Chikungunya is an abroviral disease caused by the mosquito, which frequently deteriorates into chronic inflammatory rheumatism, including post-Chikungunya arthritis, resulting in permanent incapacity in 2 5 - 6 0% of the infected individuals. This paper addresses the absence of early risk prediction by developing an AI-possessed hybrid ensemble model of XGBoost and Logistic Regression. It was trained using a combined dataset of 1001 real samples provided by Kaggle and 1129 more samples created through probability distributions using clinical summaries. The key features were demographics (age, gender) and symptoms (fever, headache, myalgia, arthralgia, rash, vomiting, swelling, joint stiffness, pain duration). The initial data was skewed (59% with arthritis); a combination of these and SMOTE eliminated this. The model scored 0.73 macro F1score and 0.761 AUC on the merged dataset with preprocessing, non-linear relationship using polynomial feature engineering, SMOTE balancing, and hyperparameter optimization, which was significantly higher than the baseline macro F1-score of 0.46. The SHAP analysis showed that among the clinical evidence arthralgia type, duration of pain, swelling were the best predictors. Robustness was verified by Sensitivity analysis. Results validate the importance of augmented datasets in data-scarce diseases. Weaknesses are synthetic biases and future efforts include deep learning and external clinical validation.