Ovidiu
Leemos los datos y definimos variables nuevas:
df=read_sav("CopiaAnalisisDesember2019.sav")
df=df %>% mutate(TumorSolido=as.factor(TumorType==1),Diameter=ifelse(CTdiameter>=6,6,CTdiameter)) %>%
mutate(
Trifecta1= (ClampingTime<=20) & Bleeding<=500 & (IntraopComplicationYN==0),
Trifecta2= (ClampingTime<20) & Bleeding <500 & (IntraopComplicationYN==0),
Trifecta3= (ClampingTime<=20) & Bleeding<=500 & (RosenthalGrade<=1),
Trifecta4= (ClampingTime< 20) & Bleeding< 500 & (RosenthalGrade<=1),
) %>%
mutate(ChangeGFRabs=GFRDischarge-PreGFR) %>%
mutate(ChangeGFRabs=ChangeGFRabs/PreGFR*100) %>%
mutate(HospitalStay4=HospitalStay>=5,
ClavienDindoGrade3=ClavienDindoGrade>=3) %>% select(-OperDate)
df2=df %>% as_factor() %>% mutate(Trifecta3=as.factor(Trifecta3))
rndr <- function(x, name, ...) {
if (length(x) == 0) {
y <- df2[[name]]
s <- rep("", length(render.default(x=y, name=name, ...)))
if (is.numeric(y)) {
p <- t.test(y ~ df2$Trifecta3)$p.value
} else {
p <- chisq.test(table(y, droplevels(df2$Trifecta3)))$p.value
}
s[2] <- sub("<", "<", format.pval(p, digits=3, eps=0.001))
s
} else {
render.default(x=x, name=name, ...)
}
}
rndr.strat <- function(label, n, ...) {
ifelse(n==0, label, render.strat.default(label, n, ...))
}table1(~ . ,data=df2)| Overall (n=122) |
|
|---|---|
| Age | |
| Mean (SD) | 61.5 (12.2) |
| Median [Min, Max] | 65.0 [18.0, 85.0] |
| Sex | |
| Man | 80 (65.6%) |
| Woman | 42 (34.4%) |
| BMI | |
| Mean (SD) | 27.5 (4.78) |
| Median [Min, Max] | 27.1 [17.3, 47.9] |
| CharlsonIndex | |
| Mean (SD) | 4.58 (1.55) |
| Median [Min, Max] | 5.00 [2.00, 9.00] |
| ASA | |
| ASA I | 6 (4.9%) |
| ASA II | 61 (50.0%) |
| ASA III | 54 (44.3%) |
| ASA IV | 1 (0.8%) |
| Hb | |
| Mean (SD) | 14.3 (1.38) |
| Median [Min, Max] | 14.5 [9.20, 17.8] |
| PreGFR | |
| Mean (SD) | 80.8 (15.6) |
| Median [Min, Max] | 89.5 [13.0, 90.0] |
| TumorType | |
| Solid | 90 (73.8%) |
| Bosniak 2 | 0 (0%) |
| Bosniak 3 | 13 (10.7%) |
| Bosniak 4 | 19 (15.6%) |
| Number | |
| Mean (SD) | 1.04 (0.237) |
| Median [Min, Max] | 1.00 [1.00, 3.00] |
| Side | |
| Right | 66 (54.1%) |
| Left | 55 (45.1%) |
| 3 | 1 (0.8%) |
| CT størrelse | |
| Mean (SD) | 2.69 (1.20) |
| Median [Min, Max] | 2.45 [1.10, 9.00] |
| RENALscore | |
| Mean (SD) | 6.42 (1.79) |
| Median [Min, Max] | 6.50 [4.00, 10.0] |
| RenalSuffix | |
| a | 26 (21.3%) |
| p | 54 (44.3%) |
| x | 42 (34.4%) |
| OperatorAndTys | |
| And | 89 (73.0%) |
| Tys | 33 (27.0%) |
| Opr.tid min. | |
| Mean (SD) | 162 (35.8) |
| Median [Min, Max] | 157 [103, 261] |
| Blødning ml. | |
| Mean (SD) | 226 (282) |
| Median [Min, Max] | 100 [50.0, 2000] |
| ClampingTime | |
| Mean (SD) | 13.1 (5.22) |
| Median [Min, Max] | 12.0 [0.00, 32.0] |
| Dren | |
| No | 22 (18.0%) |
| Yes | 100 (82.0%) |
| Handport 0/1 | |
| No | 28 (23.0%) |
| Yes | 94 (77.0%) |
| IntraopComplicationYN | |
| No | 107 (87.7%) |
| Yes | 15 (12.3%) |
| Cause | |
| Surgeon | 11 (9.0%) |
| Anestesiolog | 4 (3.3%) |
| Missing | 107 (87.7%) |
| RosenthalGrade | |
| Mean (SD) | 0.213 (0.633) |
| Median [Min, Max] | 0.00 [0.00, 3.00] |
| Tumorstørrelse pT i cm. | |
| Mean (SD) | 2.46 (1.16) |
| Median [Min, Max] | 2.20 [1.00, 8.00] |
| Histology | |
| ClearCell | 55 (45.1%) |
| Papillary | 26 (21.3%) |
| Chromophobe | 7 (5.7%) |
| ClearCell Cystisc changes | 4 (3.3%) |
| Multilocular cystic | 2 (1.6%) |
| Oncocytoma | 17 (13.9%) |
| Others | 10 (8.2%) |
| Missing | 1 (0.8%) |
| pT | |
| pT1a | 105 (86.1%) |
| pT1b | 16 (13.1%) |
| pT2a | 0 (0%) |
| pT2b | 1 (0.8%) |
| pT3a | 0 (0%) |
| Margins | |
| Negative | 112 (91.8%) |
| Positive | 10 (8.2%) |
| uncertain | 0 (0%) |
| Postopr liggetid dager | |
| Mean (SD) | 4.57 (1.79) |
| Median [Min, Max] | 4.00 [2.00, 13.0] |
| GFRDischarge | |
| Mean (SD) | 74.8 (18.7) |
| Median [Min, Max] | 82.5 [12.0, 90.0] |
| GFR1Month | |
| Mean (SD) | 73.0 (18.8) |
| Median [Min, Max] | 81.0 [31.0, 90.0] |
| Missing | 81 (66.4%) |
| CRP1day | |
| Mean (SD) | 51.9 (38.5) |
| Median [Min, Max] | 44.5 [5.00, 235] |
| CRP2day | |
| Mean (SD) | 104 (73.9) |
| Median [Min, Max] | 81.5 [13.0, 400] |
| CRPmax | |
| Mean (SD) | 131 (86.8) |
| Median [Min, Max] | 109 [13.0, 461] |
| CRPut | |
| Mean (SD) | 71.8 (39.2) |
| Median [Min, Max] | 65.0 [9.00, 213] |
| Missing | 14 (11.5%) |
| Transfusjon ant enh | |
| Mean (SD) | 0.107 (0.442) |
| Median [Min, Max] | 0.00 [0.00, 3.00] |
| PostopComplicationYN | |
| No | 89 (73.0%) |
| Yes | 33 (27.0%) |
| ClavienDindoGrade | |
| Mean (SD) | 0.541 (1.08) |
| Median [Min, Max] | 0.00 [0.00, 4.00] |
| Komplikasjoner 0/1 | |
| 50 (41.0%) | |
| 0 | 34 (27.9%) |
| 1ileus + orchitt ve side etter hjemkomst,antibiotikabeh. | 1 (0.8%) |
| 1paralytisk ileus | 1 (0.8%) |
| Atb for mistanke UVI men ikke bekreftet med dyrkning | 1 (0.8%) |
| bilat lungeemboli 4 uker senere, Albyl e ikke sep | 1 (0.8%) |
| blødning fra binyre som måtte fjernes, brokk | 1 (0.8%) |
| blødning i trokar, nyresvikt (tidligre nefrektomert) | 1 (0.8%) |
| chylus ascites | 1 (0.8%) |
| CRP stigning behandlet med cefalo | 1 (0.8%) |
| eliquis | 1 (0.8%) |
| FEBER. infe ukjent fokus | 1 (0.8%) |
| Iiskemisk kolitt | 1 (0.8%) |
| ileus, konvertert til nefrektomi | 1 (0.8%) |
| infeksjon??? Sansynlig ikke | 1 (0.8%) |
| konvertert til nefrektomi | 1 (0.8%) |
| kreat stigning | 1 (0.8%) |
| lettgradig tarmparalyse | 1 (0.8%) |
| miltskade, UVI | 1 (0.8%) |
| nyrven skade, nyrepol ischemi | 1 (0.8%) |
| reinnlagt SSF med magesmerter 12 postopr dag | 1 (0.8%) |
| reoperasjon grunnet postop blødning med hematom evakueri | 1 (0.8%) |
| Resp svikt atrieflimmer, intensiv, telemetri AB behandlet | 1 (0.8%) |
| respirasjon svikt-UVI | 1 (0.8%) |
| respirasjonsproblem,pleuravæske,aspirasjonspneumoni | 1 (0.8%) |
| respirasjonsvikt | 1 (0.8%) |
| respirasjonsvikt + lunge atelectase med Int opphold | 1 (0.8%) |
| respirasjonsvikt, ileus | 1 (0.8%) |
| retroperitonealt hematom, pleuravæske, diare | 1 (0.8%) |
| smerter | 1 (0.8%) |
| subfebrilia, AB beh | 1 (0.8%) |
| subileus | 1 (0.8%) |
| tarmparalyse | 2 (1.6%) |
| tarmparalyse,oppkast | 1 (0.8%) |
| urinretensjon | 2 (1.6%) |
| UVI | 2 (1.6%) |
| UVI uten funn på CT, innlagt i 5 dager for atb | 1 (0.8%) |
| TumorSolido | |
| FALSE | 32 (26.2%) |
| TRUE | 90 (73.8%) |
| Diameter | |
| Mean (SD) | 2.67 (1.09) |
| Median [Min, Max] | 2.45 [1.10, 6.00] |
| Trifecta1 | |
| Yes | 99 (81.1%) |
| No | 23 (18.9%) |
| Trifecta2 | |
| Yes | 93 (76.2%) |
| No | 29 (23.8%) |
| Trifecta3 | |
| FALSE | 17 (13.9%) |
| TRUE | 105 (86.1%) |
| Trifecta4 | |
| Yes | 98 (80.3%) |
| No | 24 (19.7%) |
| ChangeGFRabs | |
| Mean (SD) | -7.73 (14.4) |
| Median [Min, Max] | -0.556 [-50.6, 36.7] |
| HospitalStay4 | |
| Yes | 49 (40.2%) |
| No | 73 (59.8%) |
| ClavienDindoGrade3 | |
| Yes | 10 (8.2%) |
| No | 112 (91.8%) |
df2$Trifecta3=factor(as.integer(df2$Trifecta3),levels=1:3,labels=c("FALSE","TRUE","p-value"))
table1(~ . |Trifecta3,data=df2 , render=rndr, render.strat=rndr.strat,overall=F,droplevels = F)## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect
## Warning in chisq.test(table(y, droplevels(df2$Trifecta3))): Chi-squared
## approximation may be incorrect| FALSE (n=17) |
TRUE (n=105) |
p-value | |
|---|---|---|---|
| Age | |||
| Mean (SD) | 63.9 (11.2) | 61.1 (12.4) | 0.348 |
| Median [Min, Max] | 65.0 [42.0, 85.0] | 65.0 [18.0, 85.0] | |
| Sex | |||
| Man | 14 (82.4%) | 66 (62.9%) | 0.196 |
| Woman | 3 (17.6%) | 39 (37.1%) | |
| BMI | |||
| Mean (SD) | 29.0 (3.70) | 27.3 (4.90) | 0.104 |
| Median [Min, Max] | 28.7 [23.2, 36.6] | 26.8 [17.3, 47.9] | |
| CharlsonIndex | |||
| Mean (SD) | 5.29 (1.86) | 4.47 (1.47) | 0.097 |
| Median [Min, Max] | 5.00 [2.00, 9.00] | 4.00 [2.00, 8.00] | |
| ASA | |||
| ASA I | 1 (5.9%) | 5 (4.8%) | 0.839 |
| ASA II | 7 (41.2%) | 54 (51.4%) | |
| ASA III | 9 (52.9%) | 45 (42.9%) | |
| ASA IV | 0 (0%) | 1 (1.0%) | |
| Hb | |||
| Mean (SD) | 14.1 (1.64) | 14.4 (1.34) | 0.478 |
| Median [Min, Max] | 14.4 [9.20, 16.3] | 14.5 [10.6, 17.8] | |
| PreGFR | |||
| Mean (SD) | 79.6 (15.7) | 81.0 (15.7) | 0.747 |
| Median [Min, Max] | 89.0 [45.0, 90.0] | 90.0 [13.0, 90.0] | |
| TumorType | |||
| Solid | 13 (76.5%) | 77 (73.3%) | NA |
| Bosniak 2 | 0 (0%) | 0 (0%) | |
| Bosniak 3 | 1 (5.9%) | 12 (11.4%) | |
| Bosniak 4 | 3 (17.6%) | 16 (15.2%) | |
| Number | |||
| Mean (SD) | 1.06 (0.243) | 1.04 (0.237) | 0.746 |
| Median [Min, Max] | 1.00 [1.00, 2.00] | 1.00 [1.00, 3.00] | |
| Side | |||
| Right | 9 (52.9%) | 57 (54.3%) | 0.912 |
| Left | 8 (47.1%) | 47 (44.8%) | |
| 3 | 0 (0%) | 1 (1.0%) | |
| CT størrelse | |||
| Mean (SD) | 3.55 (1.77) | 2.56 (1.02) | 0.036 |
| Median [Min, Max] | 3.30 [1.30, 9.00] | 2.30 [1.10, 5.60] | |
| RENALscore | |||
| Mean (SD) | 7.59 (1.46) | 6.23 (1.77) | 0.002 |
| Median [Min, Max] | 8.00 [4.00, 10.0] | 6.00 [4.00, 10.0] | |
| RenalSuffix | |||
| a | 5 (29.4%) | 21 (20.0%) | 0.672 |
| p | 7 (41.2%) | 47 (44.8%) | |
| x | 5 (29.4%) | 37 (35.2%) | |
| OperatorAndTys | |||
| And | 14 (82.4%) | 75 (71.4%) | 0.518 |
| Tys | 3 (17.6%) | 30 (28.6%) | |
| Opr.tid min. | |||
| Mean (SD) | 214 (39.8) | 154 (27.2) | <0.001 |
| Median [Min, Max] | 222 [131, 261] | 151 [103, 237] | |
| Blødning ml. | |||
| Mean (SD) | 635 (554) | 160 (117) | 0.003 |
| Median [Min, Max] | 300 [100, 2000] | 100 [50.0, 500] | |
| ClampingTime | |||
| Mean (SD) | 19.5 (7.59) | 12.0 (3.86) | <0.001 |
| Median [Min, Max] | 22.0 [7.00, 32.0] | 12.0 [0.00, 20.0] | |
| Dren | |||
| No | 0 (0%) | 22 (21.0%) | 0.081 |
| Yes | 17 (100%) | 83 (79.0%) | |
| Handport 0/1 | |||
| No | 3 (17.6%) | 25 (23.8%) | 0.803 |
| Yes | 14 (82.4%) | 80 (76.2%) | |
| IntraopComplicationYN | |||
| No | 8 (47.1%) | 99 (94.3%) | <0.001 |
| Yes | 9 (52.9%) | 6 (5.7%) | |
| Cause | |||
| Surgeon | 7 (41.2%) | 4 (3.8%) | 1 |
| Anestesiolog | 2 (11.8%) | 2 (1.9%) | |
| Missing | 8 (47.1%) | 99 (94.3%) | |
| RosenthalGrade | |||
| Mean (SD) | 1.18 (1.24) | 0.0571 (0.233) | 0.002 |
| Median [Min, Max] | 1.00 [0.00, 3.00] | 0.00 [0.00, 1.00] | |
| Tumorstørrelse pT i cm. | |||
| Mean (SD) | 2.94 (1.67) | 2.39 (1.04) | 0.201 |
| Median [Min, Max] | 2.50 [1.00, 8.00] | 2.10 [1.00, 5.70] | |
| Histology | |||
| ClearCell | 8 (47.1%) | 47 (44.8%) | 0.759 |
| Papillary | 5 (29.4%) | 21 (20.0%) | |
| Chromophobe | 1 (5.9%) | 6 (5.7%) | |
| ClearCell Cystisc changes | 0 (0%) | 4 (3.8%) | |
| Multilocular cystic | 0 (0%) | 2 (1.9%) | |
| Oncocytoma | 3 (17.6%) | 14 (13.3%) | |
| Others | 0 (0%) | 10 (9.5%) | |
| Missing | 0 (0%) | 1 (1.0%) | |
| pT | |||
| pT1a | 12 (70.6%) | 93 (88.6%) | NA |
| pT1b | 4 (23.5%) | 12 (11.4%) | |
| pT2a | 0 (0%) | 0 (0%) | |
| pT2b | 1 (5.9%) | 0 (0%) | |
| pT3a | 0 (0%) | 0 (0%) | |
| Margins | |||
| Negative | 15 (88.2%) | 97 (92.4%) | NA |
| Positive | 2 (11.8%) | 8 (7.6%) | |
| uncertain | 0 (0%) | 0 (0%) | |
| Postopr liggetid dager | |||
| Mean (SD) | 5.65 (2.06) | 4.40 (1.68) | 0.028 |
| Median [Min, Max] | 5.00 [3.00, 10.0] | 4.00 [2.00, 13.0] | |
| GFRDischarge | |||
| Mean (SD) | 67.9 (22.5) | 75.9 (17.9) | 0.183 |
| Median [Min, Max] | 77.0 [32.0, 90.0] | 84.0 [12.0, 90.0] | |
| GFR1Month | |||
| Mean (SD) | 59.1 (25.4) | 76.3 (15.6) | 0.103 |
| Median [Min, Max] | 50.0 [31.0, 90.0] | 81.0 [32.0, 90.0] | |
| Missing | 9 (52.9%) | 72 (68.6%) | |
| CRP1day | |||
| Mean (SD) | 89.8 (60.5) | 45.7 (29.8) | 0.009 |
| Median [Min, Max] | 89.0 [8.00, 235] | 40.0 [5.00, 166] | |
| CRP2day | |||
| Mean (SD) | 156 (103) | 95.4 (64.7) | 0.03 |
| Median [Min, Max] | 139 [14.0, 333] | 76.0 [13.0, 400] | |
| CRPmax | |||
| Mean (SD) | 187 (125) | 121 (75.9) | 0.049 |
| Median [Min, Max] | 154 [14.0, 461] | 100 [13.0, 400] | |
| CRPut | |||
| Mean (SD) | 69.7 (34.4) | 72.1 (40.2) | 0.803 |
| Median [Min, Max] | 66.0 [9.00, 140] | 64.0 [10.0, 213] | |
| Missing | 1 (5.9%) | 13 (12.4%) | |
| Transfusjon ant enh | |||
| Mean (SD) | 0.529 (0.943) | 0.0381 (0.237) | 0.048 |
| Median [Min, Max] | 0.00 [0.00, 3.00] | 0.00 [0.00, 2.00] | |
| PostopComplicationYN | |||
| No | 8 (47.1%) | 81 (77.1%) | 0.022 |
| Yes | 9 (52.9%) | 24 (22.9%) | |
| ClavienDindoGrade | |||
| Mean (SD) | 1.47 (1.70) | 0.390 (0.860) | 0.02 |
| Median [Min, Max] | 1.00 [0.00, 4.00] | 0.00 [0.00, 4.00] | |
| Komplikasjoner 0/1 | |||
| 5 (29.4%) | 45 (42.9%) | 0.002 | |
| 0 | 2 (11.8%) | 32 (30.5%) | |
| bilat lungeemboli 4 uker senere, Albyl e ikke sep | 1 (5.9%) | 0 (0%) | |
| blødning fra binyre som måtte fjernes, brokk | 1 (5.9%) | 0 (0%) | |
| ileus, konvertert til nefrektomi | 1 (5.9%) | 0 (0%) | |
| konvertert til nefrektomi | 1 (5.9%) | 0 (0%) | |
| miltskade, UVI | 1 (5.9%) | 0 (0%) | |
| Resp svikt atrieflimmer, intensiv, telemetri AB behandlet | 1 (5.9%) | 0 (0%) | |
| respirasjon svikt-UVI | 1 (5.9%) | 0 (0%) | |
| respirasjonsproblem,pleuravæske,aspirasjonspneumoni | 1 (5.9%) | 0 (0%) | |
| respirasjonsvikt + lunge atelectase med Int opphold | 1 (5.9%) | 0 (0%) | |
| tarmparalyse | 1 (5.9%) | 1 (1.0%) | |
| 1ileus + orchitt ve side etter hjemkomst,antibiotikabeh. | 0 (0%) | 1 (1.0%) | |
| 1paralytisk ileus | 0 (0%) | 1 (1.0%) | |
| Atb for mistanke UVI men ikke bekreftet med dyrkning | 0 (0%) | 1 (1.0%) | |
| blødning i trokar, nyresvikt (tidligre nefrektomert) | 0 (0%) | 1 (1.0%) | |
| chylus ascites | 0 (0%) | 1 (1.0%) | |
| CRP stigning behandlet med cefalo | 0 (0%) | 1 (1.0%) | |
| eliquis | 0 (0%) | 1 (1.0%) | |
| FEBER. infe ukjent fokus | 0 (0%) | 1 (1.0%) | |
| Iiskemisk kolitt | 0 (0%) | 1 (1.0%) | |
| infeksjon??? Sansynlig ikke | 0 (0%) | 1 (1.0%) | |
| kreat stigning | 0 (0%) | 1 (1.0%) | |
| lettgradig tarmparalyse | 0 (0%) | 1 (1.0%) | |
| nyrven skade, nyrepol ischemi | 0 (0%) | 1 (1.0%) | |
| reinnlagt SSF med magesmerter 12 postopr dag | 0 (0%) | 1 (1.0%) | |
| reoperasjon grunnet postop blødning med hematom evakueri | 0 (0%) | 1 (1.0%) | |
| respirasjonsvikt | 0 (0%) | 1 (1.0%) | |
| respirasjonsvikt, ileus | 0 (0%) | 1 (1.0%) | |
| retroperitonealt hematom, pleuravæske, diare | 0 (0%) | 1 (1.0%) | |
| smerter | 0 (0%) | 1 (1.0%) | |
| subfebrilia, AB beh | 0 (0%) | 1 (1.0%) | |
| subileus | 0 (0%) | 1 (1.0%) | |
| tarmparalyse,oppkast | 0 (0%) | 1 (1.0%) | |
| urinretensjon | 0 (0%) | 2 (1.9%) | |
| UVI | 0 (0%) | 2 (1.9%) | |
| UVI uten funn på CT, innlagt i 5 dager for atb | 0 (0%) | 1 (1.0%) | |
| TumorSolido | |||
| FALSE | 4 (23.5%) | 28 (26.7%) | 1 |
| TRUE | 13 (76.5%) | 77 (73.3%) | |
| Diameter | |||
| Mean (SD) | 3.38 (1.27) | 2.56 (1.02) | 0.02 |
| Median [Min, Max] | 3.30 [1.30, 6.00] | 2.30 [1.10, 5.60] | |
| Trifecta1 | |||
| Yes | 0 (0%) | 99 (94.3%) | <0.001 |
| No | 17 (100%) | 6 (5.7%) | |
| Trifecta2 | |||
| Yes | 0 (0%) | 93 (88.6%) | <0.001 |
| No | 17 (100%) | 12 (11.4%) | |
| Trifecta3 | |||
| FALSE | 17 (100%) | 0 (0%) | NA |
| TRUE | 0 (0%) | 105 (100%) | |
| p-value | 0 (0%) | 0 (0%) | |
| Trifecta4 | |||
| Yes | 0 (0%) | 98 (93.3%) | <0.001 |
| No | 17 (100%) | 7 (6.7%) | |
| ChangeGFRabs | |||
| Mean (SD) | -16.2 (17.0) | -6.36 (13.5) | 0.035 |
| Median [Min, Max] | -11.2 [-48.4, 0.00] | 0.00 [-50.6, 36.7] | |
| HospitalStay4 | |||
| Yes | 12 (70.6%) | 37 (35.2%) | 0.013 |
| No | 5 (29.4%) | 68 (64.8%) | |
| ClavienDindoGrade3 | |||
| Yes | 5 (29.4%) | 5 (4.8%) | 0.003 |
| No | 12 (70.6%) | 100 (95.2%) |
Parece que nos interesa:
Trifecta3 ~ 1 + BMI + CharlsonIndex + RENALscore
d <- datadist(df)
options(datadist='d')
f<-lrm(Trifecta3 ~ 1 + BMI + CharlsonIndex + RENALscore,data=df)
summary(f)## Effects Response : Trifecta3
##
## Factor Low High Diff. Effect S.E. Lower 0.95 Upper 0.95
## BMI 24.425 30.2 5.775 -0.49410 0.34745 -1.175100 0.18689
## Odds Ratio 24.425 30.2 5.775 0.61012 NA 0.308790 1.20550
## CharlsonIndex 3.250 6.0 2.750 -1.10130 0.50410 -2.089300 -0.11330
## Odds Ratio 3.250 6.0 2.750 0.33244 NA 0.123770 0.89289
## RENALscore 5.000 8.0 3.000 -1.54410 0.54590 -2.614000 -0.47415
## Odds Ratio 5.000 8.0 3.000 0.21351 NA 0.073239 0.62241#summary(f,BMI=c(20,40),Diameter=c(0,5))
plot(nomogram(f,fun=plogis,funlabel="Probability of Trifecta"
#, fun.at=c(.05,.1,.25,.5,.75,.9,.95), #Perdidas1Mes=c(0,5,10,20,seq(50,700,by=50))
))
Vamos a explorar la curva ROC:
df$predictions=predict(f)
df$labels=2*(df$Trifecta3-0.5)
precrec_obj <- evalmod(scores = df$predictions, labels = df$labels)
autoplot(precrec_obj)
uauc1 <- evalmod(scores=df$predictions, labels = df$labels, mode="aucroc")
# print 'aucroc'
uauc1##
## === Input data ===
##
## Model name Dataset ID # of negatives # of positives
## 1 m1 1 17 105
##
##
## === AUCs ===
##
## Model name Dataset ID AUC U
## 1 m1 1 0.7582633 1353.5Bootstrap
Vamos a usar la técnica de bootstrap con la idea de tener algo de validación interna del modelo de regresión anterior. Cada cálculo dará un resultado aleatorio. Vamos a fijar una semilla para reproducibilidad:
set.seed(20201001)logit = glm(Trifecta3 ~ 1 + BMI + CharlsonIndex + RENALscore,data=df, family = binomial(link = "logit"))
logit %>% summary()##
## Call:
## glm(formula = Trifecta3 ~ 1 + BMI + CharlsonIndex + RENALscore,
## family = binomial(link = "logit"), data = df)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.4754 0.2066 0.3484 0.5815 1.1064
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 9.76122 2.61667 3.730 0.000191 ***
## BMI -0.08556 0.06016 -1.422 0.154986
## CharlsonIndex -0.40047 0.18330 -2.185 0.028901 *
## RENALscore -0.51470 0.18195 -2.829 0.004673 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 98.520 on 121 degrees of freedom
## Residual deviance: 82.789 on 118 degrees of freedom
## AIC: 90.789
##
## Number of Fisher Scoring iterations: 5Vamos a aplicar la técnica de bootstrap para estudiar la estabilidad de las estimaciones de los coeficientes de la regresión logística:.
R = 10000 # number of bootstrap samples
n = nrow(df) # sample size
k = length(coef(logit)) # number of coefficients
# set up a empty Rxn matrix B
B = matrix(nrow = R, ncol = k,
dimnames = list(paste("Sample",1:R),
names(coef(logit))))
# loop R times
for(i in 1:R){
# sample credit data with replacement
boot.data = df[sample(x = 1:n, size = n, replace = TRUE), ]
# fit the model on the boostrapped sample
boot.logit = glm(logit$formula,
data=boot.data, family = binomial(link = "logit"))
# store the coefficients
B[i,] = coef(boot.logit)
}
# function to return bootstrapped coefficients
myLogitCoef <- function(data, indices, formula) {
d <- data[indices,]
fit <- glm(formula, data=d, family = binomial(link = "logit"))
return(coef(fit))
}
# set up cluster of 4 CPU cores
cl<-makeCluster(4)
clusterExport(cl, 'myLogitCoef')
set.seed(373)
coef.boot <- boot(data=df, statistic=myLogitCoef, R=10000,
formula= logit$formula,
# process in parallel across 4 CPU cores
parallel = 'snow', ncpus=4, cl=cl)
stopCluster(cl)
coef.boot##
## ORDINARY NONPARAMETRIC BOOTSTRAP
##
##
## Call:
## boot(data = df, statistic = myLogitCoef, R = 10000, formula = logit$formula,
## parallel = "snow", ncpus = 4, cl = cl)
##
##
## Bootstrap Statistics :
## original bias std. error
## t1* 9.7612216 0.5328396901 2.68695651
## t2* -0.0855582 -0.0004506506 0.05976187
## t3* -0.4004748 -0.0224398008 0.21652931
## t4* -0.5146958 -0.0423837644 0.19574431No parece que los sesgos sean importantes. Ni en valor, ni menos tras compararlos con el tamaño del error standard. creo que basta decirlo de palabra. No será necesario enseñarlo.
Vamos a usar ahora bootstrap para estimar AUC:
prob = predict.glm(logit, type='response', newdata = df)
pred = prediction(prob, df$Trifecta3)
# AUC
auc = performance(pred,"auc")@y.values[[1]][1]
# plot the ROC curve
perf <- performance(pred,"tpr","fpr")
plot(perf, col="navyblue", cex.main=1,
main= paste("Logistic Regression ROC Curve: AUC =", round(auc,3)))
abline(a=0, b = 1, col='darkorange1')
R = 10000
n = nrow(df)
# empty Rx2 matrix for bootstrap results
B = matrix(nrow = R, ncol = 2,
dimnames = list(paste('Sample',1:R),
c("auc_orig","auc_boot")))
set.seed(701)
for(i in 1:R){
# draw a random sample
obs.boot <- sample(x = 1:n, size = n, replace = T)
data.boot <- df[obs.boot, ]
# fit the model on bootstrap sample
logit.boot <- glm(logit$formula ,
data=data.boot,
family = binomial(link = "logit"))
# apply model to original data
prob1 = predict(logit.boot, type='response', df)
pred1 = prediction(prob1, df$Trifecta3)
auc1 = performance(pred1,"auc")@y.values[[1]][1]
B[i, 1] = auc1
# apply model to bootstrap data
prob2 = predict(logit.boot, type='response', data.boot)
pred2 = prediction(prob2, data.boot$Trifecta3)
auc2 = performance(pred2,"auc")@y.values[[1]][1]
B[i, 2] = auc2
}
resultado=B %>% as_data_frame() %>% as_tibble() %>% mutate(optimism=auc_boot-auc_orig)## Warning: `as_data_frame()` is deprecated, use `as_tibble()` (but mind the new semantics).
## This warning is displayed once per session.Más o menos se obtiene los mismo. El optimismo (bias) es pequeño.
resultado %>% summarise_all(mean)## # A tibble: 1 x 3
## auc_orig auc_boot optimism
## <dbl> <dbl> <dbl>
## 1 0.752 0.782 0.0301La variabilidad de las estimacines anteriores también es pequeña.
resultado %>% summarise_all(sd)## # A tibble: 1 x 3
## auc_orig auc_boot optimism
## <dbl> <dbl> <dbl>
## 1 0.0155 0.0551 0.0565Más o menos vemos lo mismo. No tenemos problemas de sesgo. Se puede ignorar lo que hemos hecho y comentarlo de palabra.