Critically ill intensive care unit (ICU) patients frequently experience acute insulin resistance (low insulin sensitivity) manifesting as stress-induced hyperglycemia and hyperinsulinemia, especially in the early stages of the treatment. High inter/intra-patient variability makes glycemic control difficult. Stochastic TARgeted (STAR) a modelbased glycemic control, directly manages this variability using model-based insulin sensitivity (SI) and a second model of its variability. Early occurrence of insulin resistance and hyperglycemia may need a special (customized) model-based control designed only for the early phase of patient treatment. This study analyses insulin sensitivity and blood glucose levels of ICU patients from 3 different cohorts and compares the first 24h of the treatment and the rest of the treatment in order to assess the differences. Using clinical data from 717 patients treated with STAR in three independent cohorts (Hungary, New Zealand, and Malaysia), insulin sensitivity and blood glucose are compared at first between the first 24h and the rest of the treatment, then the first 24h and the successive treatment days. Results show that insulin sensitivity is lower in the first 24h compared to the rest of the treatment and in the first 24h compared to the five successive days. The differences were noticeable in the Hungarian and New Zealand cohort but not for the Malaysian cohort. Blood glucose levels were higher in all cohorts in the first 24h compared to the rest of the treatment time and in the first 24h compared to the five successive days. Patients in the early stages of ICU have low insulin sensitivity and high blood glucose levels, as expected, given the stress response physiology. Given the results, this study initiates the idea of Implementing a customized model-based control designed only for the early phase of patient treatment that can effectively handle patients' hyperglycemia and insulin resistance and create a space for further development.