Maintaining accurate temperature records is a crucial part of medical logistics for keeping supplies like blood products and vaccines safe. A lot of forecasting models fail to give exact predictions, which can make it hard to keep transportation safe. To fix this, we created X-STEP, a stacked ensemble model based on XGBoost that can predict temperatures in real time. Linear regression, decision trees, random forest, and support vector machines are some of the base models that X-STEP combines. We collected data using IoT-based temperature control systems that were put in medical transport vehicles. We used fivefold cross-validation to train and test the model to make sure the evaluation was accurate. It was found that X-STEP had a R2 of 0.912 and a root mean square error (RMSE) of 1.46 °C. These outcomes are better than those from the random forest model and the support vector regression model. X-STEP can predict what will happen in real time, which helps keep medical supplies safe while they are being shipped. It also helps improve medical logistics by lowering risks related to temperature and making the supply chain work better.