Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Meta-Analysis
. 2025 Aug 7;14(8):474-488.
doi: 10.1093/ehjacc/zuaf051.

Machine learning to optimize use of natriuretic peptides in the diagnosis of acute heart failure

Dimitrios Doudesis  1   2 , Kuan Ken Lee  1 , Mohamed Anwar  1 , Adam J Singer  3 , Judd E Hollander  4 , Camille Chenevier-Gobeaux  5 , Yann-Erick Claessens  6 , Desiree Wussler  7   8 , Dominic Weil  7 , Nikola Kozhuharov  7   9 , Ivo Strebel  7 , Zaid Sabti  7 , Christopher deFilippi  10 , Stephen Seliger  11 , Evandro Tinoco Mesquita  12 , Jan C Wiemer  13 , Martin Möckel  14 , Joel Coste  15 , Patrick Jourdain  16 , Komukai Kimiaki  17 , Michihiro Yoshimura  17 , Irwani Ibrahim  18 , Shirley Beng Suat Ooi  18 , Win Sen Kuan  18 , Alfons Gegenhuber  19 , Thomas Mueller  20 , Olivier Hanon  21 , Jean-Sébastien Vidal  21 , Peter Cameron  22 , Louisa Lam  22   23 , Ben Freedman  24 , Tommy Chung  25 , Sean P Collins  26 , Christopher J Lindsell  27 , David E Newby  1 , Alan G Japp  1 , Anoop S V Shah  1   28 , Humberto Villacorta  12 , A Mark Richards  29   30 , John J V McMurray  31 , Christian Mueller  7 , James L Januzzi  32   33 , Nicholas L Mills  1   2 ; CoDE-HF investigators
Collaborators, Affiliations
Meta-Analysis

Machine learning to optimize use of natriuretic peptides in the diagnosis of acute heart failure

Dimitrios Doudesis et al. Eur Heart J Acute Cardiovasc Care. .

Abstract

Aims: B-type natriuretic peptide (BNP) and mid-regional pro-atrial natriuretic peptide (MR-proANP) testing are guideline-recommended to aid in the diagnosis of acute heart failure. Nevertheless, the diagnostic performance of these biomarkers is uncertain.

Methods and results: We performed a systematic review and individual patient-level data meta-analysis to evaluate the diagnostic performance of BNP and MR-proANP. We subsequently developed and externally validated a decision-support tool called CoDE-HF that combines natriuretic peptide concentrations with clinical variables using machine learning to report the probability of acute heart failure. Fourteen studies from 12 countries provided individual patient-level data in 8493 patients for BNP and 3899 patients for MR-proANP, in whom, 48.3% (4105/8493) and 41.3% (1611/3899) had an adjudicated diagnosis of acute heart failure, respectively. The negative predictive value (NPV) of guideline-recommended thresholds for BNP (100 pg/mL) and MR-proANP (120 pmol/L) was 93.6% (95% confidence interval 88.4-96.6%) and 95.6% (92.2-97.6%), respectively, whilst the positive predictive value (PPV) was 68.8% (62.9-74.2%) and 64.8% (56.3-72.5%). Significant heterogeneity in the performance of these thresholds was observed across important subgroups. CoDE-HF was well calibrated with excellent discrimination in those without prior acute heart failure for both BNP and MR-proANP [area under the curve of 0.914 (0.906-0.921) and 0.929 (0.919-0.939), and Brier scores of 0.110 and 0.094, respectively]. CoDE-HF with BNP and MR-proANP identified 30% and 48% as low-probability [NPV of 98.5% (97.1-99.3%) and 98.5% (97.7-99.0%)], and 30% and 28% as high-probability [PPV of 78.6% (70.4-85.0%) and 75.1% (70.9-78.9%)], respectively, and performed consistently across subgroups.

Conclusion: The diagnostic performance of guideline-recommended BNP and MR-proANP thresholds for acute heart failure varied significantly across patient subgroups. A decision-support tool that combines natriuretic peptides and clinical variables was more accurate and supports more individualized diagnosis.

Study registration: PROSPERO number, CRD42019159407.

Keywords: Heart failure; Machine learning; Natriuretic peptide.

PubMed Disclaimer

Conflict of interest statement

Conflict of interest: A.B.-G. reports personal fees and non-financial support from Roche Diagnostics, during the conduct of the study; personal fees from Abbott, personal fees from AstraZeneca, grants, personal fees and non-financial support from Boehringer-Ingelheim, personal fees and non-financial support from Novartis, personal fees and non-financial support from Vifor, outside the submitted work. J.C.W. reports work as a biostatistician at the biotech company B·R·A·H·M·S GmbH, part of Thermo Fisher Scientific. M.M. reports grants from Health Care Research Projects, grants from Biomarker Research, personal fees from Consulting, outside the submitted work. A.S.V.S. reports speaker fees from Abbott Diagnostics, outside the submitted work. A.G.J. reports speaker fees/consultancy fees from Astra Zeneca, Novartis, Vifor, Bayer and Pharmacosmos. H.V. reports speaker fees and consulting fees from Roche Diagnostics and speaker fees from Novartis, AstraZeneca, Boehringer Ingelheim, Servier, Bayer, and Daiichi Sankyo, outside the submitted work. A.M.R. reports grants, personal fees and non-financial support from Roche Diagnostics, outside the submitted work. J.J.V.M. reports consulting fees from Alnylam, AstraZeneca, Bayer, Boehringer Ingelheim, BMS, Cardurion, Dal-Cor, GSK, Ionis, KBP Biosciences, Novartis, Pfizer, Theracos, payments for advisory boards, symposia or lectures from Abbott, Alkem Metabolics, Canadian Medical & Surgical Knowledge Translation Research Group, Eris Lifesciences, Hikma, Lupin, Sun Pharmaceuticals, Medscape/Heart.Org, ProAdWise Communications, Radcliffe Cardiology, Servier, the Corpus, participation on a Data Safety Monitoring Board or Advisory Board for Cardialysis (MONITOR study) and Merck (VICTORIA trial) and work as company director for Global Clinical Trial Partners Ltd, outside the submitted work. C.M. reports grants and non-financial support from several diagnostic companies during the conduct of the study; grants, personal fees and non-financial support from several diagnostic companies, outside the submitted work. N.L.M. reports speaker fees from Abbott Diagnostics and Siemens Healthineers, and personal fees for consultancy or advisory boards from Roche Diagnostics and LumiraDx, outside the submitted work. K.K.L., D.D., and N.L.M. are employed by the University of Edinburgh, who has filed a patent on the CoDE-HF score (UK Intellectual Property Office Reference: PCT/GB2021/051470). J.J. is a trustee of the American College of Cardiology; is a board member of Imbria Pharmaceuticals; has received grant support from Abbott, Applied Therapeutics, HeartFlow, Innolife, and Roche Diagnostics; has received consulting income from Abbott, Beckman, Boehringer Ingelheim, Bristol Myers Squibb, Janssen, Jana Care, Merck, Novartis, Pfizer, Roche Diagnostics, and Siemens; and participates in clinical endpoint committees/data safety monitoring boards for Abbott, AbbVie, CVRx, Intercept, and Takeda. S.P.C. receives research support from NIH, PCORI, DOD and AHRQ and serves as a consultant for Reprieve Cardiovascular and Abbott. C.J.L. reports grants to institution from NIH, CDC and DoD, contracts to institution from Entegrion, Endpoint Health, BioMerieux, Biomeme, Novartis, and AstraZeneca, income from VUMC, Emory University and Rocket Pharmaceuticals for research services, stock options in Bioscape Digital, patents for risk stratification in sepsis and septic shock held by Cincinnati Children’s Hospital Medical Centre, and income from ACTS for service as Editor-In-Chief of the Journal of Clinical and Translational Science.

Figures

Graphical Abstract
Graphical Abstract
Figure 1
Figure 1
BNP and MR-proANP thresholds for acute heart failure. (A) (top) NPVs of BNP concentrations to rule-out a diagnosis of acute heart failure. (bottom) Cumulative proportion of patients presenting with suspected acute heart failure with BNP concentrations below each threshold. (B) (top) NPVs of MR-proANP concentrations to rule-out a diagnosis of acute heart failure. (bottom) Cumulative proportion of patients presenting with suspected acute heart failure with MR-proANP concentrations below each threshold. * dashed horizontal line corresponds to NPV of 98%.
Figure 1
Figure 1
BNP and MR-proANP thresholds for acute heart failure. (A) (top) NPVs of BNP concentrations to rule-out a diagnosis of acute heart failure. (bottom) Cumulative proportion of patients presenting with suspected acute heart failure with BNP concentrations below each threshold. (B) (top) NPVs of MR-proANP concentrations to rule-out a diagnosis of acute heart failure. (bottom) Cumulative proportion of patients presenting with suspected acute heart failure with MR-proANP concentrations below each threshold. * dashed horizontal line corresponds to NPV of 98%.
Figure 2
Figure 2
NPV of guideline-recommended BNP and MR-proANP thresholds across patient subgroups. (A) NPV of the BNP threshold of 100 pg/mL across patient subgroups. (B) NPV of the MR-proANP threshold of 120 pmol/L across patient subgroups. COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate.
Figure 2
Figure 2
NPV of guideline-recommended BNP and MR-proANP thresholds across patient subgroups. (A) NPV of the BNP threshold of 100 pg/mL across patient subgroups. (B) NPV of the MR-proANP threshold of 120 pmol/L across patient subgroups. COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate.
Figure 3
Figure 3
Calibration plot of CoDE-HF with BNP in patients with (A) no previous heart failure and (B) previous heart failure.
Figure 4
Figure 4
Diagnostic performance of the CoDE-HF score across patient subgroups. CoDE-HF incorporates BNP concentrations as a continuous measure and predefined simple objective clinical variables (age, eGFR, haemoglobin, body mass index, heart rate, blood pressure, peripheral oedema, prior history of heart failure, COPD and ischaemic heart disease) to provide an individualized assessment of the likelihood of the diagnosis of acute heart failure. (A) NPV of the CoDE-HF rule-out score of 5.4 in patients without prior heart failure across patient subgroups. (B) PPV of the CoDE-HF rule-in score of 58.0 in patients without prior heart failure across patient subgroups. (C) PPV of the CoDE-HF rule-in score of 90.7 in patients with prior heart failure across patient subgroups.
Figure 4
Figure 4
Diagnostic performance of the CoDE-HF score across patient subgroups. CoDE-HF incorporates BNP concentrations as a continuous measure and predefined simple objective clinical variables (age, eGFR, haemoglobin, body mass index, heart rate, blood pressure, peripheral oedema, prior history of heart failure, COPD and ischaemic heart disease) to provide an individualized assessment of the likelihood of the diagnosis of acute heart failure. (A) NPV of the CoDE-HF rule-out score of 5.4 in patients without prior heart failure across patient subgroups. (B) PPV of the CoDE-HF rule-in score of 58.0 in patients without prior heart failure across patient subgroups. (C) PPV of the CoDE-HF rule-in score of 90.7 in patients with prior heart failure across patient subgroups.
Figure 4
Figure 4
Diagnostic performance of the CoDE-HF score across patient subgroups. CoDE-HF incorporates BNP concentrations as a continuous measure and predefined simple objective clinical variables (age, eGFR, haemoglobin, body mass index, heart rate, blood pressure, peripheral oedema, prior history of heart failure, COPD and ischaemic heart disease) to provide an individualized assessment of the likelihood of the diagnosis of acute heart failure. (A) NPV of the CoDE-HF rule-out score of 5.4 in patients without prior heart failure across patient subgroups. (B) PPV of the CoDE-HF rule-in score of 58.0 in patients without prior heart failure across patient subgroups. (C) PPV of the CoDE-HF rule-in score of 90.7 in patients with prior heart failure across patient subgroups.
See this image and copyright information in PMC

References

    1. National Heart Failure Audit 2019 Report. University College London: NICOR (National Institute for Cardiovascular Outcomes Research). 2019. https://www.nicor.org.uk/wp-content/uploads/2019/09/Heart-Failure-2019-R....
    1. Pang PS, Collins SP, Miró Ò, Bueno H, Diercks DB, Di Somma S, et al. Editor’s choice-the role of the emergency department in the management of acute heart failure: an international perspective on education and research. Eur Heart J Acute Cardiovasc Care 2017;6:421–429. - PubMed
    1. Roberts E, Ludman AJ, Dworzynski K, Al-Mohammad A, Cowie MR, McMurray JJV, et al. The diagnostic accuracy of the natriuretic peptides in heart failure: systematic review and diagnostic meta-analysis in the acute care setting. BMJ 2015;350:h910. - PMC - PubMed
    1. National Institute for Health and Care Excellence. Acute heart failure: diagnosis and management. Clinical Guideline 187. 2014. https://www.nice.org.uk/guidance/cg187.
    1. Januzzi JL Jr, Chen-Tournoux AA, Christenson RH, Doros G, Hollander JE, Levy PD, et al. N-Terminal Pro-B-Type natriuretic peptide in the emergency department: the ICON-RELOADED study. J Am Coll Cardiol 2018;71:1191–1200. - PubMed