Browsing by Author "EGBI, Alilu Grace"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Embargo
    Development of a Drug Recommender System
    (The Department of Computer Science and Engineering, Faculty of Technology, Obafeim Awolowo University., 2024) EGBI, Alilu Grace
    This study elicited and analysed data on patients, drugs and disease. The study then designed a drug recommender model, implemented the model and tested the performance of the system. These were with a view to developing a drug recommender system that recommends appropriate drug(s) for the treatment of an ailment. Patients’ data were elicited from Medical Information Mart for Intensive Care – IV (MIMIC-IV), a deidentified clinical data of patient admitted in ICU at Beth Isreal Deaconess Medical Center (BIDMC). Drugs data were elicited from Drugs.com. Disease treatment knowledge were elicited from the guidelines on the treatment of Peptic Ulcer Disease (PUD) provided by the Japanese Association of Gastroenterologists. These data acquired were analysed using various functions in the Pandas library. The model for the recommender system was designed based on the Hybrid recommendation approach by combining clustering algorithm, Collaborative filtering approach (CF) and Knowledge-Based filtering approach (KBF). The factors that were considered for recommending appropriate drugs were age of patient, gender of patient, body weight, allergies and drug interactions. The model designed was implemented using the Python Programming Language version 3.6.3 with Flask framework for web development and Visual Studio Code as the Integrated Development Environment (IDE). And the performance of the system was evaluated using Precision, Recall, and Root Mean Squared Error (RMSE). The evaluation was carried out in two phases; Firstly, the CF component was evaluated by splitting the dataset from MIMIV-IV into 70% (60,018) train set and 30% (25,722) test set. Secondly, the KBF component was evaluated using 30 different cases. The evaluation for this was computed manually by comparing the recommendation results from the system with that of an expert. For the CF aspect of the DRS, the system had a precision score of 85.48%, a recall score of 85.58% and a RMSE score of 0.74. The precision result shows that the system has an 85.48% bability in making relevant recommendations. The recall score shows that the system has an 85.58% ability in recommending relevant drugs from all available relevant drugs. The RMSE score of 0.74 shows that the recommended drugs are far from the actual drugs prescribed. For the KBF aspect of the DRS, the system achieved a Precision of 77%, a recall of 83% and a RMSE of 0.24. The system’s Precision and Recall scores were lower when the KBF was added. This study concluded that the addition of the KBF reduced the error rate between actual recommendations and predicted recommendations. So, the system had a high ability in recommending appropriate drugs for PUD.