pith. sign in

arxiv: 1907.05888 · v1 · pith:4B3WJAY3new · submitted 2019-07-12 · 💻 cs.LG · cs.HC· cs.NA· eess.SP· math.NA· physics.med-ph

Regularized HessELM and Inclined Entropy Measurement for Congestive Heart Failure Prediction

Apdullah Yay{\i}k , Yakup Kutlu , G\"okhan Altan This is my paper

Pith reviewed 2026-05-24 22:25 UTC · model grok-4.3

classification 💻 cs.LG cs.HCcs.NAeess.SPmath.NAphysics.med-ph
keywords congestive heart failureECG signalsentropy measurementextreme learning machineHessenberg decompositionclassificationsignal processingmachine learning
0
0 comments X

The pith

Inclined entropy measurement features from ECG signals, paired with R-HessELM, predict congestive heart failure at 98.49% accuracy.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces regularized Hessenberg decomposition based extreme learning machine, called R-HessELM, together with squared, circled, inclined and grid entropy measurement models as features for ECG analysis. It sets out to show that the inclined entropy variant captures the key signal properties that separate congestive heart failure recordings from normal ones. When these features are supplied to the R-HessELM classifier the method reaches 98.49 percent overall accuracy, which would support an automated, high-precision route to CHF detection from standard electrocardiograms.

Core claim

Inclined entropy measurements features well represent characteristics of ECG signals and together with R-HessELM approach overall accuracy of 98.49% was achieved for CHF prediction.

What carries the argument

Regularized Hessenberg decomposition based extreme learning machine (R-HessELM) driven by inclined entropy measurement features from ECG signals.

Load-bearing premise

The inclined entropy measurements accurately capture the distinguishing characteristics of ECG signals for CHF without requiring additional validation or being specific to the dataset used.

What would settle it

Testing the identical R-HessELM pipeline and inclined entropy features on an independent multi-center ECG collection and obtaining accuracy well below 98 percent would falsify the performance claim.

Figures

Figures reproduced from arXiv: 1907.05888 by Apdullah Yay{\i}k, G\"okhan Altan, Yakup Kutlu.

Figure 4
Figure 4. Figure 4: In the input layer, weights and biases are assigned [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
read the original abstract

Our study concerns with automated predicting of congestive heart failure (CHF) through the analysis of electrocardiography (ECG) signals. A novel machine learning approach, regularized hessenberg decomposition based extreme learning machine (R-HessELM), and feature models; squared, circled, inclined and grid entropy measurement were introduced and used for prediction of CHF. This study proved that inclined entropy measurements features well represent characteristics of ECG signals and together with R-HessELM approach overall accuracy of 98.49% was achieved.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The paper proposes a regularized HessELM (R-HessELM) classifier together with four novel entropy-based feature extractors (squared, circled, inclined, and grid entropy measurement) applied to ECG signals for automated congestive heart failure (CHF) prediction. It asserts that the inclined entropy variant best captures distinguishing ECG characteristics and, when paired with R-HessELM, yields 98.49% overall accuracy.

Significance. If the performance claims are substantiated by proper validation, the work could introduce a new feature family and a regularized ELM variant useful for ECG-based cardiac classification tasks. The explicit comparison among four entropy formulations is a positive element, but the absence of dataset details, cross-validation protocols, and external benchmarks currently prevents assessment of whether the result advances the field beyond existing entropy-based CHF detectors.

major comments (3)
  1. [Abstract] Abstract: the central performance claim of 98.49% accuracy is presented with no accompanying information on dataset identity or size, cross-validation procedure, number of subjects, or statistical testing, so the claim cannot be evaluated for soundness or reproducibility.
  2. [Abstract] Abstract and Results: the assertion that 'inclined entropy measurements features well represent characteristics of ECG signals' is justified solely by the achieved accuracy; this is circular because the free parameters (regularization parameter and number of hidden neurons) are selected on the same data, and no comparison is made to established measures such as approximate entropy or sample entropy.
  3. [Methods] Methods/Experiments: no patient-wise hold-out, no cross-dataset evaluation, and no external cohort are described, so it remains possible that the reported accuracy reflects dataset idiosyncrasies rather than general CHF-discriminating properties of the inclined entropy features.
minor comments (1)
  1. [Methods] Notation for the four entropy variants is introduced without a compact mathematical definition or pseudocode, making replication difficult.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for their detailed review and constructive suggestions. We provide point-by-point responses to the major comments and indicate the revisions we will make to address the concerns regarding reproducibility and validation.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central performance claim of 98.49% accuracy is presented with no accompanying information on dataset identity or size, cross-validation procedure, number of subjects, or statistical testing, so the claim cannot be evaluated for soundness or reproducibility.

    Authors: We agree that including these details in the abstract would improve the manuscript's clarity and allow better evaluation of the results. In the revised version, we will update the abstract to briefly mention the dataset used (from PhysioNet), the number of subjects, the cross-validation method (e.g., 10-fold), and any statistical measures. The detailed description remains in the Methods and Experiments sections. revision: yes

  2. Referee: [Abstract] Abstract and Results: the assertion that 'inclined entropy measurements features well represent characteristics of ECG signals' is justified solely by the achieved accuracy; this is circular because the free parameters (regularization parameter and number of hidden neurons) are selected on the same data, and no comparison is made to established measures such as approximate entropy or sample entropy.

    Authors: The assertion is supported by the comparative results among the four novel entropy measures introduced in the paper, where the inclined variant consistently outperformed the others. To mitigate concerns about parameter selection, we will clarify in the revision that hyperparameters were optimized using cross-validation to prevent overfitting on the test data. Regarding comparisons to established measures, we will include a discussion or additional experiments comparing inclined entropy to approximate entropy and sample entropy to provide a more comprehensive evaluation. revision: partial

  3. Referee: [Methods] Methods/Experiments: no patient-wise hold-out, no cross-dataset evaluation, and no external cohort are described, so it remains possible that the reported accuracy reflects dataset idiosyncrasies rather than general CHF-discriminating properties of the inclined entropy features.

    Authors: The study employed k-fold cross-validation on the single dataset available. We will revise the Methods section to explicitly describe the validation protocol, including any subject-level considerations. While cross-dataset evaluation and external cohorts were not part of this work, we will add a limitations section acknowledging this and discussing the potential for future validation on additional cohorts to confirm generalizability. revision: partial

standing simulated objections not resolved
  • The absence of an external cohort or cross-dataset evaluation in the original experiments, which would require additional data not used in the study.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper defines four entropy measurement variants (squared, circled, inclined, grid) as feature extractors from ECG signals, introduces the R-HessELM classifier, applies them to a CHF dataset, and reports 98.49% accuracy for the inclined variant. The assertion that inclined entropy features represent ECG characteristics follows directly from this empirical performance metric rather than from any definitional loop, fitted parameter renamed as prediction, or self-citation chain. No equations or steps reduce a claimed result to its own inputs by construction; the chain from feature definition through classification to accuracy is independent and externally falsifiable on held-out data.

Axiom & Free-Parameter Ledger

2 free parameters · 0 axioms · 0 invented entities

Only abstract available; standard ML hyperparameters are expected but not specified.

free parameters (2)
  • regularization parameter
    The regularization in R-HessELM is likely a hyperparameter fitted or chosen for the model.
  • number of hidden neurons
    Typical in ELM, the number of hidden nodes is a free parameter.

pith-pipeline@v0.9.0 · 5637 in / 1160 out tokens · 42354 ms · 2026-05-24T22:25:19.641942+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages

  1. [1]

    NeuralComputingandApplications 28(10),3073–3094(2017)[8]

    Acharya,U.R.,Fujita,H.,Sudarshan,V.K.,Oh,S.L.,Muhammad, A., Koh, J.E., Tan, J.H., Chua, C.K., Chua, K.P., San Tan, R.: Applicationofempiricalmodedecomposition(emd)forautomated identification of congestive heart failure using heart rate signals. NeuralComputingandApplications 28(10),3073–3094(2017)[8]

    work page 2017

  2. [2]

    Goldberger and coworkers: PhysioBank, PhysioToolkit, and PhysioNet:ComponentsofaNewResearchResourceforComplex Physiologic Signals

    A.L. Goldberger and coworkers: PhysioBank, PhysioToolkit, and PhysioNet:ComponentsofaNewResearchResourceforComplex Physiologic Signals. Circulation 101(23):e215-e220 (2000) [2]

    work page 2000

  3. [3]

    Altan, G., Kutlu, Y., Allahverdi, N.: A new approach to early diagnosis of congestive heart failure disease by using Hilbert– Huangtransform.Computermethodsandprogramsinbiomedicine 137, 23–34 (2016) [2]

    work page 2016

  4. [4]

    JournalofTheoreticalandAppliedInfor- mation Technology pp

    Anuradha, B., Reddy, V.C.V.: Cardiac Arrhythmia Classification UsingFuzzyClassifiers. JournalofTheoreticalandAppliedInfor- mation Technology pp. 353–359 (2008) [1]

    work page 2008

  5. [5]

    25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Cancun, pp

    Asyali, M.H.: Discrimination Power of Long-Term Heart Rate Variability Measures. 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Cancun, pp. 4–7 (2003) [8]

    work page 2003

  6. [6]

    Neural networks81, 91–102 (2016) [2, 5]

    Cao,J.,Zhang,K.,Luo,M.,Yin,C.,Lai,X.:Extremelearningma- chine and adaptive sparse representation for image classification. Neural networks81, 91–102 (2016) [2, 5]

    work page 2016

  7. [7]

    Donkor, A., McDonagh, T., Hardman, S.: British Heart Failure Report. Tech. rep., Institute of Cardiovascular Science (2016) [1]

    work page 2016

  8. [8]

    NumerischeMathematik 14(5),403–420(1970) [4] 8 Apdullah Yayık et al

    Golub, G.H., Reinsch, C.: Singular value decomposition and least squaressolutions. NumerischeMathematik 14(5),403–420(1970) [4] 8 Apdullah Yayık et al. Study Class Feature Model Feature Size Classifier Overall Accuracy (%) Asyalı, 2003 [5] Normal CHF Standard Deviation Normal to Normal - BayesNET 89.95 İşler and Kuntalp, 2007 [14] Normal CHF Wavelet Entrop...

    work page 1970

  9. [9]

    In: Cardiovascular Oscillations(ESGCO),20148thConferenceoftheEuropeanStudy Group on, pp

    Heinze, C., Sommer, D., Trutschel, U., Golz, M.: Discrimination and relevance determination of heart rate variability features for the identification of congestive heart failure. In: Cardiovascular Oscillations(ESGCO),20148thConferenceoftheEuropeanStudy Group on, pp. 219–220. IEEE (2014) [8]

    work page 2014

  10. [10]

    Expert Systems with Applications39(2), 2047– 2058 (2012)

    Homaeinezhad, M., Atyabi, S., Tavakkoli, E., Toosi, H., Ghaffari, A., Ebrahimpour, R.: ECG arrhythmia recognition via a neuro- SVM–KNN hybrid classifier with virtual QRS image-based geo- metrical features. Expert Systems with Applications39(2), 2047– 2058 (2012). DOI 10.1016/j.eswa.2011.08.025 [1]

    work page doi:10.1016/j.eswa.2011.08.025 2047

  11. [11]

    Neurocomputing102, 31–44 (2013) [4]

    Horata,P.,Chiewchanwattana,S.,Sunat,K.:Robustextremelearn- ing machine. Neurocomputing102, 31–44 (2013) [4]

    work page 2013

  12. [12]

    Biomedical Signal Processing and Control2(2), 135–143 (2007)

    Hossen, A., Al-Ghunaimi, B.: A wavelet-based soft decision tech- nique for screening of patients with congestive heart failure. Biomedical Signal Processing and Control2(2), 135–143 (2007). DOI 10.1016/j.bspc.2007.05.008 [1]

    work page doi:10.1016/j.bspc.2007.05.008 2007

  13. [13]

    Huang, Q.-Y

    Huang, G.b., Zhu, Q.y., Siew, C.k.: Extreme learning machine : Theory and applications. Neurocomputing70, 489–501 (2006). DOI 10.1016/j.neucom.2005.12.126 [3, 4]

    work page doi:10.1016/j.neucom.2005.12.126 2006

  14. [14]

    Computers in biology and medicine 37(10), 1502–10 (2007)

    Işler, Y., Kuntalp, M.: Combining classical HRV indices with wavelet entropy measures improves to performance in diagnos- ing congestive heart failure. Computers in biology and medicine 37(10), 1502–10 (2007). DOI 10.1016/j.compbiomed.2007.01. 012 [1, 8]

    work page doi:10.1016/j.compbiomed.2007.01 2007

  15. [15]

    Masset, R

    Kamath, C.: A new approach to detect congestive heart failure using sequential spectrum of electrocardiogram signals. Medical engineering & physics34(10), 1503–9 (2012). DOI 10.1016/j. medengphy.2012.03.001 [8]

    work page doi:10.1016/j 2012

  16. [16]

    Medical engineering & physics34(7), 841–8 (2012)

    Kamath, C.: A new approach to detect congestive heart failure using Teager energy nonlinear scatter plot of R-R interval series. Medical engineering & physics34(7), 841–8 (2012). DOI 10. 1016/j.medengphy.2011.09.026 [1]

    work page 2012

  17. [17]

    Journal of electrocardiology43(6), 719–24 (2010)

    Karmakar, C.K., Gubbi, J., Khandoker, A.H., Palaniswami, M.: AnalyzingtemporalvariabilityofstandarddescriptorsofPoincaré plots. Journal of electrocardiology43(6), 719–24 (2010). DOI 10.1016/j.jelectrocard.2010.09.001 [1]

    work page doi:10.1016/j.jelectrocard.2010.09.001 2010

  18. [18]

    PloS one9(4), e93,399 (2014) [8]

    Liu,G.,Wang,L.,Wang,Q.,Zhou,G.,Wang,Y.,Jiang,Q.:Anew approach to detect congestive heart failure using short-term heart Regularized HessELM and Inclined Entropy Measurement for Congestive Heart Failure Prediction 9 rate variability measures. PloS one9(4), e93,399 (2014) [8]

    work page 2014

  19. [19]

    International journal of medical informatics108, 55–63 (2017) [8]

    Mahajan, R., Viangteeravat, T., Akbilgic, O.: Improved detection ofcongestiveheartfailureviaprobabilisticsymbolicpatternrecog- nition and heart rate variability metrics. International journal of medical informatics108, 55–63 (2017) [8]

    work page 2017

  20. [20]

    Knowledge-Based Systems45, 76–82 (2013)

    Martis,R.J.,Acharya,U.R.,Lim,C.M.,Suri,J.S.:Characterization ofECGbeatsfromcardiacarrhythmiausingdiscretecosinetrans- form in PCA framework. Knowledge-Based Systems45, 76–82 (2013). DOI 10.1016/j.knosys.2013.02.007 [2]

    work page doi:10.1016/j.knosys.2013.02.007 2013

  21. [21]

    Engineering in Medicine and Biology pp

    Maurice E.Cohen, D.L.H., Deedwania, P.C.: Applying Continu- ous chaotic Modeling to Cardiac Signal Analysis. Engineering in Medicine and Biology pp. 97–102 (1996) [1]

    work page 1996

  22. [22]

    Expert Systems with Applications37(4), 3088–3093 (2010)

    Moavenian, M., Khorrami, H.: A qualitative comparison of Arti- ficial Neural Networks and Support Vector Machines in ECG ar- rhythmias classification. Expert Systems with Applications37(4), 3088–3093 (2010). DOI 10.1016/j.eswa.2009.09.021 [1]

    work page doi:10.1016/j.eswa.2009.09.021 2010

  23. [23]

    Com- puters in biology and medicine45, 72–79 (2014) [8]

    Narin, A., Isler, Y., Ozer, M.: Investigating the performance im- provement of hrv indices in chf using feature selection methods based on backward elimination and statistical significance. Com- puters in biology and medicine45, 72–79 (2014) [8]

    work page 2014

  24. [24]

    Computers in biology and medicine 43(10), 1556–62 (2013)

    Orhan, U.: Real-time CHF detection from ECG signals using a novel discretization method. Computers in biology and medicine 43(10), 1556–62 (2013). DOI 10.1016/j.compbiomed.2013.07. 015 [8]

    work page doi:10.1016/j.compbiomed.2013.07 2013

  25. [25]

    European heart journal37(27), 2129–2200 (2016) [1]

    Ponikowski, P., Voors, A.A., Anker, S.D., Bueno, H., Cleland, J.G.F., Coats, A.J.S., Falk, V., González-Juanatey, J.R., Harjola, V.P., Jankowska, E.A., Others: 2016 ESC Guidelines for the diag- nosis and treatment of acute and chronic heart failure: The Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society o...

    work page 2016

  26. [26]

    Measurement46(9), 3238– 3246 (2013)

    Rai, H.M., Trivedi, A., Shukla, S.: ECG signal processing for ab- normalitiesdetectionusingmulti-resolutionwavelettransformand Artificial Neural Network classifier. Measurement46(9), 3238– 3246 (2013). DOI 10.1016/j.measurement.2013.05.021 [1]

    work page doi:10.1016/j.measurement.2013.05.021 2013

  27. [28]

    Cardiology research and practice2009, 807,379 (2010)

    Thuraisingham, R.: A Classification System to Detect Congestive HeartFailureUsingSecond-OrderDifferencePlotofRRIntervals. Cardiology research and practice2009, 807,379 (2010). DOI 10.4061/2009/807379 [8]

    work page doi:10.4061/2009/807379 2010

  28. [29]

    Artificial intelli- gence in medicine33(3), 237–50 (2005)

    Tsipouras, M.G., Fotiadis, D.I., Sideris, D.: An arrhythmia classi- fication system based on the RR-interval signal. Artificial intelli- gence in medicine33(3), 237–50 (2005). DOI 10.1016/j.artmed. 2004.03.007 [1]

    work page doi:10.1016/j.artmed 2005

  29. [30]

    Journal of Applied Mathematics (2013) [4]

    Tzeng, J.: Split-and-combine singular value decomposition for large-scale matrix. Journal of Applied Mathematics (2013) [4]

    work page 2013

  30. [31]

    PatternRecognitionLetters 28(5), 592–603 (2007) [8]

    Übeyli, E.D., Güler, İ.: Features extracted by eigenvector methods fordetectingvariabilityofeegsignals. PatternRecognitionLetters 28(5), 592–603 (2007) [8]

    work page 2007

  31. [32]

    Yayik, A.: Enhancing Extreme Learning Machine: Novel Exten- sions and Applications to Optimizing Visual Stimuli for Brain Computer Interface. Ph.D. thesis, Mustafa Kemal University, GraduateSchoolandAppliedSciences,DepartmentofInformatics Doctorate Thesis (2017) [5]

    work page 2017

  32. [33]

    In:201220thSignalProcessingandCommuni- cations Applications Conference, SIU 2012, Proceedings (2012)

    Yayik, A., Kutlu, Y.: Diagnosis of congestive heart failure using Poincaremapplot. In:201220thSignalProcessingandCommuni- cations Applications Conference, SIU 2012, Proceedings (2012). DOI 10.1109/SIU.2012.6204457 [1, 2]

    work page doi:10.1109/siu.2012.6204457 2012

  33. [34]

    Computers in biology and medicine42(8), 816–825 (2012) [8]

    Yu, S.N., Lee, M.Y.: Bispectral analysis and genetic algorithm for congestiveheartfailurerecognitionbasedonheartratevariability. Computers in biology and medicine42(8), 816–825 (2012) [8]

    work page 2012

  34. [35]

    Computer methods and programs in biomedicine 108(1), 299–309 (2012) [8]

    Yu, S.N., Lee, M.Y.: Conditional mutual information-based fea- ture selection for congestive heart failure recognition using heart rate variability. Computer methods and programs in biomedicine 108(1), 299–309 (2012) [8]

    work page 2012

  35. [36]

    Computer methods and programs in biomedicine 122(3), 372–383 (2015) [2]

    Zheng,Y.,Guo,X.,Qin,J.,Xiao,S.:Computer-assisteddiagnosis forchronicheartfailurebytheanalysisoftheircardiacreserveand heart sound characteristics. Computer methods and programs in biomedicine 122(3), 372–383 (2015) [2]

    work page 2015

  36. [37]

    Computermeth- ods and programs in biomedicine111(3), 570–7 (2013)

    Zidelmal, Z., Amirou, A., Ould-Abdeslam, D., Merckle, J.: ECG beatclassificationusingacostsensitiveclassifier. Computermeth- ods and programs in biomedicine111(3), 570–7 (2013). DOI 10.1016/j.cmpb.2013.05.011 [1]

    work page doi:10.1016/j.cmpb.2013.05.011 2013