Safe Personalized Treatment Recommendation System for Multimorbidity

Please enable javascript in your browser.

Ghasempour, Elahe | 2020

445 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 53189 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Habibi, Jafar
Abstract:
Multimorbidity in an individual is a typical and increasing condition and a major challenge in healthcare. The goal of treatment recommendation for such patients is to decide the most effective combination of treatments. Doctors typically prescribe medication based on their intuition and experience. However, due to knowledge gaps or unintended biases, often times these clinical decisions can be sub-optimal. Broad adoption and usage of electronic health records (EHRs) in the last decade has opened up a great opportunity to leverage healthcare data to improve clinical decisions.In this approach we use EHR and drug-drug interaction (DDI) information to prescribe medications. We use multiple neural networks, DDI relations and patient related data including demographic information, disease codes, lab results and used medications to dynamically prescribe safe medication for a patient. Effective usage of patient related data in this work resulted in about 20 percent more accuracy for the prescribed medications
Keywords:
Electronic Health ; Medication Recommendation ; Multimorbidity ; Drug-Drug Interaction ; Electronic Health Records (EHRs)

No TOC