Healthcare analytics systems based on artificial intelligence (AI) have been developed to enable early risk assessment and personalized patient treatment, thanks to the increasing rise of electronic health records (EHRs). Pressure ulcers (PUs) are a major hospital-acquired complication that require timely identification and preventive interventions. This study proposes an AI-enabled healthcare analytics framework for personalized patient risk assessment and pressure ulcer prediction using the MIMIC-IV clinical dataset. The proposed framework includes data extraction, preprocessing, missing value handling, feature engineering, normalization, and feature selection using Random Forest-based feature importance analysis. Clinically relevant features, including demographic characteristics, ICU-related parameters, vital signs, laboratory measurements, Braden scale factors, and diagnosis information, were utilized for predictive modeling. Decision Tree (DT) and Light Gradient Boosting Machine (LightGBM) are two ML models that were fine-tuned using hyperparameter tweaking to make better risk predictions. The LightGBM model was analysed using explainable AI based on SHAP (SHapley Additive exPlanations) to identify the clinical factors that contribute to pressure ulcer risk prediction. The experimental results demonstrate that the proposed LightGBM model outperformed traditional machine learning approaches, achieving 96.1% accuracy (ACC), 94.8% precision (PRE), 91.9% recall (REC), 93.1% F1-score (F1), and 99.04% ROC-AUC on average. To improve model transparency, the SHAP study identified important clinical variables linked to pressure ulcer formation. The suggested architecture offers a dependable method for pressure ulcer prediction and may find use in intelligent clinical decision-support systems and real-time healthcare monitoring.