<h3>Purpose/Objective(s)</h3> Cervical cancer (CC) is one of the common gynecological tumors. It has been reported that patients with pathological complete response (pCR) after neoadjuvant chemoradiotherapy (nCRT) had better disease-free survival and longer overall survival than patients with pathological partial response (pPR). Radiomics can extract high-throughput information from medical images to quantitatively evaluate lesions. Prediction models based on deep learning may be a useful analytical tool for predicting pathological reactions of CC after radiotherapy. The model performance may be improved by combining clinical information with image information, and the decision model performance may be better by combining manual feature model with deep feature model. Therefore, this study will establish a model to automatically predict treatment response (pCR or non-pCR) for locally advanced cervical cancer (LACC) based on T2-weighted MR images and clinical parameters. <h3>Materials/Methods</h3> A total of 138 patients with LACC from 2010 to 2014 were enrolled in this study, and t2-weighted magnetic resonance images and clinical information of the patients before treatment were collected. Clinical information includes age, stage, pathological type, SCC level, and lymph node status. Combining the domain specific features from computational radiomics system, the abstract features from deep learning network and the clinical parameters, ensemble learning classifier composed of multiple weak classifiers was used to predict pCR. The area under the receiver operating characteristics curve (AUC), accuracy (ACC), true positive rate (TPR), true negative rate (TNR) and precision were used as evaluation metrics. <h3>Results</h3> Among 138 cervical cancer patients, 74 were in the pCR group and 64 were in the non-pCR group. There was no significant difference between the two cohorts in terms of tumor diameter, lymph node and stage before radiotherapy, p=0.787, 0.068, 0.846, respectively. The average AUC, ACC, TPR, TNR and precision of the proposed hybrid model was about 0.80, 0.71, 0.75, 0.66 and 0.71. The AUC values of using clinical parameters, domain specific features, abstract features alone are 0.66, 0.67 and 0.76, respectively. This shows that radiomic features extracted from MRI images are more efficient than clinical parameters in predicting treatment response, and abstract features from deep learning network are the best representation for the LACC. The AUC value of model without ensemble learning is 0.76, which shows ensemble learning classifier is better than single classifier. <h3>Conclusion</h3> The proposed hybrid model can predict well the treatment response of patients with cervical cancer, which might help radiation oncologist to make personalized treatment plans for patients.