Patient Similarity Using Population Statistics and Multiple Kernel Learning

Bryan Conroy, Minnan Xu-Wilson, Asif Rahman
; Proceedings of the 2nd Machine Learning for Healthcare Conference, PMLR 68:191-203, 2017.

Abstract

We present a multiple kernel learning framework to learn similarity functions that compare physiological state between patients. A powerful ensemble kernel is learned from many base kernels evaluated on individual features. Our proposed framework captures two aspects of patient similarity: that patient similarity should be dependent on clinical context and that similarity should be modulated by the frequency and specificity of individual feature values. We validate our model on ICU data to predict hemodynamic instability and present analyses on using the patient similarity function to construct personalized cohorts. Our experiments show that the statistical properties learned by the kernels functions based on feature population distributions are significantly more predictive than naive stationary kernels (e.g. RBFs). Population-based kernels outperform RBF’s in identifying patient cohorts based on abnormality of their vitals and lab measurements and at predicting mortality.

Cite this Paper


BibTeX
@InProceedings{pmlr-v68-conroy17a, title = {Patient Similarity Using Population Statistics and Multiple Kernel Learning}, author = {Bryan Conroy and Minnan Xu-Wilson and Asif Rahman}, booktitle = {Proceedings of the 2nd Machine Learning for Healthcare Conference}, pages = {191--203}, year = {2017}, editor = {Finale Doshi-Velez and Jim Fackler and David Kale and Rajesh Ranganath and Byron Wallace and Jenna Wiens}, volume = {68}, series = {Proceedings of Machine Learning Research}, address = {Boston, Massachusetts}, month = {18--19 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v68/conroy17a/conroy17a.pdf}, url = {http://proceedings.mlr.press/v68/conroy17a.html}, abstract = {We present a multiple kernel learning framework to learn similarity functions that compare physiological state between patients. A powerful ensemble kernel is learned from many base kernels evaluated on individual features. Our proposed framework captures two aspects of patient similarity: that patient similarity should be dependent on clinical context and that similarity should be modulated by the frequency and specificity of individual feature values. We validate our model on ICU data to predict hemodynamic instability and present analyses on using the patient similarity function to construct personalized cohorts. Our experiments show that the statistical properties learned by the kernels functions based on feature population distributions are significantly more predictive than naive stationary kernels (e.g. RBFs). Population-based kernels outperform RBF’s in identifying patient cohorts based on abnormality of their vitals and lab measurements and at predicting mortality.} }
Endnote
%0 Conference Paper %T Patient Similarity Using Population Statistics and Multiple Kernel Learning %A Bryan Conroy %A Minnan Xu-Wilson %A Asif Rahman %B Proceedings of the 2nd Machine Learning for Healthcare Conference %C Proceedings of Machine Learning Research %D 2017 %E Finale Doshi-Velez %E Jim Fackler %E David Kale %E Rajesh Ranganath %E Byron Wallace %E Jenna Wiens %F pmlr-v68-conroy17a %I PMLR %J Proceedings of Machine Learning Research %P 191--203 %U http://proceedings.mlr.press %V 68 %W PMLR %X We present a multiple kernel learning framework to learn similarity functions that compare physiological state between patients. A powerful ensemble kernel is learned from many base kernels evaluated on individual features. Our proposed framework captures two aspects of patient similarity: that patient similarity should be dependent on clinical context and that similarity should be modulated by the frequency and specificity of individual feature values. We validate our model on ICU data to predict hemodynamic instability and present analyses on using the patient similarity function to construct personalized cohorts. Our experiments show that the statistical properties learned by the kernels functions based on feature population distributions are significantly more predictive than naive stationary kernels (e.g. RBFs). Population-based kernels outperform RBF’s in identifying patient cohorts based on abnormality of their vitals and lab measurements and at predicting mortality.
APA
Conroy, B., Xu-Wilson, M. & Rahman, A.. (2017). Patient Similarity Using Population Statistics and Multiple Kernel Learning. Proceedings of the 2nd Machine Learning for Healthcare Conference, in PMLR 68:191-203

Related Material