The healthcare industry has witnessed an unprecedented integration of unstructured data from diverse sources including medical images, clinical notes, patient records, etc. Traditional methods often struggle to effectively analyze and make sense of this heterogeneous data. In response to the growing integration of diverse unstructured healthcare data, this study introduces a novel approach leveraging a hybrid model that combines Deep Recurrent Neural Network and Butterfly Optimization Algorithm (BOA-DRNN). The proposed study aims is to enhance the analysis of complex and unstructured healthcare data by examining hierarchical representations and temporal dependencies. The proposed mechanism begins with the collection and pre-processing of unstructured healthcare data. The DRNN component examines the hierarchical representations and temporal dependencies within the unstructured data. By capturing patterns in the data, the DRNN facilitates improved predictive analysis and offers enhanced decision-making within healthcare units. In addition, the BOA approach was utilized for optimizing the DRNN training by fine-tuning its hyperparameters. This optimized training of DRNN ensures greater effective analysis of data and provides improved prediction performances. The presented framework was validated with the publicly available unstructured healthcare dataset and the results are examined in terms of accuracy, precision, recall and f-measure.