Exploiting Convolutional Neural Network for Risk Prediction with Medical Feature Embedding

In this paper, our focus is on the problems of high dimensionality and temporality. We explore deep neural network models with learned medical feature embedding to deal with these issues. Specifically, we use a multi-layer convolutional neural network (CNN) to parameterize the model and is thus able to capture complex non-linear longitudinal evolution of EHRs. Different from recent proposed deep learning approaches such as stacked auto-encoders [11] and recurrent neural network [5, 2], our model can effectively capture local/short temporal dependency in EHRs, which is beneficial for risk prediction. To account for high dimensionality, instead of using the raw EHR data as the input, we use the embedding medical features in the CNN model. Based on the medical context, each medical event is compressed into a given length vector with medical feature embedding. Similar to the word embedding [8], the event embedding presented in our model holds its natural medical concept. Our initial experiments produce promising results, and demonstrate the effectiveness of both the medical feature embedding and the proposed convolutional neural network in risk prediction on cohorts of congestive heart failure and diabetes patients, compared with several strong baselines.