# install.packages("readxl")
library(readxl)  

data <- read_excel("E://raw data.xlsx")
head(data)

data$Risk_Score <- 0.011 * data$`Times of antibiotic use` + 0.005 * data$`Length of hospital stay`

# install.packages("pROC")
library(pROC)

predicted_probs <- data$Risk_Score  

roc_curve <- roc(data$`Symptomatic urinary tract infection`, predicted_probs)

auc_value <- auc(roc_curve)
cat("AUC: ", auc_value)

plot(roc_curve, main = "ROC Curve for Multivariate Logistic Regression Model",
     col = "blue", lwd = 2, xlab = "1 - Specificity", ylab = "Sensitivity")

# install.packages("ResourceSelection")
library(ResourceSelection)

hoslem_test <- hoslem.test(data$`Symptomatic urinary tract infection`, predicted_probs)
print(hoslem_test)

# install.packages("boot")
library(boot)

bootstrap_auc_function <- function(data, indices) {
  sample_data <- data[indices, ]
  predicted_probs <- sample_data$Risk_Score
  roc_curve <- roc(sample_data$`Symptomatic urinary tract infection`, predicted_probs)
  return(auc(roc_curve))
}

bootstrap_auc_results <- boot(data = data, statistic = bootstrap_auc_function, R = 1000)
print(bootstrap_auc_results)

boot.ci(bootstrap_auc_results, type = "bca")

# install.packages("rms")
library(rms)

calibration_model <- lrm(`Symptomatic urinary tract infection` ~ `Times of antibiotic use` + 
                           `Length of hospital stay`, data = data)
calibration_slope <- calibration_model$coefficients[2]   
cat("Calibration Slope: ", calibration_slope)

bootstrap_calibration_function <- function(data, indices) {
  sample_data <- data[indices, ]
  calibration_model <- lrm(`Symptomatic urinary tract infection` ~ `Times of antibiotic use` + 
                             `Length of hospital stay`, data = sample_data)
  return(calibration_model$coefficients[2])  
}

bootstrap_calibration_results <- boot(data = data, statistic = bootstrap_calibration_function, R = 1000)
print(bootstrap_calibration_results)

boot.ci(bootstrap_calibration_results, type = "bca")

# install.packages("car") 
library(car)

multi_model <- glm(`Symptomatic urinary tract infection` ~ `Times of indwelling catheter use` + 
                     Department + `Times of antibiotic use` + `Length of hospital stay` + 
                     Polytrauma + `Injury level` + `AIS grade`, 
                   data = data, family = "binomial")
vif_values <- vif(multi_model)
print(vif_values)