A feature-based hybrid recommender system for risk prediction : Machine learning approach (Preprint)

Abstract

In the era of health informatics, exponential growth of information generated by health information systems and healthcare organizations demands expert and intelligent recommendation systems. It has become one of the most valuable tools as it reduces problems such as information overload while selecting and suggesting doctors, hospitals, medicine, diagnosis etc according to patients’ interests.

Recommendation uses Hybrid Filtering as one of the most popular approaches, but the major limitations of this approach are selectivity and data integrity issues.Mostly existing recommendation systems & risk prediction algorithms focus on a single domain, on the other end cross-domain hybrid filtering is able to alleviate the degree of selectivity and data integrity problems to a better extent.

We propose a novel algorithm for recommendation & predictive model using KNN algorithm with machine learning algorithms and artificial intelligence (AI). We find the factors that directly impact on diseases and propose an approach for predicting the correct diagnosis of different diseases. We have constructed a series of models with good reliability for predicting different surgery complications and identified several novel clinical associations. We proposed a novel algorithm pr-KNN to use KNN for prediction and recommendation of diseases

Beside that we compared the performance of our algorithm with other machine algorithms and found better performance of our algorithm, with predictive accuracy improving by +3.61%.

The potential to directly integrate these predictive tools into EHRs may enable personalized medicine and decision-making at the point of care for patient counseling and as a teaching tool.

dataset for the trials of patient attached