Temporal Poisson Square Root Graphical Models

Sinong Geng, Zhaobin Kuang, Peggy Peissig, David Page
Proceedings of the 35th International Conference on Machine Learning, PMLR 80:1714-1723, 2018.

Abstract

We propose temporal Poisson square root graphical models (TPSQRs), a generalization of Poisson square root graphical models (PSQRs) specifically designed for modeling longitudinal event data. By estimating the temporal relationships for all possible pairs of event types, TPSQRs can offer a holistic perspective about whether the occurrences of any given event type could excite or inhibit any other type. A TPSQR is learned by estimating a collection of interrelated PSQRs that share the same template parameterization. These PSQRs are estimated jointly in a pseudo-likelihood fashion, where Poisson pseudo-likelihood is used to approximate the original more computationally intensive pseudo-likelihood problem stemming from PSQRs. Theoretically, we demonstrate that under mild assumptions, the Poisson pseudolikelihood approximation is sparsistent for recovering the underlying PSQR. Empirically, we learn TPSQRs from a real-world large-scale electronic health record (EHR) with millions of drug prescription and condition diagnosis events, for adverse drug reaction (ADR) detection. Experimental results demonstrate that the learned TPSQRs can recover ADR signals from the EHR effectively and efficiently.

Cite this Paper

BibTeX
@InProceedings{pmlr-v80-geng18a, title = {Temporal Poisson Square Root Graphical Models}, author = {Geng, Sinong and Kuang, Zhaobin and Peissig, Peggy and Page, David}, booktitle = {Proceedings of the 35th International Conference on Machine Learning}, pages = {1714--1723}, year = {2018}, editor = {Dy, Jennifer and Krause, Andreas}, volume = {80}, series = {Proceedings of Machine Learning Research}, month = {10--15 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v80/geng18a/geng18a.pdf}, url = {https://proceedings.mlr.press/v80/geng18a.html}, abstract = {We propose temporal Poisson square root graphical models (TPSQRs), a generalization of Poisson square root graphical models (PSQRs) specifically designed for modeling longitudinal event data. By estimating the temporal relationships for all possible pairs of event types, TPSQRs can offer a holistic perspective about whether the occurrences of any given event type could excite or inhibit any other type. A TPSQR is learned by estimating a collection of interrelated PSQRs that share the same template parameterization. These PSQRs are estimated jointly in a pseudo-likelihood fashion, where Poisson pseudo-likelihood is used to approximate the original more computationally intensive pseudo-likelihood problem stemming from PSQRs. Theoretically, we demonstrate that under mild assumptions, the Poisson pseudolikelihood approximation is sparsistent for recovering the underlying PSQR. Empirically, we learn TPSQRs from a real-world large-scale electronic health record (EHR) with millions of drug prescription and condition diagnosis events, for adverse drug reaction (ADR) detection. Experimental results demonstrate that the learned TPSQRs can recover ADR signals from the EHR effectively and efficiently.} }
Endnote
%0 Conference Paper %T Temporal Poisson Square Root Graphical Models %A Sinong Geng %A Zhaobin Kuang %A Peggy Peissig %A David Page %B Proceedings of the 35th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2018 %E Jennifer Dy %E Andreas Krause %F pmlr-v80-geng18a %I PMLR %P 1714--1723 %U https://proceedings.mlr.press/v80/geng18a.html %V 80 %X We propose temporal Poisson square root graphical models (TPSQRs), a generalization of Poisson square root graphical models (PSQRs) specifically designed for modeling longitudinal event data. By estimating the temporal relationships for all possible pairs of event types, TPSQRs can offer a holistic perspective about whether the occurrences of any given event type could excite or inhibit any other type. A TPSQR is learned by estimating a collection of interrelated PSQRs that share the same template parameterization. These PSQRs are estimated jointly in a pseudo-likelihood fashion, where Poisson pseudo-likelihood is used to approximate the original more computationally intensive pseudo-likelihood problem stemming from PSQRs. Theoretically, we demonstrate that under mild assumptions, the Poisson pseudolikelihood approximation is sparsistent for recovering the underlying PSQR. Empirically, we learn TPSQRs from a real-world large-scale electronic health record (EHR) with millions of drug prescription and condition diagnosis events, for adverse drug reaction (ADR) detection. Experimental results demonstrate that the learned TPSQRs can recover ADR signals from the EHR effectively and efficiently.
APA
Geng, S., Kuang, Z., Peissig, P. & Page, D.. (2018). Temporal Poisson Square Root Graphical Models. Proceedings of the 35th International Conference on Machine Learning, in Proceedings of Machine Learning Research 80:1714-1723 Available from https://proceedings.mlr.press/v80/geng18a.html.

Related Material