Deep Learning Based Model for Automatic Multimorbidity Pattern Prognosis

Abstract

One of the main contemporary health issues in the healthcare system is multimorbidity, which is the co-existence of two or more diseases in the same patient. With the emerging of big data era, the need of analytical tools to extract actionable knowledge from high volumes of medical data is increasing to better understand Multimorbidity and manage health care system. In this work we propose to develop a recurrent neural network based model for the prediction of the multimorbidity sequences patterns. The proposed method is illustrated on real data and evaluated against Multi layer perceptron model and other baseline algorithms. The preliminary results show that recurrent neural model can outperform these baseline algorithmes according to F1 and precision scores which suggest recurrent neural network as a promising method when predicting multimorbid diseases.