SU‐D‐137‐03

Bayesian Belief Network Based Personalized Adaptive Decision Support to Individualize Response‐Based Adaptive Therapy

Y. Luo, D. Mcshan, Feng Ming Kong, M. Matuszak, M. Schipper, R. Ten Haken

Research output: Contribution to journalArticle

Abstract

Purpose: To develop a personalized adaptive decision support tool to aid in assigning tumor dose before actual dose delivery and then adjusted for individualized treatment during the course of therapy to maximize each patient's overall survival while maintaining a specified population‐based overall toxicity rate. Methods: As an example, a simple personalized adaptive decision support tool was developed based on data prospectively gathered from previously treated lung cancer patients and a Bayesian decision network which seeks an appropriate tumor dose (TD) and mean lung dose (MLD) combination for each individual patient at two time points, before the actual dose delivery AND during the course of treatment. The inputs of the decision support tool are initial/current TD and MLD; identified (via data mining) patient's specifics such as age, gross tumor volume (GTV), lung cancer staging; and then, as weighted by preferences to the outcome of radiation treatment (tradeoffs between predicted survival and potential for complications). Moreover, each patient's individual biomarkers before (IL‐8) and during the treatment (TGF‐β1) are important inputs as well. Satisfaction values associated with these scenarios are evaluated from the pair‐wise comparison based on an analytic hierarchy process, and are determined by the individualized preference during the course of treatment. Results: Each patient's local tumor control and toxicity can be predicted based on these inputs, and a decision alert/suggestion to adjust current TD and MLD based on preference for the patient under consideration is provided as the output of the system. Conclusion: Although our decision support tool is still in its infancy, it is a data‐driven, response‐based and patient‐centered approach, and it can help the physician make more efficient decision and improve the patient's therapeutic gain in the radiation treatment. R01CA142840.

Original languageEnglish (US)
Number of pages1
JournalMedical Physics
Volume40
Issue number6
DOIs
StatePublished - Jan 1 2013

Fingerprint

Neoplasms
Lung
Therapeutics
Lung Neoplasms
Radiation
Data Mining
Survival
Patient Preference
Neoplasm Staging
Tumor Burden
Interleukin-8
Biomarkers
Physicians

ASJC Scopus subject areas

  • Biophysics
  • Radiology Nuclear Medicine and imaging

Cite this

SU‐D‐137‐03 : Bayesian Belief Network Based Personalized Adaptive Decision Support to Individualize Response‐Based Adaptive Therapy. / Luo, Y.; Mcshan, D.; Kong, Feng Ming; Matuszak, M.; Schipper, M.; Ten Haken, R.

In: Medical Physics, Vol. 40, No. 6, 01.01.2013.

Research output: Contribution to journalArticle

Luo, Y. ; Mcshan, D. ; Kong, Feng Ming ; Matuszak, M. ; Schipper, M. ; Ten Haken, R. / SU‐D‐137‐03 : Bayesian Belief Network Based Personalized Adaptive Decision Support to Individualize Response‐Based Adaptive Therapy. In: Medical Physics. 2013 ; Vol. 40, No. 6.
@article{1f0a47ae3167443cb89b3ba2f97edd81,
title = "SU‐D‐137‐03: Bayesian Belief Network Based Personalized Adaptive Decision Support to Individualize Response‐Based Adaptive Therapy",
abstract = "Purpose: To develop a personalized adaptive decision support tool to aid in assigning tumor dose before actual dose delivery and then adjusted for individualized treatment during the course of therapy to maximize each patient's overall survival while maintaining a specified population‐based overall toxicity rate. Methods: As an example, a simple personalized adaptive decision support tool was developed based on data prospectively gathered from previously treated lung cancer patients and a Bayesian decision network which seeks an appropriate tumor dose (TD) and mean lung dose (MLD) combination for each individual patient at two time points, before the actual dose delivery AND during the course of treatment. The inputs of the decision support tool are initial/current TD and MLD; identified (via data mining) patient's specifics such as age, gross tumor volume (GTV), lung cancer staging; and then, as weighted by preferences to the outcome of radiation treatment (tradeoffs between predicted survival and potential for complications). Moreover, each patient's individual biomarkers before (IL‐8) and during the treatment (TGF‐β1) are important inputs as well. Satisfaction values associated with these scenarios are evaluated from the pair‐wise comparison based on an analytic hierarchy process, and are determined by the individualized preference during the course of treatment. Results: Each patient's local tumor control and toxicity can be predicted based on these inputs, and a decision alert/suggestion to adjust current TD and MLD based on preference for the patient under consideration is provided as the output of the system. Conclusion: Although our decision support tool is still in its infancy, it is a data‐driven, response‐based and patient‐centered approach, and it can help the physician make more efficient decision and improve the patient's therapeutic gain in the radiation treatment. R01CA142840.",
author = "Y. Luo and D. Mcshan and Kong, {Feng Ming} and M. Matuszak and M. Schipper and {Ten Haken}, R.",
year = "2013",
month = "1",
day = "1",
doi = "10.1118/1.4814006",
language = "English (US)",
volume = "40",
journal = "Medical Physics",
issn = "0094-2405",
publisher = "AAPM - American Association of Physicists in Medicine",
number = "6",

}

TY - JOUR

T1 - SU‐D‐137‐03

T2 - Bayesian Belief Network Based Personalized Adaptive Decision Support to Individualize Response‐Based Adaptive Therapy

AU - Luo, Y.

AU - Mcshan, D.

AU - Kong, Feng Ming

AU - Matuszak, M.

AU - Schipper, M.

AU - Ten Haken, R.

PY - 2013/1/1

Y1 - 2013/1/1

N2 - Purpose: To develop a personalized adaptive decision support tool to aid in assigning tumor dose before actual dose delivery and then adjusted for individualized treatment during the course of therapy to maximize each patient's overall survival while maintaining a specified population‐based overall toxicity rate. Methods: As an example, a simple personalized adaptive decision support tool was developed based on data prospectively gathered from previously treated lung cancer patients and a Bayesian decision network which seeks an appropriate tumor dose (TD) and mean lung dose (MLD) combination for each individual patient at two time points, before the actual dose delivery AND during the course of treatment. The inputs of the decision support tool are initial/current TD and MLD; identified (via data mining) patient's specifics such as age, gross tumor volume (GTV), lung cancer staging; and then, as weighted by preferences to the outcome of radiation treatment (tradeoffs between predicted survival and potential for complications). Moreover, each patient's individual biomarkers before (IL‐8) and during the treatment (TGF‐β1) are important inputs as well. Satisfaction values associated with these scenarios are evaluated from the pair‐wise comparison based on an analytic hierarchy process, and are determined by the individualized preference during the course of treatment. Results: Each patient's local tumor control and toxicity can be predicted based on these inputs, and a decision alert/suggestion to adjust current TD and MLD based on preference for the patient under consideration is provided as the output of the system. Conclusion: Although our decision support tool is still in its infancy, it is a data‐driven, response‐based and patient‐centered approach, and it can help the physician make more efficient decision and improve the patient's therapeutic gain in the radiation treatment. R01CA142840.

AB - Purpose: To develop a personalized adaptive decision support tool to aid in assigning tumor dose before actual dose delivery and then adjusted for individualized treatment during the course of therapy to maximize each patient's overall survival while maintaining a specified population‐based overall toxicity rate. Methods: As an example, a simple personalized adaptive decision support tool was developed based on data prospectively gathered from previously treated lung cancer patients and a Bayesian decision network which seeks an appropriate tumor dose (TD) and mean lung dose (MLD) combination for each individual patient at two time points, before the actual dose delivery AND during the course of treatment. The inputs of the decision support tool are initial/current TD and MLD; identified (via data mining) patient's specifics such as age, gross tumor volume (GTV), lung cancer staging; and then, as weighted by preferences to the outcome of radiation treatment (tradeoffs between predicted survival and potential for complications). Moreover, each patient's individual biomarkers before (IL‐8) and during the treatment (TGF‐β1) are important inputs as well. Satisfaction values associated with these scenarios are evaluated from the pair‐wise comparison based on an analytic hierarchy process, and are determined by the individualized preference during the course of treatment. Results: Each patient's local tumor control and toxicity can be predicted based on these inputs, and a decision alert/suggestion to adjust current TD and MLD based on preference for the patient under consideration is provided as the output of the system. Conclusion: Although our decision support tool is still in its infancy, it is a data‐driven, response‐based and patient‐centered approach, and it can help the physician make more efficient decision and improve the patient's therapeutic gain in the radiation treatment. R01CA142840.

UR - http://www.scopus.com/inward/record.url?scp=85024824556&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85024824556&partnerID=8YFLogxK

U2 - 10.1118/1.4814006

DO - 10.1118/1.4814006

M3 - Article

VL - 40

JO - Medical Physics

JF - Medical Physics

SN - 0094-2405

IS - 6

ER -