Cognitive AI for Precision Medicine: Integrative Drug Recommendation Through Intelligent Architectures
K. Kavitha, M Sivaraman, Yamasani Srinivas Reddy, N. Amsaveni, M. Umamaheswari, R. Ganga Devi · 2025
A Drug Recommendation System (DRS) is an advanced technology for recommending appropriate medicines in healthcare, based on patient-reported symptoms. This helps improve care and treatment outcomes. It also predicts how patients will respond to drugs and helps avoid side effects caused by recommended medications. Machine learning (ML) recommends the most suitable medications, by analyzing patients’ health data such as health records, medical images (for example, scan reports) and other sensor data generated by smart wearable devices. Conventional machine learning models enable DRS systems to provide straightforward predictions that are easy to understand. However, due to disorganized information on symptoms, heavy feature engineering is required. As a result, these models fail to grasp the full context of data. On the other hand, deep learning (DL) classifiers can understand raw text on their own, allowing them to extract more detailed information about symptoms. However, this comes at a higher computational cost due to the presence of many layers. To integrate the strengths of ML and DL and to address these research challenges, a new DRS framework is designed using both XGBoost, LightGBM and CNN. This hybrid combination of ML and DL models outperforms other approaches in terms of the evaluation accuracy achieving 94.8%. As a result, the proposed model provides better recommendations on medications.