosnegativos) C esp=C0+CV+Pr(V+) +CVPr(V) +CF+Pr(F+) +CFPr(F)

8>>>>>><

>>>>>:C

0Costebase

C

V+Costedeverdaderospositivos

C

VCostedeverdaderosnegativos

C

F+Costedefalsospositivos

C

FCostedefalsosnegativos

8>>>< >>:P r(V+) =Pr(Enfermo)Pr(Y= 1jEnfermo) =pS prevalenciasensibilidad P r(V) =Pr(Sano)Pr(Y= 0jSano) = (1p)E(1prevalencia)especificidad P r(F+) =Pr(Sano)Pr(Y= 1jSano) = (1p)(1E) (1prevalencia)(1especificidad) P r(F) =Pr(Enfermo)Pr(Y= 0jEnfermo) =p(1S)prevalencia(1sensibilidad) C esp=C0+CV+pS+CV(1p)E+CF+(1p)(1E)+CFp(1S) =quotesdbs_dbs18.pdfusesText_24