Septic shock, a condition with high mortality, necessitates early detection and management, especially in clean room environments where predicting it presents unique challenges. This study developed a machine learning model to predict septic shock in immunocompromised patients within such settings. Utilizing a dataset of 1223 admission records from 1160 patients who underwent hematopoietic cell transplant, the study employed a retrospective observational approach. The dataset was divided into five cross-validation folds and included thirty-five clinical variables like vital signs, laboratory tests, and cleanroom-related factors. Both time series and static data were incorporated, with missing values being linearly interpolated. The model's hybrid design integrated these diverse data forms to predict septic shock.The model's predictive ability was evaluated using metrics like the area under the precision-recall curve (AU-PRC) and the area under the receiver operating characteristic curve (AU-ROC). Performance assessments were conducted across different input variable sets and lengths, ranging from vital signs only to a combination of vital signs, ICU, and cleanroom-related variables. The results indicated that the model performed promisingly in predicting septic shock in the target patient group. By leveraging a comprehensive range of variables and understanding the temporal dynamics of the data, the model offers valuable insights for septic shock prediction, contributing significantly to early detection and improved management in immunocompromised patients, particularly in clean room environments.