國立虎尾科技大學 |

Statistical Machine Learning Methodology for Individualized Treatment Rule Estimation in Precision Medicine.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Statistical Machine Learning Methodology for Individualized Treatment Rule Estimation in Precision Medicine./
Author:	Shah, Kushal S.
Description:	1 online resource (124 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-02, Section: B.
Contained By:	Dissertations Abstracts International85-02B.
Subject:	Biostatistics. -
Online resource:	click for full text (PQDT)
ISBN:	9798380134163

Statistical Machine Learning Methodology for Individualized Treatment Rule Estimation in Precision Medicine.
Shah, Kushal S.

Statistical Machine Learning Methodology for Individualized Treatment Rule Estimation in Precision Medicine. - 1 online resource (124 pages)

Source: Dissertations Abstracts International, Volume: 85-02, Section: B.

Thesis (Ph.D.)--The University of North Carolina at Chapel Hill, 2023.

Includes bibliographical references

Precision medicine aims to deliver optimal, individualized treatments for patients by accounting for their unique characteristics. With a foundation in reinforcement learning, decision theory, and causal inference, the field of precision medicine has seen many advancements in recent years. Significant focus has been placed on creating algorithms to estimate individualized treatment rules (ITRs), which map from patient covariates to the space of available treatments with the goal of maximizing patient outcome. In Chapter 1, we extend ITR estimation methodology in the scenario where variance of the outcome is heterogeneous with respect to treatment and covariates. Accordingly, we propose Stabilized Direct Learning (SD-Learning), which utilizes heteroscedasticity in the error term through a residual reweighting framework that models residual variance via flexible machine learning algorithms such as XGBoost and random forests. We also develop an internal cross-validation scheme which determines the best residual model among competing models. Further, we extend this methodology to multi-arm treatment scenarios. In Chapter 2, we develop ITR estimation methodology for situations where clinical decision-making involves balancing multiple outcomes of interest. Our proposed framework estimates an ITR which maximizes a combination of the multiple clinical outcomes, accounting for the fact that patients may ascribe importance to outcomes differently (utility heterogeneity). This approach employs inverse reinforcement learning (IRL) techniques through an expert-augmentation solution, whereby physicians provide input to guide the utility estimation and ITR learning processes.In Chapter 3, we apply an end-to-end precision medicine workflow to novel data from older adults with Type 1 Diabetes in order to understand the heterogeneous treatment effects of continuous glucose monitoring (CGM) and develop an interpretable ITR to reveal patients for which CGM confers a major safety benefit. The results from this analysis elucidate the demographic and clinical markers which moderate CGM's success, provide the basis for using diagnostic CGM to inform therapeutic CGM decisions, and serve to augment clinical decision-making. Finally, in Chapter 4, as a future research direction, we propose a deep autoencoder framework which simultaneously performs feature selection and ITR optimization, contributing to methodology built for direct consumption of unstructured, high-dimensional data in the precision medicine pipeline.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798380134163Subjects--Topical Terms:

783654
Biostatistics.
Subjects--Index Terms:

Decision theoryIndex Terms--Genre/Form:

554714
Electronic books.

Statistical Machine Learning Methodology for Individualized Treatment Rule Estimation in Precision Medicine.
LDR:04000ntm a22003977 4500 001 1146618
005 20240812064658.5
006 m o d
007 cr bn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798380134163
035 $a (MiAaPQ)AAI30571713
035 $a AAI30571713
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Shah, Kushal S. $3 1472041
245 1 0 $a Statistical Machine Learning Methodology for Individualized Treatment Rule Estimation in Precision Medicine.
264 0 $c 2023
300 $a 1 online resource (124 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-02, Section: B.
500 $a Advisor: Kosorok, Michael R.;Fu, Haoda.
502 $a Thesis (Ph.D.)--The University of North Carolina at Chapel Hill, 2023.
504 $a Includes bibliographical references
520 $a Precision medicine aims to deliver optimal, individualized treatments for patients by accounting for their unique characteristics. With a foundation in reinforcement learning, decision theory, and causal inference, the field of precision medicine has seen many advancements in recent years. Significant focus has been placed on creating algorithms to estimate individualized treatment rules (ITRs), which map from patient covariates to the space of available treatments with the goal of maximizing patient outcome. In Chapter 1, we extend ITR estimation methodology in the scenario where variance of the outcome is heterogeneous with respect to treatment and covariates. Accordingly, we propose Stabilized Direct Learning (SD-Learning), which utilizes heteroscedasticity in the error term through a residual reweighting framework that models residual variance via flexible machine learning algorithms such as XGBoost and random forests. We also develop an internal cross-validation scheme which determines the best residual model among competing models. Further, we extend this methodology to multi-arm treatment scenarios. In Chapter 2, we develop ITR estimation methodology for situations where clinical decision-making involves balancing multiple outcomes of interest. Our proposed framework estimates an ITR which maximizes a combination of the multiple clinical outcomes, accounting for the fact that patients may ascribe importance to outcomes differently (utility heterogeneity). This approach employs inverse reinforcement learning (IRL) techniques through an expert-augmentation solution, whereby physicians provide input to guide the utility estimation and ITR learning processes.In Chapter 3, we apply an end-to-end precision medicine workflow to novel data from older adults with Type 1 Diabetes in order to understand the heterogeneous treatment effects of continuous glucose monitoring (CGM) and develop an interpretable ITR to reveal patients for which CGM confers a major safety benefit. The results from this analysis elucidate the demographic and clinical markers which moderate CGM's success, provide the basis for using diagnostic CGM to inform therapeutic CGM decisions, and serve to augment clinical decision-making. Finally, in Chapter 4, as a future research direction, we propose a deep autoencoder framework which simultaneously performs feature selection and ITR optimization, contributing to methodology built for direct consumption of unstructured, high-dimensional data in the precision medicine pipeline.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Biostatistics. $3 783654
650 4 $a Statistics. $3 556824
653 $a Decision theory
653 $a Precision health
653 $a Precision medicine
653 $a Reinforcement learning
653 $a Statistical machine learning
655 7 $a Electronic books. $2 local $3 554714
690 $a 0308
690 $a 0463
690 $a 0800
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a The University of North Carolina at Chapel Hill. $b Biostatistics. $3 1193116
773 0 $t Dissertations Abstracts International $g 85-02B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30571713 $z click for full text (PQDT)