IVP-VAE: Modeling EHR Time Series with Initial Value Problem Solvers

Continuous-time models such as Neural ODEs and Neural Flows have shown promising results in analyzing irregularly sampled time series frequently encountered in electronic health records. Based on these models, time series are typically processed with a hybrid of an initial value problem (IVP) solver...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Xiao, Jingge, Basso, Leonie, Nejdl, Wolfgang, Ganguly, Niloy, Sikdar, Sandipan
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Xiao, Jingge
Basso, Leonie
Nejdl, Wolfgang
Ganguly, Niloy
Sikdar, Sandipan
description Continuous-time models such as Neural ODEs and Neural Flows have shown promising results in analyzing irregularly sampled time series frequently encountered in electronic health records. Based on these models, time series are typically processed with a hybrid of an initial value problem (IVP) solver and a recurrent neural network within the variational autoencoder architecture. Sequentially solving IVPs makes such models computationally less efficient. In this paper, we propose to model time series purely with continuous processes whose state evolution can be approximated directly by IVPs. This eliminates the need for recurrent computation and enables multiple states to evolve in parallel. We further fuse the encoder and decoder with one IVP solver utilizing its invertibility, which leads to fewer parameters and faster convergence. Experiments on three real-world datasets show that the proposed method can systematically outperform its predecessors, achieve state-of-the-art results, and have significant advantages in terms of data efficiency.
doi_str_mv 10.48550/arxiv.2305.06741
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2305_06741</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2305_06741</sourcerecordid><originalsourceid>FETCH-LOGICAL-a671-aabc10c13745499ef7025dd1c554aa121c919179b472bf8c578daa9fbecae2703</originalsourceid><addsrcrecordid>eNotz7FOwzAUQFEvDKjwAUz4BxL8HLuO2aoSaKQiqjbKGr04L8WS0yCnFPh71MJ0tysdxu5ApCrXWjxg_PanVGZCp2JuFFyzp7LeJPWieOSvY0fBH_a8WG155QfiO4qeJv7lj--8PPijx8BrDJ_EN3FsAw18N4YTxemGXfUYJrr974xVz0W1XCXrt5dyuVgnODeQILYOhIPMKK2spd4IqbsOnNYKESQ4CxaMbZWRbZ87bfIO0fYtOSRpRDZj93_bC6P5iH7A-NOcOc2Fk_0CApBDsw</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>IVP-VAE: Modeling EHR Time Series with Initial Value Problem Solvers</title><source>arXiv.org</source><creator>Xiao, Jingge ; Basso, Leonie ; Nejdl, Wolfgang ; Ganguly, Niloy ; Sikdar, Sandipan</creator><creatorcontrib>Xiao, Jingge ; Basso, Leonie ; Nejdl, Wolfgang ; Ganguly, Niloy ; Sikdar, Sandipan</creatorcontrib><description>Continuous-time models such as Neural ODEs and Neural Flows have shown promising results in analyzing irregularly sampled time series frequently encountered in electronic health records. Based on these models, time series are typically processed with a hybrid of an initial value problem (IVP) solver and a recurrent neural network within the variational autoencoder architecture. Sequentially solving IVPs makes such models computationally less efficient. In this paper, we propose to model time series purely with continuous processes whose state evolution can be approximated directly by IVPs. This eliminates the need for recurrent computation and enables multiple states to evolve in parallel. We further fuse the encoder and decoder with one IVP solver utilizing its invertibility, which leads to fewer parameters and faster convergence. Experiments on three real-world datasets show that the proposed method can systematically outperform its predecessors, achieve state-of-the-art results, and have significant advantages in terms of data efficiency.</description><identifier>DOI: 10.48550/arxiv.2305.06741</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Learning</subject><creationdate>2023-05</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2305.06741$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2305.06741$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Xiao, Jingge</creatorcontrib><creatorcontrib>Basso, Leonie</creatorcontrib><creatorcontrib>Nejdl, Wolfgang</creatorcontrib><creatorcontrib>Ganguly, Niloy</creatorcontrib><creatorcontrib>Sikdar, Sandipan</creatorcontrib><title>IVP-VAE: Modeling EHR Time Series with Initial Value Problem Solvers</title><description>Continuous-time models such as Neural ODEs and Neural Flows have shown promising results in analyzing irregularly sampled time series frequently encountered in electronic health records. Based on these models, time series are typically processed with a hybrid of an initial value problem (IVP) solver and a recurrent neural network within the variational autoencoder architecture. Sequentially solving IVPs makes such models computationally less efficient. In this paper, we propose to model time series purely with continuous processes whose state evolution can be approximated directly by IVPs. This eliminates the need for recurrent computation and enables multiple states to evolve in parallel. We further fuse the encoder and decoder with one IVP solver utilizing its invertibility, which leads to fewer parameters and faster convergence. Experiments on three real-world datasets show that the proposed method can systematically outperform its predecessors, achieve state-of-the-art results, and have significant advantages in terms of data efficiency.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz7FOwzAUQFEvDKjwAUz4BxL8HLuO2aoSaKQiqjbKGr04L8WS0yCnFPh71MJ0tysdxu5ApCrXWjxg_PanVGZCp2JuFFyzp7LeJPWieOSvY0fBH_a8WG155QfiO4qeJv7lj--8PPijx8BrDJ_EN3FsAw18N4YTxemGXfUYJrr974xVz0W1XCXrt5dyuVgnODeQILYOhIPMKK2spd4IqbsOnNYKESQ4CxaMbZWRbZ87bfIO0fYtOSRpRDZj93_bC6P5iH7A-NOcOc2Fk_0CApBDsw</recordid><startdate>20230511</startdate><enddate>20230511</enddate><creator>Xiao, Jingge</creator><creator>Basso, Leonie</creator><creator>Nejdl, Wolfgang</creator><creator>Ganguly, Niloy</creator><creator>Sikdar, Sandipan</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230511</creationdate><title>IVP-VAE: Modeling EHR Time Series with Initial Value Problem Solvers</title><author>Xiao, Jingge ; Basso, Leonie ; Nejdl, Wolfgang ; Ganguly, Niloy ; Sikdar, Sandipan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a671-aabc10c13745499ef7025dd1c554aa121c919179b472bf8c578daa9fbecae2703</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Xiao, Jingge</creatorcontrib><creatorcontrib>Basso, Leonie</creatorcontrib><creatorcontrib>Nejdl, Wolfgang</creatorcontrib><creatorcontrib>Ganguly, Niloy</creatorcontrib><creatorcontrib>Sikdar, Sandipan</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Xiao, Jingge</au><au>Basso, Leonie</au><au>Nejdl, Wolfgang</au><au>Ganguly, Niloy</au><au>Sikdar, Sandipan</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>IVP-VAE: Modeling EHR Time Series with Initial Value Problem Solvers</atitle><date>2023-05-11</date><risdate>2023</risdate><abstract>Continuous-time models such as Neural ODEs and Neural Flows have shown promising results in analyzing irregularly sampled time series frequently encountered in electronic health records. Based on these models, time series are typically processed with a hybrid of an initial value problem (IVP) solver and a recurrent neural network within the variational autoencoder architecture. Sequentially solving IVPs makes such models computationally less efficient. In this paper, we propose to model time series purely with continuous processes whose state evolution can be approximated directly by IVPs. This eliminates the need for recurrent computation and enables multiple states to evolve in parallel. We further fuse the encoder and decoder with one IVP solver utilizing its invertibility, which leads to fewer parameters and faster convergence. Experiments on three real-world datasets show that the proposed method can systematically outperform its predecessors, achieve state-of-the-art results, and have significant advantages in terms of data efficiency.</abstract><doi>10.48550/arxiv.2305.06741</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2305.06741
ispartof
issn
language eng
recordid cdi_arxiv_primary_2305_06741
source arXiv.org
subjects Computer Science - Artificial Intelligence
Computer Science - Learning
title IVP-VAE: Modeling EHR Time Series with Initial Value Problem Solvers
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-24T21%3A40%3A56IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=IVP-VAE:%20Modeling%20EHR%20Time%20Series%20with%20Initial%20Value%20Problem%20Solvers&rft.au=Xiao,%20Jingge&rft.date=2023-05-11&rft_id=info:doi/10.48550/arxiv.2305.06741&rft_dat=%3Carxiv_GOX%3E2305_06741%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true