東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Linked to FindBook

Google Book

Amazon

博客來

MISL: Multiple Imputation by Super Learning and Its Applications in Population Health.

Record Type:	Electronic resources : Monograph/item
Title/Author:	MISL: Multiple Imputation by Super Learning and Its Applications in Population Health./
Author:	Carpenito, Thomas Michael.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2022,
Description:	188 p.
Notes:	Source: Dissertations Abstracts International, Volume: 83-12, Section: B.
Contained By:	Dissertations Abstracts International83-12B.
Subject:	Public health. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29253389
ISBN:	9798834000662

MISL: Multiple Imputation by Super Learning and Its Applications in Population Health.
Carpenito, Thomas Michael.

MISL: Multiple Imputation by Super Learning and Its Applications in Population Health. - Ann Arbor : ProQuest Dissertations & Theses, 2022 - 188 p.

Source: Dissertations Abstracts International, Volume: 83-12, Section: B.

Thesis (Ph.D.)--Northeastern University, 2022.

This item must not be sold to any third party vendors.

BackgroundMissing data are an artifact of almost all health research. While advancements in machine learning have greatly increased our capacity in forming predictive models, their practicality is far less understood when generating imputations used for inference. We investigate a plausible imputation procedure and examine its use in various applications of population health.MethodsWe develop Multiple Imputation by Super Learning (MISL), a novel open-source-ensemble-based-multiple-imputation procedure and evaluate its use in several different simulation environments. We then use MISL to investigate how the quality of imputation impacts variable selection among correlated environmental mixtures of exposures and their association with preterm birth in mothers from the Puerto Rico Test site for Exploring Contamination Threats cohort study. Finally, we again use MISL to better understand the national distribution of firearm injury intent using data from the National Electronic Injury Surveillance System - Firearm Injury Surveillance Study.ResultsWe observe that MISL outperforms current leading methods for multiple imputation both in the presence and absence of interaction effects. Next, we observe that while imputations have little impact on the variable selection process, caution is advised to use a variety of different imputation methods to generate, rather than confirm, hypotheses in the presence of missing data and variable selection. Finally, with MISL, we discover the distribution of assault type incidents over time has remained constant, rather than what was previously believed to be steadily increasing.ConclusionsMISL is a demonstration of how classical statistical methodology can be updated to align with advancements in modern machine learning. The algorithm is adaptable and can be used in a wide variety of different situations, especially population health research.

ISBN: 9798834000662Subjects--Topical Terms:

534748
Public health.
Subjects--Index Terms:

Firearm injuries

MISL: Multiple Imputation by Super Learning and Its Applications in Population Health.
LDR:03195nmm a2200421 4500 001 2349048
005 20220920134654.5
008 241004s2022 ||||||||||||||||| ||eng d
020 $a 9798834000662
035 $a (MiAaPQ)AAI29253389
035 $a AAI29253389
040 $a MiAaPQ $c MiAaPQ
100 1 $a Carpenito, Thomas Michael. $3 3688436
245 1 0 $a MISL: Multiple Imputation by Super Learning and Its Applications in Population Health.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2022
300 $a 188 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-12, Section: B.
500 $a Advisor: Manjourides, Justin.
502 $a Thesis (Ph.D.)--Northeastern University, 2022.
506 $a This item must not be sold to any third party vendors.
520 $a BackgroundMissing data are an artifact of almost all health research. While advancements in machine learning have greatly increased our capacity in forming predictive models, their practicality is far less understood when generating imputations used for inference. We investigate a plausible imputation procedure and examine its use in various applications of population health.MethodsWe develop Multiple Imputation by Super Learning (MISL), a novel open-source-ensemble-based-multiple-imputation procedure and evaluate its use in several different simulation environments. We then use MISL to investigate how the quality of imputation impacts variable selection among correlated environmental mixtures of exposures and their association with preterm birth in mothers from the Puerto Rico Test site for Exploring Contamination Threats cohort study. Finally, we again use MISL to better understand the national distribution of firearm injury intent using data from the National Electronic Injury Surveillance System - Firearm Injury Surveillance Study.ResultsWe observe that MISL outperforms current leading methods for multiple imputation both in the presence and absence of interaction effects. Next, we observe that while imputations have little impact on the variable selection process, caution is advised to use a variety of different imputation methods to generate, rather than confirm, hypotheses in the presence of missing data and variable selection. Finally, with MISL, we discover the distribution of assault type incidents over time has remained constant, rather than what was previously believed to be steadily increasing.ConclusionsMISL is a demonstration of how classical statistical methodology can be updated to align with advancements in modern machine learning. The algorithm is adaptable and can be used in a wide variety of different situations, especially population health research.
590 $a School code: 0160.
650 4 $a Public health. $3 534748
650 4 $a Epidemiology. $3 568544
650 4 $a Biostatistics. $3 1002712
650 4 $a Health sciences. $3 3168359
650 4 $a Artificial intelligence. $3 516317
650 4 $a Information science. $3 554358
653 $a Firearm injuries
653 $a Machine Learning
653 $a Missing data
653 $a Super Learning
653 $a Variable selection
653 $a Multiple Imputation by Super Learning
690 $a 0573
690 $a 0766
690 $a 0308
690 $a 0566
690 $a 0800
690 $a 0723
710 2 $a Northeastern University. $b Health Professions. $3 3688437
773 0 $t Dissertations Abstracts International $g 83-12B.
790 $a 0160
791 $a Ph.D.
792 $a 2022
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29253389