Using Contextual Learning to Improve Diagnostic Accuracy: Application in Breast Cancer Screening

Linqi Song, William Hsu, Jie Xu, Mihaela Van Der Schaar

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

Clinicians need to routinely make management decisions about patients who are at risk for a disease such as breast cancer. This paper presents a novel clinical decision support tool that is capable of helping physicians make diagnostic decisions. We apply this support system to improve the specificity of breast cancer screening and diagnosis. The system utilizes clinical context (e.g., demographics, medical history) to minimize the false positive rates while avoiding false negatives. An online contextual learning algorithm is used to update the diagnostic strategy presented to the physicians over time. We analytically evaluate the diagnostic performance loss of the proposed algorithm, in which the true patient distribution is not known and needs to be learned, as compared with the optimal strategy where all information is assumed known, and prove that the false positive rate of the proposed learning algorithm asymptotically converges to the optimum. In addition, our algorithm also has the important merit that it can provide individualized confidence estimates about the accuracy of the diagnosis recommendation. Moreover, the relevancy of contextual features is assessed, enabling the approach to identify specific contextual features that provide the most value of information in reducing diagnostic errors. Experiments were conducted using patient data collected at a large academic medical center. Our proposed approach outperforms the current clinical practice by 36% in terms of false positive rate given a 2% false negative rate.

Original languageEnglish (US)
Article number7064753
Pages (from-to)902-914
Number of pages13
JournalIEEE Journal of Biomedical and Health Informatics
Volume20
Issue number3
DOIs
StatePublished - May 1 2016
Externally publishedYes

Fingerprint

Early Detection of Cancer
Learning algorithms
Screening
Learning
Breast Neoplasms
Clinical Decision Support Systems
Physicians
Diagnostic Errors
Experiments
Demography

Keywords

  • breast cancer
  • Computer-aided diagnosis system
  • contextual learning
  • multi-armed bandit
  • online learning

ASJC Scopus subject areas

  • Biotechnology
  • Computer Science Applications
  • Electrical and Electronic Engineering
  • Health Information Management

Cite this

Using Contextual Learning to Improve Diagnostic Accuracy : Application in Breast Cancer Screening. / Song, Linqi; Hsu, William; Xu, Jie; Van Der Schaar, Mihaela.

In: IEEE Journal of Biomedical and Health Informatics, Vol. 20, No. 3, 7064753, 01.05.2016, p. 902-914.

Research output: Contribution to journalArticle

@article{87c6f0b0290e4f47b6c1554e72444ec3,
title = "Using Contextual Learning to Improve Diagnostic Accuracy: Application in Breast Cancer Screening",
abstract = "Clinicians need to routinely make management decisions about patients who are at risk for a disease such as breast cancer. This paper presents a novel clinical decision support tool that is capable of helping physicians make diagnostic decisions. We apply this support system to improve the specificity of breast cancer screening and diagnosis. The system utilizes clinical context (e.g., demographics, medical history) to minimize the false positive rates while avoiding false negatives. An online contextual learning algorithm is used to update the diagnostic strategy presented to the physicians over time. We analytically evaluate the diagnostic performance loss of the proposed algorithm, in which the true patient distribution is not known and needs to be learned, as compared with the optimal strategy where all information is assumed known, and prove that the false positive rate of the proposed learning algorithm asymptotically converges to the optimum. In addition, our algorithm also has the important merit that it can provide individualized confidence estimates about the accuracy of the diagnosis recommendation. Moreover, the relevancy of contextual features is assessed, enabling the approach to identify specific contextual features that provide the most value of information in reducing diagnostic errors. Experiments were conducted using patient data collected at a large academic medical center. Our proposed approach outperforms the current clinical practice by 36{\%} in terms of false positive rate given a 2{\%} false negative rate.",
keywords = "breast cancer, Computer-aided diagnosis system, contextual learning, multi-armed bandit, online learning",
author = "Linqi Song and William Hsu and Jie Xu and {Van Der Schaar}, Mihaela",
year = "2016",
month = "5",
day = "1",
doi = "10.1109/JBHI.2015.2414934",
language = "English (US)",
volume = "20",
pages = "902--914",
journal = "IEEE Journal of Biomedical and Health Informatics",
issn = "2168-2194",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "3",

}

TY - JOUR

T1 - Using Contextual Learning to Improve Diagnostic Accuracy

T2 - Application in Breast Cancer Screening

AU - Song, Linqi

AU - Hsu, William

AU - Xu, Jie

AU - Van Der Schaar, Mihaela

PY - 2016/5/1

Y1 - 2016/5/1

N2 - Clinicians need to routinely make management decisions about patients who are at risk for a disease such as breast cancer. This paper presents a novel clinical decision support tool that is capable of helping physicians make diagnostic decisions. We apply this support system to improve the specificity of breast cancer screening and diagnosis. The system utilizes clinical context (e.g., demographics, medical history) to minimize the false positive rates while avoiding false negatives. An online contextual learning algorithm is used to update the diagnostic strategy presented to the physicians over time. We analytically evaluate the diagnostic performance loss of the proposed algorithm, in which the true patient distribution is not known and needs to be learned, as compared with the optimal strategy where all information is assumed known, and prove that the false positive rate of the proposed learning algorithm asymptotically converges to the optimum. In addition, our algorithm also has the important merit that it can provide individualized confidence estimates about the accuracy of the diagnosis recommendation. Moreover, the relevancy of contextual features is assessed, enabling the approach to identify specific contextual features that provide the most value of information in reducing diagnostic errors. Experiments were conducted using patient data collected at a large academic medical center. Our proposed approach outperforms the current clinical practice by 36% in terms of false positive rate given a 2% false negative rate.

AB - Clinicians need to routinely make management decisions about patients who are at risk for a disease such as breast cancer. This paper presents a novel clinical decision support tool that is capable of helping physicians make diagnostic decisions. We apply this support system to improve the specificity of breast cancer screening and diagnosis. The system utilizes clinical context (e.g., demographics, medical history) to minimize the false positive rates while avoiding false negatives. An online contextual learning algorithm is used to update the diagnostic strategy presented to the physicians over time. We analytically evaluate the diagnostic performance loss of the proposed algorithm, in which the true patient distribution is not known and needs to be learned, as compared with the optimal strategy where all information is assumed known, and prove that the false positive rate of the proposed learning algorithm asymptotically converges to the optimum. In addition, our algorithm also has the important merit that it can provide individualized confidence estimates about the accuracy of the diagnosis recommendation. Moreover, the relevancy of contextual features is assessed, enabling the approach to identify specific contextual features that provide the most value of information in reducing diagnostic errors. Experiments were conducted using patient data collected at a large academic medical center. Our proposed approach outperforms the current clinical practice by 36% in terms of false positive rate given a 2% false negative rate.

KW - breast cancer

KW - Computer-aided diagnosis system

KW - contextual learning

KW - multi-armed bandit

KW - online learning

UR - http://www.scopus.com/inward/record.url?scp=84969931477&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84969931477&partnerID=8YFLogxK

U2 - 10.1109/JBHI.2015.2414934

DO - 10.1109/JBHI.2015.2414934

M3 - Article

C2 - 25807575

AN - SCOPUS:84969931477

VL - 20

SP - 902

EP - 914

JO - IEEE Journal of Biomedical and Health Informatics

JF - IEEE Journal of Biomedical and Health Informatics

SN - 2168-2194

IS - 3

M1 - 7064753

ER -