Quality-Guaranteed and Cost-Effective Population Health Profiling: A Deep Active Learning Approach

Long Chen; Jiangtao Wang; Piyushimita (Vonu) Thakuriah

doi:10.1145/3617179

Quality-Guaranteed and Cost-Effective Population Health Profiling: A Deep Active Learning Approach

Long Chen, Jiangtao Wang, Piyushimita (Vonu) Thakuriah

Global Education

Rutgers University

Research output: Contribution to journal › Article › peer-review

1 Downloads (Pure)

Abstract

Reliability and cost are two primary consideration for profiling population-scale prevalence ( PPP ) of multiple None Communicable Diseases ( NCDs ). In this paper, we exploit intra-disease and inter-disease correlation in different traditionally-sensed-areas ( TS-A ) to reduce the required number of the profiling task allocated without compromising the data reliability. Specifically, we propose a novel approach called Compressive Population Health TS-A Selection ( CPH-TS ), which blends the state-of-the-art profile inference, data augmentation and active learning in a unified deep learning framework. It can actively select a minimum number of TS-A regions for profiling task allocation in each profiling cycle, while deducting of the missing data of the unprofiled regions with a probabilistic guarantee of reliability. We evaluate our approach on real-world prevalence datasets of London, which shows the effectiveness of CPH-TS . In general, CPH-TS assigned 11.1-27.3% fewer tasks than baselines, assigning tasks to only 34.7% of the sub-regions while the profiling error below 5% for 95% of the cycles.

Original language	English
Pages (from-to)	(In-Press)
Number of pages	19
Journal	ACM Transactions on Computing for Healthcare
Volume	(In-Press)
Early online date	25 Aug 2023
DOIs	https://doi.org/10.1145/3617179
Publication status	Published - 25 Aug 2023

Bibliographical note

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honoured. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org

Funder

This work was supported by EPSRC New Investigator Award under Grant No EP/V043544/1.

Keywords

Profiling of Prevalence
Spatio-temporal Correlations
Computer Science Applications
Generative Adversarial Network,
Convolutional Neural Networks (CNN)

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1145/3617179

3617179 (1)Accepted author manuscript, 1.21 MB

Cite this

@article{f445ce39085f4110b4da188a648e1930,

title = "Quality-Guaranteed and Cost-Effective Population Health Profiling: A Deep Active Learning Approach",

abstract = "Reliability and cost are two primary consideration for profiling population-scale prevalence ( PPP ) of multiple None Communicable Diseases ( NCDs ). In this paper, we exploit intra-disease and inter-disease correlation in different traditionally-sensed-areas ( TS-A ) to reduce the required number of the profiling task allocated without compromising the data reliability. Specifically, we propose a novel approach called Compressive Population Health TS-A Selection ( CPH-TS ), which blends the state-of-the-art profile inference, data augmentation and active learning in a unified deep learning framework. It can actively select a minimum number of TS-A regions for profiling task allocation in each profiling cycle, while deducting of the missing data of the unprofiled regions with a probabilistic guarantee of reliability. We evaluate our approach on real-world prevalence datasets of London, which shows the effectiveness of CPH-TS . In general, CPH-TS assigned 11.1-27.3% fewer tasks than baselines, assigning tasks to only 34.7% of the sub-regions while the profiling error below 5% for 95% of the cycles.",

keywords = "Profiling of Prevalence, Spatio-temporal Correlations, Computer Science Applications, Generative Adversarial Network,, Convolutional Neural Networks (CNN)",

author = "Long Chen and Jiangtao Wang and Thakuriah, {Piyushimita (Vonu)}",

note = "Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honoured. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org",

year = "2023",

month = aug,

day = "25",

doi = "10.1145/3617179",

language = "English",

volume = "(In-Press)",

pages = "(In--Press)",

journal = "ACM Transactions on Computing for Healthcare",

issn = "2691-1957",

publisher = "Association for Computing Machinery (ACM)",

}

TY - JOUR

T1 - Quality-Guaranteed and Cost-Effective Population Health Profiling: A Deep Active Learning Approach

AU - Chen, Long

AU - Wang, Jiangtao

AU - Thakuriah, Piyushimita (Vonu)

N1 - Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honoured. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org

PY - 2023/8/25

Y1 - 2023/8/25

N2 - Reliability and cost are two primary consideration for profiling population-scale prevalence ( PPP ) of multiple None Communicable Diseases ( NCDs ). In this paper, we exploit intra-disease and inter-disease correlation in different traditionally-sensed-areas ( TS-A ) to reduce the required number of the profiling task allocated without compromising the data reliability. Specifically, we propose a novel approach called Compressive Population Health TS-A Selection ( CPH-TS ), which blends the state-of-the-art profile inference, data augmentation and active learning in a unified deep learning framework. It can actively select a minimum number of TS-A regions for profiling task allocation in each profiling cycle, while deducting of the missing data of the unprofiled regions with a probabilistic guarantee of reliability. We evaluate our approach on real-world prevalence datasets of London, which shows the effectiveness of CPH-TS . In general, CPH-TS assigned 11.1-27.3% fewer tasks than baselines, assigning tasks to only 34.7% of the sub-regions while the profiling error below 5% for 95% of the cycles.

AB - Reliability and cost are two primary consideration for profiling population-scale prevalence ( PPP ) of multiple None Communicable Diseases ( NCDs ). In this paper, we exploit intra-disease and inter-disease correlation in different traditionally-sensed-areas ( TS-A ) to reduce the required number of the profiling task allocated without compromising the data reliability. Specifically, we propose a novel approach called Compressive Population Health TS-A Selection ( CPH-TS ), which blends the state-of-the-art profile inference, data augmentation and active learning in a unified deep learning framework. It can actively select a minimum number of TS-A regions for profiling task allocation in each profiling cycle, while deducting of the missing data of the unprofiled regions with a probabilistic guarantee of reliability. We evaluate our approach on real-world prevalence datasets of London, which shows the effectiveness of CPH-TS . In general, CPH-TS assigned 11.1-27.3% fewer tasks than baselines, assigning tasks to only 34.7% of the sub-regions while the profiling error below 5% for 95% of the cycles.

KW - Profiling of Prevalence

KW - Spatio-temporal Correlations

KW - Computer Science Applications

KW - Generative Adversarial Network,

KW - Convolutional Neural Networks (CNN)

U2 - 10.1145/3617179

DO - 10.1145/3617179

M3 - Article

SN - 2691-1957

VL - (In-Press)

SP - (In-Press)

JO - ACM Transactions on Computing for Healthcare

JF - ACM Transactions on Computing for Healthcare

ER -

Quality-Guaranteed and Cost-Effective Population Health Profiling: A Deep Active Learning Approach

Abstract

Bibliographical note

Funder

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this