Mechanistic machine learning: How data assimilation leverages physiologic knowledge using Bayesian inference to forecast the future, infer the present, and phenotype

David J. Albers, Matthew E. Levine, Andrew Stuart, Lena Mamykina, Bruce Gluckman, George Hripcsak

Research output: Contribution to journalReview article

1 Citation (Scopus)

Abstract

We introduce data assimilation as a computational method that uses machine learning to combine data with human knowledge in the form of mechanistic models in order to forecast future states, to impute missing data from the past by smoothing, and to infer measurable and unmeasurable quantities that represent clinically and scientifically important phenotypes. We demonstrate the advantages it affords in the context of type 2 diabetes by showing how data assimilation can be used to forecast future glucose values, to impute previously missing glucose values, and to infer type 2 diabetes phenotypes. At the heart of data assimilation is the mechanistic model, here an endocrine model. Such models can vary in complexity, contain testable hypotheses about important mechanics that govern the system (eg, nutrition's effect on glucose), and, as such, constrain the model space, allowing for accurate estimation using very little data.

Original languageEnglish (US)
Pages (from-to)1392-1401
Number of pages10
JournalJournal of the American Medical Informatics Association
Volume25
Issue number10
DOIs
StatePublished - Oct 1 2018

Fingerprint

Phenotype
Glucose
Type 2 Diabetes Mellitus
Space Simulation
Mechanics
Machine Learning

All Science Journal Classification (ASJC) codes

  • Health Informatics

Cite this

@article{4e46548e87894709b991a4e0ab04671c,
title = "Mechanistic machine learning: How data assimilation leverages physiologic knowledge using Bayesian inference to forecast the future, infer the present, and phenotype",
abstract = "We introduce data assimilation as a computational method that uses machine learning to combine data with human knowledge in the form of mechanistic models in order to forecast future states, to impute missing data from the past by smoothing, and to infer measurable and unmeasurable quantities that represent clinically and scientifically important phenotypes. We demonstrate the advantages it affords in the context of type 2 diabetes by showing how data assimilation can be used to forecast future glucose values, to impute previously missing glucose values, and to infer type 2 diabetes phenotypes. At the heart of data assimilation is the mechanistic model, here an endocrine model. Such models can vary in complexity, contain testable hypotheses about important mechanics that govern the system (eg, nutrition's effect on glucose), and, as such, constrain the model space, allowing for accurate estimation using very little data.",
author = "Albers, {David J.} and Levine, {Matthew E.} and Andrew Stuart and Lena Mamykina and Bruce Gluckman and George Hripcsak",
year = "2018",
month = "10",
day = "1",
doi = "10.1093/jamia/ocy106",
language = "English (US)",
volume = "25",
pages = "1392--1401",
journal = "Journal of the American Medical Informatics Association : JAMIA",
issn = "1067-5027",
publisher = "Oxford University Press",
number = "10",

}

Mechanistic machine learning : How data assimilation leverages physiologic knowledge using Bayesian inference to forecast the future, infer the present, and phenotype. / Albers, David J.; Levine, Matthew E.; Stuart, Andrew; Mamykina, Lena; Gluckman, Bruce; Hripcsak, George.

In: Journal of the American Medical Informatics Association, Vol. 25, No. 10, 01.10.2018, p. 1392-1401.

Research output: Contribution to journalReview article

TY - JOUR

T1 - Mechanistic machine learning

T2 - How data assimilation leverages physiologic knowledge using Bayesian inference to forecast the future, infer the present, and phenotype

AU - Albers, David J.

AU - Levine, Matthew E.

AU - Stuart, Andrew

AU - Mamykina, Lena

AU - Gluckman, Bruce

AU - Hripcsak, George

PY - 2018/10/1

Y1 - 2018/10/1

N2 - We introduce data assimilation as a computational method that uses machine learning to combine data with human knowledge in the form of mechanistic models in order to forecast future states, to impute missing data from the past by smoothing, and to infer measurable and unmeasurable quantities that represent clinically and scientifically important phenotypes. We demonstrate the advantages it affords in the context of type 2 diabetes by showing how data assimilation can be used to forecast future glucose values, to impute previously missing glucose values, and to infer type 2 diabetes phenotypes. At the heart of data assimilation is the mechanistic model, here an endocrine model. Such models can vary in complexity, contain testable hypotheses about important mechanics that govern the system (eg, nutrition's effect on glucose), and, as such, constrain the model space, allowing for accurate estimation using very little data.

AB - We introduce data assimilation as a computational method that uses machine learning to combine data with human knowledge in the form of mechanistic models in order to forecast future states, to impute missing data from the past by smoothing, and to infer measurable and unmeasurable quantities that represent clinically and scientifically important phenotypes. We demonstrate the advantages it affords in the context of type 2 diabetes by showing how data assimilation can be used to forecast future glucose values, to impute previously missing glucose values, and to infer type 2 diabetes phenotypes. At the heart of data assimilation is the mechanistic model, here an endocrine model. Such models can vary in complexity, contain testable hypotheses about important mechanics that govern the system (eg, nutrition's effect on glucose), and, as such, constrain the model space, allowing for accurate estimation using very little data.

UR - http://www.scopus.com/inward/record.url?scp=85054889195&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054889195&partnerID=8YFLogxK

U2 - 10.1093/jamia/ocy106

DO - 10.1093/jamia/ocy106

M3 - Review article

C2 - 30312445

AN - SCOPUS:85054889195

VL - 25

SP - 1392

EP - 1401

JO - Journal of the American Medical Informatics Association : JAMIA

JF - Journal of the American Medical Informatics Association : JAMIA

SN - 1067-5027

IS - 10

ER -