Logo: to the web site of Uppsala University

uu.sePublications from Uppsala University
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
High variability of the R-wave amplitude predicts incident heart failure in the elderly: a cohort study using machine learning
Uppsala University, Disciplinary Domain of Medicine and Pharmacy, Faculty of Medicine, Department of Medical Sciences.
Uppsala University, Disciplinary Domain of Medicine and Pharmacy, Faculty of Medicine, Department of Medical Sciences.
Uppsala University, Disciplinary Domain of Medicine and Pharmacy, Faculty of Medicine, Department of Medical Sciences.
(English)Manuscript (preprint) (Other academic)
Abstract [en]

Aims: Early identification of individuals at risk for heart failure (HF) may improve their poor prognosis. The aim was to test if a prediction model of ECG variables added to traditional risk factors could improve the prediction of incident HF versus traditional risk factors alone.

Methods and Results: The PIVUS (Prospective Investigation of the Vasculature in Uppsala Seniors) study (1016 individuals aged 70 years) was used for analysis of 23 ECG variables measured in 6 precordial leads. Out of 6 machine learning models used in a training dataset, the one with the best accuracy was used for the testing dataset. 

During 15 years of follow-up, 107 of 836 included individuals at risk were diagnosed with HF. Adding the 8 best ECG variables, identified by random forest in the training dataset, to traditional risk factors resulted in an improvement of the area under the ROC curve by 11.7% (p=0.0043) compared to the traditional risk factor model alone. A high beat-to-beat variation of the R amplitude (SD Ramp) in V1 was the most powerful predictive ECG variable. A decreased low Frequency/high frequency ratio, a heart rate variability index, was correlated to a high SD Ramp in V1 (p=0.002). 

Conclusion: Adding ECG variables to traditional cardiovascular risk factors were valuable for prediction of incident HF in an elderly population. The improvement in C statistics by adding the 8 identified ECG variables was quite substantial, which if reproduced in other populations, might be used as a screening tool for HF risk in clinical practice.

 

       

National Category
Cardiac and Cardiovascular Systems
Identifiers
URN: urn:nbn:se:uu:diva-518486OAI: oai:DiVA.org:uu-518486DiVA, id: diva2:1821219
Available from: 2023-12-19 Created: 2023-12-19 Last updated: 2024-01-09
In thesis
1. Risk factors for incident heart failure and atrial fibrillation in an elderly population: The role of cardiac conduction and heart rate variability
Open this publication in new window or tab >>Risk factors for incident heart failure and atrial fibrillation in an elderly population: The role of cardiac conduction and heart rate variability
2024 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

Heart failure (HF) and atrial fibrillation (AF) are epidemic diseases, frequently coexisting, sharing risk factors and conferring poor prognosis. Identification of individuals at high risk of HF and AF may enable early treatment and improve the prognosis. Reliable prediction models for daily clinical practice are lacking. Early modification and treatment of risk factors may reduce the incidence of AF and HF. Because atrial structure and function abnormalities increase the risk of AF, ECG indices reflecting atrial pathology may prove useful in predicting AF and HF.

The main objectives were to evaluate whether:

  • P-wave duration (Pdur) and PR-interval in V1 predicted incident HF and incident AF (Paper I-II)
  • low frequency/high frequency (L-F/H-F) ratio, a marker of autonomic balance, predicted incident HF (Paper IV)
  • combining selected ECG variables or the L-F/H-F ratio with traditional risk factors improved the performance of the traditional HF prediction model (Paper III-IV).

The Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) with 15 years of follow-up was used for all four studies. After applying the exclusion criteria, 836 subjects were evaluated for incident HF (Paper I, III-IV) and 877 subjects for incident AF (Paper II). Cox proportional hazard analysis related ECG-derived variables to incident HF and incident AF. Study III used machine learning to determine which ECG variables correlated to incident HF. C-statistic was used to test whether adding selected ECG variables to traditional HF risk factors improved the performance of the HF prediction model.

Short Pdur was significantly associated with incident HF (Paper I) and incident AF (Paper II). Of 134 ECG variables, high R-wave amplitude variation (SD Ramp) had the highest predictive value for HF (Paper III). A decreased L-F/H-F ratio significantly predicted HF (Paper IV). Adding eight selected ECG variables (Paper III) and the L-F/H-F ratio (Paper IV) to the traditional risk factors significantly improved HF predictive performance by 11.7% and 3.3%, respectively.

In conclusion, the ECG may prove useful for predicting incident HF and AF beyond the traditional risk factors. An autonomic imbalance may precede the development of HF.

Place, publisher, year, edition, pages
Uppsala: Acta Universitatis Upsaliensis, 2024. p. 57
Series
Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine, ISSN 1651-6206 ; 2007
Keywords
incident heart failure, incident atrial fibrillation, prediction of heart failure, short P-wave duration, heart rate variability.
National Category
Cardiac and Cardiovascular Systems
Research subject
Cardiology
Identifiers
urn:nbn:se:uu:diva-518489 (URN)978-91-513-2002-1 (ISBN)
Public defence
2024-03-19, Enghoffsalen, Ing 50, Akademiska Sjukhuset, Uppsala, 13:00 (Swedish)
Opponent
Supervisors
Available from: 2024-02-13 Created: 2023-12-19 Last updated: 2024-03-18

Open Access in DiVA

No full text in DiVA

By organisation
Department of Medical Sciences
Cardiac and Cardiovascular Systems

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 17 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf