Cardiac Comorbidity Risk Score: Zero-Burden Machine Learning to Improve Prediction of Postoperative Major Adverse Cardiac Events in Hip and Knee Arthroplasty

Dmytro Onishchenko; Daniel S. Rubin; James R. van Horne; R. Parker Ward; Ishanu Chattopadhyay

doi:10.1161/JAHA.121.023745

Cardiac Comorbidity Risk Score: Zero-Burden Machine Learning to Improve Prediction of Postoperative Major Adverse Cardiac Events in Hip and Knee Arthroplasty

Dmytro Onishchenko
, Daniel S. Rubin
, James R. van Horne
, R. Parker Ward
, Ishanu Chattopadhyay

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

10 Scopus citations

Abstract

BACKGROUND: In this retrospective, observational study we introduce the Cardiac Comorbidity Risk Score, predicting perio-perative major adverse cardiac events (MACE) after elective hip and knee arthroplasty. MACE is a rare but important driver of mortality, and existing tools, eg, the Revised Cardiac Risk Index demonstrate only modest accuracy. We demonstrate an artificial intelligence-based approach to identify patients at high risk of MACE within 4 weeks (primary outcome) of arthroplasty, that imposes zero additional burden of cost/resources. METHODS AND RESULTS: Cardiac Comorbidity Risk Score calculation uses novel machine learning to estimate MACE risk from patient electronic health records, without requiring blood work or access to any demographic data beyond that of sex and age, and accounts for variable/missing/incomplete information across patient records. Validated on a deidentified cohort (age >45 years, n=445 391), performance was evaluated using the area under the receiver operator characteristics curve (AUROC), sensitivity/specificity, positive predictive value, and positive/negative likelihood ratios. In our cohort (age 63.5±10.5 years, 58.2% women, 34.2%/65.8% hip/knee procedures), 0.19% (882) experienced the primary outcome. Cardiac Comorbidity Risk Score achieved area under the receiver operator characteristics curve=80.0±0.4% (95% CI) for women and 80.1±0.5% (95% CI) for males, with 36.4% and 35.1% sensitivities, respectively, at 95% specificity, significantly outperforming Revised Cardiac Risk Index across all studied age-, sex-, risk-, and comorbidity-based subgroups. CONCLUSIONS: Cardiac Comorbidity Risk Score, a novel artificial intelligence-based screening tool using known and unknown comorbidity patterns, outperforms state-of-the-art in predicting MACE within 4 weeks postarthroplasty, and can identify patients at high risk that do not demonstrate traditional risk factors.

Original language	English
Article number	e023745
Journal	Journal of the American Heart Association
Volume	11
Issue number	15
DOIs	https://doi.org/10.1161/JAHA.121.023745
State	Published - Aug 2 2022

Bibliographical note

Publisher Copyright:
© 2022 The Authors.

Funding

This work is funded in part by the Defense Advanced Research Projects Agency project number HR00111890043/P00004. The claims made in this study do not reflect the position or the policy of the US Government. Dr Chattopadhyay is a founder and shareholder of Zero Burden Laboratories, Inc., a company formed to commercialize biomedical applications of machine learning. He has not taken any salary or money from the company. He has received funding from the United States Department of Defense, the National Institutes of Health, and the Neubauer Collegium for Culture and Society. Dr Rubin is the President of DRDR Mobile Health, a company that creates mobile applications for health care, including functional capacity assessment applications. He has not taken any salary or money from the company. He has engaged in consulting for mobile applications as well. Mr Onishchenko is a founder and shareholder of Zero Burden Laboratories. He has not taken any salary or money from the company. The remaining authors have no disclosures to report.

Funders	Funder number
Neubauer Collegium for Culture and Society
National Institutes of Health
U.S. Department of Defense
Defense Advanced Research Projects Agency	HR00111890043/P00004
Government of South Australia

Keywords

hip and knee arthroplasty
machine learning
Revised Cardiac Risk Index
risk of MACE

ASJC Scopus subject areas

Cardiology and Cardiovascular Medicine

Access to Document

10.1161/JAHA.121.023745

Cite this

@article{d9fd5bfc3cf34400b879eb1ee992ca2e,

title = "Cardiac Comorbidity Risk Score: Zero-Burden Machine Learning to Improve Prediction of Postoperative Major Adverse Cardiac Events in Hip and Knee Arthroplasty",

abstract = "BACKGROUND: In this retrospective, observational study we introduce the Cardiac Comorbidity Risk Score, predicting perio-perative major adverse cardiac events (MACE) after elective hip and knee arthroplasty. MACE is a rare but important driver of mortality, and existing tools, eg, the Revised Cardiac Risk Index demonstrate only modest accuracy. We demonstrate an artificial intelligence-based approach to identify patients at high risk of MACE within 4 weeks (primary outcome) of arthroplasty, that imposes zero additional burden of cost/resources. METHODS AND RESULTS: Cardiac Comorbidity Risk Score calculation uses novel machine learning to estimate MACE risk from patient electronic health records, without requiring blood work or access to any demographic data beyond that of sex and age, and accounts for variable/missing/incomplete information across patient records. Validated on a deidentified cohort (age >45 years, n=445 391), performance was evaluated using the area under the receiver operator characteristics curve (AUROC), sensitivity/specificity, positive predictive value, and positive/negative likelihood ratios. In our cohort (age 63.5±10.5 years, 58.2\% women, 34.2\%/65.8\% hip/knee procedures), 0.19\% (882) experienced the primary outcome. Cardiac Comorbidity Risk Score achieved area under the receiver operator characteristics curve=80.0±0.4\% (95\% CI) for women and 80.1±0.5\% (95\% CI) for males, with 36.4\% and 35.1\% sensitivities, respectively, at 95\% specificity, significantly outperforming Revised Cardiac Risk Index across all studied age-, sex-, risk-, and comorbidity-based subgroups. CONCLUSIONS: Cardiac Comorbidity Risk Score, a novel artificial intelligence-based screening tool using known and unknown comorbidity patterns, outperforms state-of-the-art in predicting MACE within 4 weeks postarthroplasty, and can identify patients at high risk that do not demonstrate traditional risk factors.",

keywords = "hip and knee arthroplasty, machine learning, Revised Cardiac Risk Index, risk of MACE",

author = "Dmytro Onishchenko and Rubin, \{Daniel S.\} and \{van Horne\}, \{James R.\} and Ward, \{R. Parker\} and Ishanu Chattopadhyay",

note = "Publisher Copyright: {\textcopyright} 2022 The Authors.",

year = "2022",

month = aug,

day = "2",

doi = "10.1161/JAHA.121.023745",

language = "English",

volume = "11",

number = "15",

}

Cardiac Comorbidity Risk Score: Zero-Burden Machine Learning to Improve Prediction of Postoperative Major Adverse Cardiac Events in Hip and Knee Arthroplasty. / Onishchenko, Dmytro; Rubin, Daniel S.; van Horne, James R. et al.
In: Journal of the American Heart Association, Vol. 11, No. 15, e023745, 02.08.2022.

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

TY - JOUR

T1 - Cardiac Comorbidity Risk Score

T2 - Zero-Burden Machine Learning to Improve Prediction of Postoperative Major Adverse Cardiac Events in Hip and Knee Arthroplasty

AU - Onishchenko, Dmytro

AU - Rubin, Daniel S.

AU - van Horne, James R.

AU - Ward, R. Parker

AU - Chattopadhyay, Ishanu

PY - 2022/8/2

Y1 - 2022/8/2

N2 - BACKGROUND: In this retrospective, observational study we introduce the Cardiac Comorbidity Risk Score, predicting perio-perative major adverse cardiac events (MACE) after elective hip and knee arthroplasty. MACE is a rare but important driver of mortality, and existing tools, eg, the Revised Cardiac Risk Index demonstrate only modest accuracy. We demonstrate an artificial intelligence-based approach to identify patients at high risk of MACE within 4 weeks (primary outcome) of arthroplasty, that imposes zero additional burden of cost/resources. METHODS AND RESULTS: Cardiac Comorbidity Risk Score calculation uses novel machine learning to estimate MACE risk from patient electronic health records, without requiring blood work or access to any demographic data beyond that of sex and age, and accounts for variable/missing/incomplete information across patient records. Validated on a deidentified cohort (age >45 years, n=445 391), performance was evaluated using the area under the receiver operator characteristics curve (AUROC), sensitivity/specificity, positive predictive value, and positive/negative likelihood ratios. In our cohort (age 63.5±10.5 years, 58.2% women, 34.2%/65.8% hip/knee procedures), 0.19% (882) experienced the primary outcome. Cardiac Comorbidity Risk Score achieved area under the receiver operator characteristics curve=80.0±0.4% (95% CI) for women and 80.1±0.5% (95% CI) for males, with 36.4% and 35.1% sensitivities, respectively, at 95% specificity, significantly outperforming Revised Cardiac Risk Index across all studied age-, sex-, risk-, and comorbidity-based subgroups. CONCLUSIONS: Cardiac Comorbidity Risk Score, a novel artificial intelligence-based screening tool using known and unknown comorbidity patterns, outperforms state-of-the-art in predicting MACE within 4 weeks postarthroplasty, and can identify patients at high risk that do not demonstrate traditional risk factors.

AB - BACKGROUND: In this retrospective, observational study we introduce the Cardiac Comorbidity Risk Score, predicting perio-perative major adverse cardiac events (MACE) after elective hip and knee arthroplasty. MACE is a rare but important driver of mortality, and existing tools, eg, the Revised Cardiac Risk Index demonstrate only modest accuracy. We demonstrate an artificial intelligence-based approach to identify patients at high risk of MACE within 4 weeks (primary outcome) of arthroplasty, that imposes zero additional burden of cost/resources. METHODS AND RESULTS: Cardiac Comorbidity Risk Score calculation uses novel machine learning to estimate MACE risk from patient electronic health records, without requiring blood work or access to any demographic data beyond that of sex and age, and accounts for variable/missing/incomplete information across patient records. Validated on a deidentified cohort (age >45 years, n=445 391), performance was evaluated using the area under the receiver operator characteristics curve (AUROC), sensitivity/specificity, positive predictive value, and positive/negative likelihood ratios. In our cohort (age 63.5±10.5 years, 58.2% women, 34.2%/65.8% hip/knee procedures), 0.19% (882) experienced the primary outcome. Cardiac Comorbidity Risk Score achieved area under the receiver operator characteristics curve=80.0±0.4% (95% CI) for women and 80.1±0.5% (95% CI) for males, with 36.4% and 35.1% sensitivities, respectively, at 95% specificity, significantly outperforming Revised Cardiac Risk Index across all studied age-, sex-, risk-, and comorbidity-based subgroups. CONCLUSIONS: Cardiac Comorbidity Risk Score, a novel artificial intelligence-based screening tool using known and unknown comorbidity patterns, outperforms state-of-the-art in predicting MACE within 4 weeks postarthroplasty, and can identify patients at high risk that do not demonstrate traditional risk factors.

KW - hip and knee arthroplasty

KW - machine learning

KW - Revised Cardiac Risk Index

KW - risk of MACE

UR - https://www.scopus.com/pages/publications/85135212561

UR - https://www.scopus.com/pages/publications/85135212561#tab=citedBy

U2 - 10.1161/JAHA.121.023745

DO - 10.1161/JAHA.121.023745

M3 - Article

C2 - 35904198

AN - SCOPUS:85135212561

SN - 2047-9980

VL - 11

JO - Journal of the American Heart Association

JF - Journal of the American Heart Association

IS - 15

M1 - e023745

ER -

Cardiac Comorbidity Risk Score: Zero-Burden Machine Learning to Improve Prediction of Postoperative Major Adverse Cardiac Events in Hip and Knee Arthroplasty

Abstract

Bibliographical note

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this