-
Notifications
You must be signed in to change notification settings - Fork 1
/
interaction.R
149 lines (108 loc) · 4.58 KB
/
interaction.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
# pm1.5 contrast
pm_contrast <- function(model, pm0, pm1, conf.level = 0.95) {
lvl <- 1 - ((1 - conf.level)/2)
z <- qnorm(lvl, mean = 0, sd = 1)
X <- data.frame(onMeds = c(0,1,0,1), pm = c(pm0, pm0, pm1, pm1))
X$pm_nomed <- ifelse(X$onMeds == 0, X$pm, 8)
X$pm_med <- ifelse(X$onMeds == 1, X$pm, 8)
W <- model.matrix(delete.response(model$term[1:3]), X)[,-1]
lp <- c(W%*%model$coefficients)
V <- model$var
h2 <- c(lp[2] - lp[1])
h3 <- c(lp[3] - lp[1])
h4 <- c(lp[4] - lp[2])
w2 <- c(W[2,] - W[1,])
w3 <- c(W[3,] - W[1,])
w4 <- c(W[4,] - W[2,])
est <- c(h2, h3, h4)
estvar <- diag(rbind(w2,w3,w4) %*% V %*% cbind(w2,w3,w4))
lower <- exp(est - z*sqrt(estvar))
upper <- exp(est + z*sqrt(estvar))
cbind(log.hr = est, log.var = estvar, hr = exp(est),
lower = lower, upper = upper,
pm0 = rep(pm0, 3), pm1 = rep(pm1,3))
}
# Cox model additive interaction with Splines
add_interact_cox <- function(model, pm0, pm1, conf.level = 0.95) {
lvl <- 1 - ((1 - conf.level)/2)
z <- qnorm(lvl, mean = 0, sd = 1)
X <- data.frame(onMeds = c(0,1,0,1), pm = c(pm0, pm0, pm1, pm1))
X$pm_nomed <- ifelse(X$onMeds == 0, X$pm, 8)
X$pm_med <- ifelse(X$onMeds == 1, X$pm, 8)
W <- model.matrix(delete.response(model$terms[1:3]), X)[,-1]
lp <- c(W%*%model$coefficients)
V <- model$var
col1 <- length(model$coefficients)
col0 <- (col1 - 1)/2 + 1
# these indexes change based on df of pspline()
a2 <- W[2,2:col0] - W[1,2:col0]
a3 <- W[3,2:col0] - W[1,2:col0]
a4 <- W[4,2:col0] - W[1,2:col0]
b2 <- W[2,(col0 + 1):col1] - W[1,(col0 + 1):col1]
b3 <- W[3,(col0 + 1):col1] - W[1,(col0 + 1):col1]
b4 <- W[4,(col0 + 1):col1] - W[1,(col0 + 1):col1]
h0 <- c(exp(lp[4] - lp[1]) - exp(lp[2] - lp[1]))
h1 <- c(a4*exp(lp[4] - lp[1]) - a3*exp(lp[3] - lp[1]) - a2*exp(lp[2] - lp[1]))
h2 <- c(b4*exp(lp[4] - lp[1]) - b3*exp(lp[3] - lp[1]) - b2*exp(lp[2] - lp[1]))
reri.var <- t(c(h0,h1,h2)) %*% V %*% c(h0,h1,h2)
reri.se <- sqrt(reri.var)
reri.p <- c(exp(lp[4] - lp[1]) - exp(lp[3] - lp[1]) - exp(lp[2] - lp[1])) + 1
reri.l <- reri.p - (z * reri.se)
reri.u <- reri.p + (z * reri.se)
rval <- c(est = unname(reri.p), lower = reri.l, upper = reri.u, pm0 = pm0, pm1 = pm1)
return(rval)
}
# Aalen model additive interaction
add_interact_aalen <- function(model, pm0, pm1, mu.pm, idx = NULL,
conf.level = 0.95, monotone = TRUE) {
lvl <- 1 - ((1 - conf.level)/2)
z <- qnorm(lvl, mean = 0, sd = 1)
pm1 <- pm1 - mu.pm
pm0 <- pm0 - mu.pm
if (class(model)[1] != "aalen")
stop("Error: model must be aalen object")
design <- rbind(c(1,0,pm0,0),
c(1,1,pm0,pm0),
c(1,0,pm1,0),
c(1,1,pm1,pm1))
if (monotone) {
out <- sapply(1:length(idx), function(i, ...) {
beta <- model$cum[idx[i],-1]
time <- unname(model$cum[idx[i],1])
lp <- c(design%*%beta)
V <- model$covariance[[idx[i]]]
reri.p <- c(exp(-lp[2]) + exp(-lp[3]) - exp(-lp[4]) - exp(-lp[1]))
h0 <- -c(reri.p)
h1 <- c(exp(-lp[4]) - exp(-lp[2]))
h2 <- c(pm1*exp(-lp[4]) - pm1*exp(-lp[3]) - pm0*exp(-lp[2]) + pm0*exp(-lp[1]))
h3 <- c(pm1*exp(-lp[4]) - pm0*exp(-lp[2]))
reri.var <- t(c(h0,h1,h2,h3)) %*% V %*% c(h0,h1,h2,h3)
reri.se <- sqrt(reri.var)
reri.l <- reri.p - (z * reri.se)
reri.u <- reri.p + (z * reri.se)
rval <- c(time = time, pm0 = pm0 + mu.pm, pm1 = pm1 + mu.pm,
est = reri.p, lower = reri.l, upper = reri.u)
return(rval)
})
} else {
out <- sapply(1:length(idx), function(i, ...) {
beta <- model$cum[idx[i],-1]
time <- unname(model$cum[idx[i],1])
lp <- c(design%*%beta)
V <- model$covariance[[idx[i]]]
reri.p <- c(exp(-lp[2]) + exp(-lp[3]) - exp(-lp[4]) - 1)
h0 <- -c(reri.p + 1)
h1 <- c(exp(-lp[4]) - exp(-lp[2]))
h2 <- c(pm1*exp(-lp[4]) - pm1*exp(-lp[3]) - pm0*exp(-lp[2]))
h3 <- c(pm1*exp(-lp[4]) - pm0*exp(-lp[2]))
reri.var <- t(c(h0,h1,h2,h3)) %*% V %*% c(h0,h1,h2,h3)
reri.se <- sqrt(reri.var)
reri.l <- reri.p - (z * reri.se)
reri.u <- reri.p + (z * reri.se)
rval <- c(time = time, pm0 = pm0 + mu.pm, pm1 = pm1 + mu.pm,
est = reri.p, lower = reri.l, upper = reri.u)
return(rval)
})
}
return(out)
}