Unsupervised patient representations from clinical notes with interpretable classification decisions

We have two main contributions in this work: 1. We explore the usage of a stacked denoising autoencoder, and a paragraph vector model to learn task-independent dense patient representations directly from clinical notes. We evaluate these representations by using them as features in multiple supervised setups, and compare their performance with those of sparse representations. 2. To understand and interpret the representations, we explore the best encoded features within the patient representations obtained from the autoencoder model. Further, we calculate the significance of the input features of the trained classifiers when we use these pretrained representations as input.

[1]  Jianying Hu,et al.  Towards Personalized Medicine: Leveraging Patient Similarity and Drug Similarity Analytics , 2014, AMIA Joint Summits on Translational Science proceedings. AMIA Joint Summits on Translational Science.

[2]  Quoc V. Le,et al.  Distributed Representations of Sentences and Documents , 2014, ICML.

[3]  Fei Wang,et al.  Medical prognosis based on patient similarity and expert feedback , 2012, Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012).

[4]  Anita Burgun-Parenthoine,et al.  Finding patients using similarity measures in a rare diseases-oriented clinical data warehouse: Dr. Warehouse and the needle in the needle stack , 2017, J. Biomed. Informatics.

[5]  Pascal Vincent,et al.  Stacked Denoising Autoencoders: Learning Useful Representations in a Deep Network with a Local Denoising Criterion , 2010, J. Mach. Learn. Res..

[6]  Georgiana Dinu,et al.  Don’t count, predict! A systematic comparison of context-counting vs. context-predicting semantic vectors , 2014, ACL.

[7]  Katherine E Henson,et al.  Risk of Suicide After Cancer Diagnosis in England , 2018, JAMA psychiatry.

[8]  Florian Schmidt,et al.  Neural Document Embeddings for Intensive Care Patient Mortality Prediction , 2016, NIPS 2016.

[9]  Li Li,et al.  Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records , 2016, Scientific Reports.

[10]  Yoshua Bengio,et al.  Random Search for Hyper-Parameter Optimization , 2012, J. Mach. Learn. Res..

[11]  Andries P. Engelbrecht,et al.  Feature Extraction from Feedforward Neural Networks using Sensitivity Analysis , 1998 .

[12]  Antal van den Bosch.  Ucto: Unicode Tokeniser , 2012 .

[13]  Peter Szolovits,et al.  The Use of Autoencoders for Discovering Patient Phenotypes , 2017, ArXiv.

[14]  W. Beyer.  CRC Standard Probability And Statistics Tables and Formulae , 1990 .

[15]  Benjamin S. Glicksberg,et al.  Identification of type 2 diabetes subgroups through topological analysis of patient similarity , 2015, Science Translational Medicine.

[16]  Hong Yu,et al.  Learning for Biomedical Information Extraction: Methodological Review of Recent Advances , 2016, ArXiv.

[17]  Yu Tian,et al.  An Electronic Medical Record System with Treatment Recommendations Based on Patient Similarity , 2015, Journal of Medical Systems.

[18]  Yohan Jo,et al.  Combining LSTM and Latent Topic Modeling for Mortality Prediction , 2017, ArXiv.

[19]  Kenney Ng,et al.  Personalized Predictive Modeling and Risk Factor Identification using Patient Similarity , 2015, AMIA Joint Summits on Translational Science proceedings. AMIA Joint Summits on Translational Science.

[20]  R. Sharan,et al.  A method for inferring medical diagnoses from patient similarities , 2013, BMC Medicine.

[21]  S. T. Buckland,et al.  Computer-Intensive Methods for Testing Hypotheses. , 1990 .

[22]  Peter Szolovits,et al.  MIMIC-III, a freely accessible critical care database , 2016, Scientific Data.