Learning covariate relations in disease progression models using symbolic neural networks
Sundell, Jesper LU ; Wahlquist, Ylva LU ; Kjellsson, Maria ; Karlsson, Mats and Soltesz, Kristian LU
(2026)
In CPT: Pharmacometrics and Systems Pharmacology
15(3).
- Abstract
- Covariate modeling provides individual predictions of outcomes by disease progression models. Current methodology for mapping covariates onto model parameters is limited by predefined parametric functions which can result in inadequate covariate selection and biased predictions by the final model. Furthermore, present methodology scales poorly to high-dimensional data due to combinatorial limitations. In the present study, a novel method for automation of covariate model identification in disease progression models is described. Symbolic neural networks are used to simultaneously identify the parametric covariate functions and optimize model parameters of a Markov chain. By stepwise pruning of initially fully connected dense symbolic... (More)
- Covariate modeling provides individual predictions of outcomes by disease progression models. Current methodology for mapping covariates onto model parameters is limited by predefined parametric functions which can result in inadequate covariate selection and biased predictions by the final model. Furthermore, present methodology scales poorly to high-dimensional data due to combinatorial limitations. In the present study, a novel method for automation of covariate model identification in disease progression models is described. Symbolic neural networks are used to simultaneously identify the parametric covariate functions and optimize model parameters of a Markov chain. By stepwise pruning of initially fully connected dense symbolic networks, humanly readable functions representing the covariate relations are produced. The presented methodology is applied to a dataset containing disease progression observations for type 2 diabetes patients. Although utilizing fewer covariates, the resulting model demonstrates predictive performance similar to that of a model which was developed on the same data using state-of-the-art modeling methodology. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/8586624a-b90c-475c-ab0e-ea16e9dd9a22
- author
- Sundell, Jesper
LU
; Wahlquist, Ylva
LU
; Kjellsson, Maria
; Karlsson, Mats
and Soltesz, Kristian
LU
- organization
- publishing date
- 2026
- type
- Contribution to journal
- publication status
- published
- subject
- in
- CPT: Pharmacometrics and Systems Pharmacology
- volume
- 15
- issue
- 3
- pages
- 9 pages
- publisher
- American Society for Clinical Pharmacology and Therapeutics
- external identifiers
-
- scopus:105032238687
- pmid:41806166
- ISSN
- 2163-8306
- DOI
- 10.1002/psp4.70214
- project
- Learning pharmacometric model structures from data
- language
- English
- LU publication?
- yes
- id
- 8586624a-b90c-475c-ab0e-ea16e9dd9a22
- date added to LUP
- 2026-02-03 07:05:39
- date last changed
- 2026-05-22 03:00:07
@article{8586624a-b90c-475c-ab0e-ea16e9dd9a22,
abstract = {{Covariate modeling provides individual predictions of outcomes by disease progression models. Current methodology for mapping covariates onto model parameters is limited by predefined parametric functions which can result in inadequate covariate selection and biased predictions by the final model. Furthermore, present methodology scales poorly to high-dimensional data due to combinatorial limitations. In the present study, a novel method for automation of covariate model identification in disease progression models is described. Symbolic neural networks are used to simultaneously identify the parametric covariate functions and optimize model parameters of a Markov chain. By stepwise pruning of initially fully connected dense symbolic networks, humanly readable functions representing the covariate relations are produced. The presented methodology is applied to a dataset containing disease progression observations for type 2 diabetes patients. Although utilizing fewer covariates, the resulting model demonstrates predictive performance similar to that of a model which was developed on the same data using state-of-the-art modeling methodology.}},
author = {{Sundell, Jesper and Wahlquist, Ylva and Kjellsson, Maria and Karlsson, Mats and Soltesz, Kristian}},
issn = {{2163-8306}},
language = {{eng}},
number = {{3}},
publisher = {{American Society for Clinical Pharmacology and Therapeutics}},
series = {{CPT: Pharmacometrics and Systems Pharmacology}},
title = {{Learning covariate relations in disease progression models using symbolic neural networks}},
url = {{http://dx.doi.org/10.1002/psp4.70214}},
doi = {{10.1002/psp4.70214}},
volume = {{15}},
year = {{2026}},
}