ABSTRACT
Automating diagnosis coding is a predictive healthcare application that involves building predictive models for large-scale patient data. This chapter proposes an approach for automatically predicting suitable diagnostic procedures given patients' medical records. The problem essentially is viewed as one of neural machine translation, where the sequence is translated from the diagnosis space to the procedure space. The results are presented in the form of Bilingual Evaluation Understudy Score (BLEU) scores, because the performance of the proposed approach as a machine translation problem is to be measured. The method used in this chapter is a novel idea proposed in the field of predictive health care that provides an intuitive insight into the relation between the diagnoses and their corresponding procedures.
