A Generalized Machine Learning Model for Identifying Congenital Heart Defects (CHDs) Using ICD Codes

IF 1.6 4区医学 Q4 DEVELOPMENTAL BIOLOGY

Birth Defects Research Pub Date : 2025-01-31 DOI:10.1002/bdr2.2440

Haoming Shi, Wendy M. Book, Lindsey C. Ivey, Fred H. Rodriguez III, Cheryl Raskind-Hood, Karrie F. Downing, Sherry L. Farr, Courtney E. McCracken, Vinita O. Leedom, Susan E. Haynes, Sandra Amouzou, Reza Sameni, Rishikesan Kamaleswaran

{"title":"A Generalized Machine Learning Model for Identifying Congenital Heart Defects (CHDs) Using ICD Codes","authors":"Haoming Shi, Wendy M. Book, Lindsey C. Ivey, Fred H. Rodriguez III, Cheryl Raskind-Hood, Karrie F. Downing, Sherry L. Farr, Courtney E. McCracken, Vinita O. Leedom, Susan E. Haynes, Sandra Amouzou, Reza Sameni, Rishikesan Kamaleswaran","doi":"10.1002/bdr2.2440","DOIUrl":null,"url":null,"abstract":"<div>\n \n \n <section>\n \n <h3> Background</h3>\n \n <p><i>International Classification of Diseases</i> (ICD) codes utilized for congenital heart defect (CHD) case identification in datasets have substantial false-positive (FP) rates. Incorporating machine learning (ML) algorithms following case selection by ICD codes may improve the accuracy of CHD identification, enhancing surveillance efforts.</p>\n </section>\n \n <section>\n \n <h3> Methods</h3>\n \n <p>Traditional ML methods were applied to four encounter-level datasets, 2010–2019, for 3334 patients with validated diagnoses and with at least one CHD ICD code identified. A 5-fold cross-validation approach was applied to the dataset to determine the set of overlapping important features best classifying CHD cases. Training and testing combinations were explored to determine the approach yielding the most accurate CHD classification.</p>\n </section>\n \n <section>\n \n <h3> Results</h3>\n \n <p>CHD ICD positive predictive values (PPVs) by site ranged from 53.2% to 84.0%. The ML algorithm achieved a PPV of 95% (1273/1340) for the four-site dataset with a false-negative (FN) rate of 33% (639/1912) by choosing an operating point prioritizing PPV from the PPV–FN rate curve. XGBoost reduced 2105 Clinical Classification Software (CCS) features to 137 that identified those with true-positive (TP) CHD and false-positive FP classification.</p>\n </section>\n \n <section>\n \n <h3> Conclusion</h3>\n \n <p>Applying ML algorithms following case selection by CHD-related ICD codes improved the accuracy of identifying TP true-positive CHD cases.</p>\n </section>\n </div>","PeriodicalId":9121,"journal":{"name":"Birth Defects Research","volume":"117 2","pages":""},"PeriodicalIF":1.6000,"publicationDate":"2025-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Birth Defects Research","FirstCategoryId":"3","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1002/bdr2.2440","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"DEVELOPMENTAL BIOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Background

International Classification of Diseases (ICD) codes utilized for congenital heart defect (CHD) case identification in datasets have substantial false-positive (FP) rates. Incorporating machine learning (ML) algorithms following case selection by ICD codes may improve the accuracy of CHD identification, enhancing surveillance efforts.

Methods

Traditional ML methods were applied to four encounter-level datasets, 2010–2019, for 3334 patients with validated diagnoses and with at least one CHD ICD code identified. A 5-fold cross-validation approach was applied to the dataset to determine the set of overlapping important features best classifying CHD cases. Training and testing combinations were explored to determine the approach yielding the most accurate CHD classification.

Results

CHD ICD positive predictive values (PPVs) by site ranged from 53.2% to 84.0%. The ML algorithm achieved a PPV of 95% (1273/1340) for the four-site dataset with a false-negative (FN) rate of 33% (639/1912) by choosing an operating point prioritizing PPV from the PPV–FN rate curve. XGBoost reduced 2105 Clinical Classification Software (CCS) features to 137 that identified those with true-positive (TP) CHD and false-positive FP classification.

Conclusion

Applying ML algorithms following case selection by CHD-related ICD codes improved the accuracy of identifying TP true-positive CHD cases.

查看原文本刊更多论文

求助全文

约1分钟内获得全文求助全文

来源期刊

Birth Defects Research Medicine-Embryology

CiteScore

3.60

自引率

9.50%

发文量

153

期刊介绍： The journal Birth Defects Research publishes original research and reviews in areas related to the etiology of adverse developmental and reproductive outcome. In particular the journal is devoted to the publication of original scientific research that contributes to the understanding of the biology of embryonic development and the prenatal causative factors and mechanisms leading to adverse pregnancy outcomes, namely structural and functional birth defects, pregnancy loss, postnatal functional defects in the human population, and to the identification of prenatal factors and biological mechanisms that reduce these risks. Adverse reproductive and developmental outcomes may have genetic, environmental, nutritional or epigenetic causes. Accordingly, the journal Birth Defects Research takes an integrated, multidisciplinary approach in its organization and publication strategy. The journal Birth Defects Research contains separate sections for clinical and molecular teratology, developmental and reproductive toxicology, and reviews in developmental biology to acknowledge and accommodate the integrative nature of research in this field. Each section has a dedicated editor who is a leader in his/her field and who has full editorial authority in his/her area.