low accuracy in random forest model

Question

HCV.Egy.Data <- read.csv("~/Stat/Metaheuristic ML/hepatitis+c+virus+hcv+for+egyptian+patients/HCV-Egy-Data.csv")
table(HCV.Egy.Data$Baselinehistological.staging)
HCV.Egy.Data$Baselinehistological.staging <- 
    as.factor(HCV.Egy.Data$Baselinehistological.staging)
library("randomForest")
library(caTools)
split <- sample.split(HCV.Egy.Data, SplitRatio = 0.7) 
split 
train <- subset(HCV.Egy.Data, split == "TRUE") 
test <- subset(HCV.Egy.Data, split == "FALSE") 
set.seed(120)  # Setting seed 
classifier_RF = randomForest(x = train[,-c(28,29)], 
                         y = train$Baselinehistological.staging, 
                             ntree = 500, type="classification") 
classifier_RF$classes
y_pred = predict(classifier_RF, newdata = test[, -c(28,29)]) 
confusion_mtx = table(test[, 29], y_pred) 
confusion_mtx

table(train[, 29], classifier_RF$predicted)

I have low accuracy (20-30%) for my ML models for this model using a lot of different ML methods (random forest, XGboost, adaboost, LR, etc.) Here is the confusion matrix from running the above code:

 y_pred
     1  2  3  4
  1 13 22 36 38
  2 21 17 39 33
  3 15 20 38 27
  4 13 16 37 45

We even applied SMOTE and the discretization steps to increase the accuracy to only 33%. Does anyone know why the papers that have used this dataset are getting accuracies in the 70-95% range? Is something glaringly missing?

For example, here's a paper: https://www.researchgate.net/profile/Md-Satu/publication/341987762_Predicting_Infectious_State_of_Hepatitis_C_Virus_Affected_Patient%27s_Applying_Machine_Learning_Methods/links/5edcab8a45851529453fc609/Predicting-Infectious-State-of-Hepatitis-C-Virus-Affected-Patients-Applying-Machine-Learning-Methods.pdf

We also applied SMOTE 8 times but accuracy only increased to 33%.

Here is the dataset: https://archive.ics.uci.edu/dataset/503/hepatitis+c+virus+hcv+for+egyptian+patients If you know any fixes, it would be much appreciated.