The clinical illness known as sepsis encompasses a wide range of variations. A more targeted approach to treatment may be possible after sepsis subphenotypes have been identified. However, there aren’t enough diagnostic models to pin down the subphenotypes in these people. Thus, in a large sepsis cohort, the researchers intended to recognize potential subphenotypes and evaluate clinical outcomes for subphenotypes. Cluster analysis based on machine learning was conducted on the Medical Information Mart in Intensive Care (MIMIC)-IV database, which contains information about intensive care patients. For this study, they included all adult (>18) patients admitted to the intensive care unit (ICU) with a diagnosis of sepsis within the first 24 hours of their stay. The number of groups was determined using a K-means clustering technique. Multivariable logistic regression models calculated the correlation between sepsis subphenotypes and hospital mortality. There were a total of 8,817 sepsis patients enrolled. Females comprised 38.1% of the total population (3,361 out of 8,817), while the median age was 66.8 (IQR: 55.9-77.1). 11 commonly collected clinical indicators within the first 24 hours of ICU admission were used to divide patients into 2 subphenotypes optimally. Subphenotype B participants had lower body temperature, a lower platelet count, a lower systolic blood pressure, a lower haemoglobin, and a lower PaO2/FiO2 ratio compared to subphenotype A participants. In addition, subphenotype B patients had a considerably greater in-hospital mortality rate than subphenotype A patients (29.4% vs. 8.5%, P<0.001). After accounting for confounding variables, the difference remained statistically significant (adjusted OR 2,214; 95% CI, 1.780-2.754, P<0.001). Using K-means clustering analysis on routinely collected clinical data, 2 sepsis subphenotypes with distinct clinical outcomes were quickly found. With this information, doctors may be able to determine the subphenotype more quickly during patient visits.