Extubation failure (EF), defined as the necessity for reintubation within 48 h after extubation, is associated with higher mortality rates in patients requiring invasive res- piratory support in intensive care units (ICUs). Current meth- ods for predicting EF lack consistent effectiveness, and research on post-extubation indicators that could enhance the manage- ment of respiratory distress is limited. This study examines fea- tures linked to EF using the MIMIC-III dataset, which con- tains health records of over 40,000 critically ill patients. From this dataset, we identified 1,998 patients who underwent invasive ventilation for more than 24 h, including 86 EF. We detail the data pre-processing approach used to extract relevant data from the MIMIC-III dataset and present preliminary findings. Initial analysis reveals a statistically significant lower (p < 0.05) mean heart rate of the EF group during the first 24 h post- extubation. This finding underscores the utility of the MIMIC- III dataset in identifying critical parameters associated with ex- tubation failure in the post-extubation period.