Abstract
Patients with type 2 diabetes have higher risk of post operative complications i.e. surgical wound infections and sepsis. If not treated with proper treatment that may lead to prolonged hospitalization and also increase morbidity rate. Identification of high risk patients diabetic patients can improve surgical outcomes and optimize clinical management.
This study aim to design and develop a model based on random forest algorithm to predict the high risk patients with post operstive complications.
A retrospective analysis was conducted using the dataset collected from local hospitals of hyderabad. The dataset developed included demographic information, labortory indicators and surgical characteristics i.e. procedure specialty, operative duration and anesthesia type. The dataset was divided into 80/20 ratio in which 80% was for training set and 20% was reserved for testing to evaluate the model performance.
This study presents the development and evaluation of a predictive model for identifying postoperative complications, specifically wound infection and sepsis. The model’s performance was assessed using standard classification metrics, including accuracy, precision, recall, F1-score, and receiver operating characteristic (ROC) analysis. The results demonstrate strong predictive capability, with accuracy values of 93.8% for wound infection and 94% for sepsis, indicating a high rate of correct classifications. Precision scores of 89.2% and 87.1% for wound infection and sepsis, respectively, reflect the model’s effectiveness in producing reliable positive predictions while minimizing false positives. Similarly, recall values of 90.1% for wound infection and 89.6% for sepsis highlight the model’s ability to correctly identify the majority of actual positive cases. The F1-scores, recorded at 90.5% for wound infection and 89.1% for sepsis, further confirm a balanced performance between precision and recall. The findings highlight the effectiveness of accurate prdiction of postoperative complications using random forest approach.