fNIRS Parkinson DT study - Diagnostics

Diagnostics

# File names
files <- c("model_aim1_oa",
           "model_aim1_pd",
           "model_aim2_1_oa",
           "model_aim2_1_pd",
           "model_aim2_2_oa",
           "model_aim2_2_pd",
           "model_aim2_3_oa",
           "model_aim2_3_pd",
           "model_aim3_oa",
           "model_aim3_pd")

# Loop through each file
for (file in files) {
  
  folder = "../data/model_eval/"
  coeff_file = paste(folder, file, "_estimates.csv", sep="")
  data_file = paste(folder, file, "_data.csv", sep="")
  formula_file = paste(folder, file, "_formula.txt", sep="")
  formula_description = paste(folder, file, "_description.txt", sep="")
  
  # Read the coefficients from the CSV file
  coeffs <- read.csv(coeff_file)
  model_data <- read.csv(data_file)
  model_formula <- read_file(formula_file)
  model_formula <- sub("beta ~ ", "beta ~ -1 + ", model_formula)
  model_formula_descr <- read_file(formula_description)
  
  # Print a headline for the current file
  cat("\n\n## Analysis for", file, "\n\n")
  cat("\nFormula: ", model_formula_descr, "\n\n")

  # Set formula
  formula <- as.formula(model_formula)
  
  # Import data points
  n = length(model_data$beta)
  data <- model_data %>% select(-Residuals, -FittedValues)

  # Set up the model structure
  model <- lm(formula, data = data)
  
  # Manually set the model fields to exported data
  model$coefficients <- setNames(coeffs$Coefficients, coeffs$Predictors)
  model$residuals <- model_data$Residuals
  model$fitted.values <- model_data$FittedValues
  model$call <- formula
  model$terms <- terms(formula)
  model$qr <- qr(model.matrix(formula, data))
  model$rank <- ncol(model.matrix(formula, data))
  model$df.residual <- n - length(model$coefficients)
  
  # Diagnostic plots
  plot(model)
  influencePlot(model, main="Influence Plot", sub="Circle size is proportial to Cook's Distance")
}

Analysis for model_aim1_oa

Formula: beta ~ -1 + cond + st_step_time_var + age

Analysis for model_aim1_pd

Formula: beta ~ -1 + cond + st_step_time_var + age + updrs_3_motor

Analysis for model_aim2_1_oa

Formula: beta ~ -1 + cond + age + dt_cost_walk_speed

Analysis for model_aim2_1_pd

Formula: beta ~ -1 + cond + age + updrs_3_motor + dt_cost_walk_speed

Analysis for model_aim2_2_oa

Formula: beta ~ -1 + cond + age + dt_cost_stroop_time

Analysis for model_aim2_2_pd

Formula: beta ~ -1 + cond + age + updrs_3_motor + dt_cost_stroop_time

Analysis for model_aim2_3_oa

Formula: beta ~ -1 + cond + age + prio

Analysis for model_aim2_3_pd

Formula: beta ~ -1 + cond + age + updrs_3_motor + prio

Analysis for model_aim3_oa

Formula: beta ~ -1 + cond + st_step_time_var*tmt4 + age + edu

Analysis for model_aim3_pd

Formula: beta ~ -1 + cond + st_step_time_var*tmt4 + age + updrs_3_motor + edu