Diagnosis Prediction using Ensemble Learning

Abstract: The data in healthcare has increased in volume, intricacy and comprehensiveness. This growth leads to extensive application of artificial intelligence and machine learning in the healthcare sector. This study aims to examine the application of deep learning models and ensemble learning in diagnosis prediction. We apply Natural Language Processing techniques on medical notes to predict diagnosis. Real-life healthcare datasets, like MIMIC-2, contain tables with medical notes which can be pre-processed and used to train ML models. This paper presents an analysis of a diagnosis prediction algorithm. This facilitates the creation of autonomous medical systems which can be used to aid or act in place of healthcare professionals.

Keywords: MIMIC 2, EHR, Clinical notes, NLP, Bidirectional LSTM, BERT, Ensemble learning.