1

I am fitting a logistic growth model for each subject.

When I run my loop to fit the model for every patient. There is an error as some patients the model does not converge as they do not have a logistic growth. I want my loop to continue if there is an error in fitting the model and just have NA's in the dataset where I am capturing the coefficients

    growth_rate_l <- function(patient, data) {
  data0 <- data %>% dplyr::filter(pat== patient)
  fit0 <- nls(MRDRSLT ~ SSlogis(TIME, phi1, phi2, phi3), data = data0)
  t_doubling <- (summary(fit0)$coefficients)
  aic <-AIC(fit0, k=3)
  outdata <- c(t_doubling,aic)
  outdata
}

for (i in unique(doub1log$pat)) {
  doub1log$phi1[doub1log$pat== i] <- growth_rate_l(i, doub1log)[1]
  doub1log$phi2[doub1log$pat== i] <- growth_rate_l(i, doub1log)[2]
  doub1log$phi3[doub1log$pat == i] <- growth_rate_l(i, doub1log)[3]
  doub1log$AIC[doub1log$pat== i] <- growth_rate_l(i, doub1log)[13]
}

For some subjects the model will not fit and I get an error from nls

Error in nls(y ~ 1/(1 + exp((xmid - x)/scal)), data = xy, start = list(xmid = aux[[1L]],  : 
  step factor 0.000488281 reduced below 'minFactor' of 0.000976562

Once it errors the loop stops, but I want it to carry until the end and have NA's for subjects whose models cannot be fitted.

I tried "try" and "tryCatch" but it is not intuitive any ideas on how to deal with this issue?

I tried tryCatch(, error = function(e) { skip_to_next <<- TRUE})    if(skip_to_next) { next }  ``` i tired this in the loop above

I put this in the loop above but it is not working. I am unsure whether I need to capture the error in the loop or in the function.

I used to be a stata user and it had a command called capture which allowed a loop to run regardless of error.

danronmoon
  • 3,814
  • 5
  • 34
  • 56
Bobby
  • 13
  • 3

1 Answers1

0
growth_rate_l <- function(patient, data) {
  data0 <- data %>% dplyr::filter(pat== patient)
  fit0 <- nls(MRDRSLT ~ SSlogis(TIME, phi1, phi2, phi3), data = data0, na.action = na.omit)
  t_doubling <- (summary(fit0)$coefficients)
  aic <-AIC(fit0, k=3)
  outdata <- c(t_doubling,aic)
  outdata
}

for (i in unique(doub1log$pat)) {
tryCatch({
  doub1log$phi1[doub1log$pat== i] <- growth_rate_l(i, doub1log)[1]
  doub1log$phi2[doub1log$pat== i] <- growth_rate_l(i, doub1log)[2]
  doub1log$phi3[doub1log$pat== i] <- growth_rate_l(i, doub1log)[3]
  doub1log$AIC[doub1log$pat== i] <- growth_rate_l(i, doub1log)[13]
}, error=function(e){})
}

You need add na.omit in the formula and inlcude a try catch in the loop for it to work. Skipping error in for-loop Fit multiple Gompertz curves and skip errors in R (nlsList and SSgompertz)

Bobby
  • 13
  • 3