Forecasting the duration of hospitalization in the Intensive Care Unit (ICU) is a crucial component in providing quality healthcare to patients while diminishing costs. Various clinical factors may impact how long a patient stays, and machine learning approaches such as regression models have shown impressive ability in making precise predictions. However, despite numerous studies on length of stay (LOS) in various healthcare facilities, none have been able to accurately predict this outcome based solely on a limited number of features. The main aim of this article is to predict LOS in General Intensive Care Unit (GICU) ward at Hospital Canselor Tuanku Muhriz (HCTM) for operation theatre (OT) scheduling purpose. A study was done on GICU dataset from 2015 to 2020, involving of 3165 cases with GICU admissions in HCTM. Demographic and clinical data of these subject are analyzed, which comprising patient’s age, gender, race, primary team (specialty), and diagnosis. Features that influenced the LOS at the GICU were also studied by applying correlation analysis method. Four out of the five features were significant (p < 0.05) with LOS which are age, race, gender and primary team (specialty). In the prediction model, we considered the regression output of ±2 days was true and it was found that an Artificial Neural Network (ANN) was able to predict patient LOS with a Root Mean Square Error (RMSE) of 1.65 days, which is comparable to other studies despite utilizing only four features. These findings are significant for hospitals seeking to reduce surgery cancellation rates by incorporating LOS as one of the parameters in their Operating Theater (OT) scheduling system.