Entity resolution for symptom vs disease for top-K treatments

摘要

The sufficient information on the web of data calls for an efficient entity resolution techniques in biomedical records, symptom vs. disease where a particular symptom, subjected to ambiguity. There may be several terms that refer to the same symptom. Thus, Entity Resolution becomes an essential task to identify a particular disease, for a given symptom. This work aims at suggesting the best alternate treatments to the health care professionals based on the patient's disease. A hybrid recommender system that recommends alternate treatments to the healthcare professionals based on their patient's disease, symptoms, age, and gender is designed and developed. Nowadays, there are new on-going treatments which are much successful know from the clinical trials for a particular disease. The content-based filtering, find the different treatments that are available for user's disease based on their outcome obtained by sentiment analysis. The collaborative filtering uses the similarity measure to find the similarity between the user and the patients by considering their age, gender, location, symptoms, and diseases. The treatments obtained from both these modules are then ranked by assigning a score based on their effectiveness and side effects. Finally, the top-k treatments for the disease are recommended to the health care professionals.