CoChat
Sign Up
Cited by 96Privacy-Preserving Technologies in DataArtificial Intelligence in HealthcareInternet of Things and AI

Hybrid Classifier-Based Federated Learning in Health Service Providers for Cardiovascular Disease Prediction

Open access
Yaqoob, Muhammad Mateen, Nazir, Muhammad, Khan, Muhammad Amir, Qureshi, Sajida, Al-Rasheed, Amal·2023·Applied Sciences
DOI 10.3390/app13031911
Open this paper in CoChat — free accountOpen PDF

Abstract

One of the deadliest diseases, heart disease, claims millions of lives every year worldwide. The biomedical data collected by health service providers (HSPs) contain private information about the patient and are subject to general privacy concerns, and the sharing of the data is restricted under global privacy laws. Furthermore, the sharing and collection of biomedical data have a significant network communication cost and lead to delayed heart disease prediction. To address the training latency, communication cost, and single point of failure, we propose a hybrid framework at the client end of HSP consisting of modified artificial bee colony optimization with support vector machine (MABC-SVM) for optimal feature selection and classification of heart disease. For the HSP server, we proposed federated matched averaging to overcome privacy issues in this paper. We tested and evaluated our proposed technique and compared it with the standard federated learning techniques on the combined cardiovascular disease dataset. Our experimental results show that the proposed hybrid technique improves the prediction accuracy by 1.5%, achieves 1.6% lesser classification error, and utilizes 17.7% lesser rounds to reach the maximum accuracy.

AI snapshot

Written from the paper’s full text.

TL;DR

The paper proposes a hybrid federated learning framework for cardiovascular disease prediction, using MABC-SVM at health service provider clients and FedMA at the server. It reports improved accuracy, lower classification error, and fewer communication rounds than FedAvg-SVM, FedMA-SVM, and FedMA with GA-SVM.

Key findings

  • The proposed FedMA with MABC-RB-SVM achieved 93.8% accuracy, 94.2% precision, 96.6 sensitivity, 81.8 specificity, 90.1 F-measure, and 11.9 classification error on the combined dataset (Table 4).
  • Compared with baselines, the conclusion reports 1.5% greater accuracy, 1.6% lower classification error, 4.7% higher sensitivity, and 17.7% fewer rounds to reach maximum accuracy.
  • It reached 91% accuracy in 2408 rounds, versus 3810 for FedAvg-SVM, 3425 for FedMA-SVM, and 3046 for FedMA with GA-SVM; Table 3 lists a 37.8% difference for the proposed method.
  • The paper reports consuming 15–25% less communication size in GB for improved target accuracy and reaching 93.8% accuracy within 4500 communication rounds.

Methods

  • The framework uses M-ABC for optimal feature selection and RB-SVM for classification at HSP client nodes, while FedMA performs matched averaging and global model updates at the HSP server.
  • Simulation was run in Python/PyTorch on an Intel Core i7 @ 4 GHz with 16 GB RAM, using 5000 communication rounds, five HSP client nodes, one HSP server node, local epochs varied from 10 to 160, and communication volumes varied from 0.6 to 6.0 GB.
  • The dataset combines five heart disease datasets (Cleveland, Stalog, Hungary, Long Beach, Switzerland) with 918 entries and 11 common features; baselines are FedAvg-SVM, FedMA-SVM, and FedMA with GA-SVM.

Limitations

  • The paper does not state limitations. Evident from the text: evaluation is simulation-only with five HSP clients and one server, not a real clinical or HSP deployment.
  • Evident from the text: the dataset is a combined public dataset of 918 entries, and no external clinical validation is described.
  • Evident from the text: privacy is addressed by keeping CVD data local, but no formal privacy or security guarantees or adversarial analysis are provided.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/app13031911

Cites

41 of 41 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 3 cited sources checked · 0 supported · 0 flagged

  • Findingreasoning

    The number of rounds in the proposed technique is reduced by 37% compared to existing methods.

    “The number of rounds in our suggested technique is reduced by 37% when compared to existing methods, as shown in Table 3”
    Quote in the paperSubstantial claim
  • Findingown data

    The proposed hybrid technique utilizes 17.7% lesser rounds to reach maximum accuracy compared to standard federated learning techniques on the combined cardiovascular disease dataset.

    “Our experimental results show that the proposed hybrid technique improves the prediction accuracy by 1.5%, achieves 1.6% lesser classification error, and utilizes 17.7% lesser rounds to reach the maximum accuracy.”
    Quote in the paperSubstantial claim
  • Findingown data

    The proposed hybrid approach outperforms traditional FL techniques and FedMA with GA-SVM in terms of accuracy at both low and large communication volumes.

    “The recorded results show that our hybrid approach outperforms traditional FL techniques and FedMA with the GA-SVM in terms of accuracy at both low and large communication volumes.”
    Quote in the paperSubstantial claim
  • Findingown data

    The proposed hybrid technique improves prediction accuracy by 1.5% compared to standard federated learning techniques on the combined cardiovascular disease dataset.

    “Our experimental results show that the proposed hybrid technique improves the prediction accuracy by 1.5%, achieves 1.6% lesser classification error, and utilizes 17.7% lesser rounds to reach the maximum accuracy.”
    Quote in the paperSubstantial claim
  • Findingown data

    The proposed hybrid technique achieves 1.6% lesser classification error compared to standard federated learning techniques on the combined cardiovascular disease dataset.

    “Our experimental results show that the proposed hybrid technique improves the prediction accuracy by 1.5%, achieves 1.6% lesser classification error, and utilizes 17.7% lesser rounds to reach the maximum accuracy.”
    Quote in the paperSubstantial claim
  • May combine two claims. 12
  • Classified as probably not a substantial claim. 5

  • May combine two claims.

    Findingreasoning

    The proposed hybrid framework accomplishes better accuracy and reduces the amount of communication rounds used.

    “Consequently, the proposed hybrid framework accomplishes better accuracy and reduces the amount of communication rounds used.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The proposed framework reaches 93.8% accuracy within 4500 communication rounds, which is better than FedAvg-SVM, FedMA-SVM, and FedMA with GA-SVM.

    “Our proposed framework reaches 93.8% accuracy within the 4500 rounds of communication, which is better than the existing FedAvg-SVM, FedMA-SVM, and FedMA with GA-SVM algorithms.”
    • Also stated in: § 5.3. Results and Discussion · boosted
      “Table 3 demonstrates that the proposed framework performs better on the heart disease dataset than FL state-of-the-art methods in terms of prediction accuracy (93.8%)”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The proposed framework achieves precision 94.2%, sensitivity 96.6, and specificity 81.8 on the heart disease dataset.

    “Table 3 demonstrates that the proposed framework performs better on the heart disease dataset than FL state-of-the-art methods in terms of prediction accuracy (93.8%), precision (94.2%), sensitivity (96.6), and specificity (81.8)”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Findingnone given

    The learning rate is increased in the GA-SVM with FedMA, but the convergence is consumed more by the classification and feature selection.

    “The learning rate is increased in the GA-SVM with FedMA, but the convergence is consumed more by the classification and feature selection.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Performance metrics including prediction accuracy, classification error, sensitivity, precision, and communication efficiency all showed a considerable improvement under the suggested paradigm.

    “Performance metrics including prediction accuracy, classification error, sensitivity, precision, and communication efficiency all showed a considerable improvement under the suggested paradigm.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The suggested strategies produce outcomes with 1.5% greater accuracy, 1.6% lower classification error, 4.7% higher sensitivity, and 17.7% fewer rounds needed to reach the greatest degree of accuracy.

    “Our findings indicated that the suggested strategies produce outcomes with 1.5% greater accuracy, 1.6% lower classification error, 4.7% higher sensitivity, and 17.7% fewer rounds needed to reach the greatest degree of accuracy.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    In the case of GA-SVM with FedMA, the accuracy does not decrease much after 100 local epochs, which is attributable to the GA algorithm.

    “However, in the case of the GA-SVM with FedMA, the accuracy does not decrease much after 100 local epochs, which is attributable to the GA algorithm.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The suggested framework can train for a longer period and supports a higher rate of convergence because it produces a better HSP global model on the local model with a higher model quality.

    “The findings show that the suggested framework can train for a longer period and supports a higher rate of convergence because it produces a better HSP global model on the local model with a higher model quality.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Around the world, laws are implemented to protect the privacy of data.

    “Around the world, laws are implemented to protect the privacy of data [6][7][8].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [6][7][8]

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationreasoning

    FedMA is more suitable than alternative approaches for the FL-based HSP environment where there are few HSP client devices and data per device are insufficient for independent model training.

    “In the FL-based HSP environment, we have a small number of HSP client devices and the data for independent model training on each device are not sufficient; an FedMA approach is more suitable.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: In the FL-based HSP environment, there are few HSP client devices and the data on each device are insufficient for independent model training.

  • May combine two claims.

    Interpretationown data

    The proposed framework is more equipped to provide increased heart disease prediction accuracy while maintaining privacy compared to existing baseline FL techniques.

    “Our proposed framework is therefore more equipped to provide increased heart disease prediction accuracy while maintaining privacy when compared to existing baseline FL techniques.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The MABC optimization technique for feature selection and the RB-SVM classification in the proposed framework result in decreased classification errors.

    “In addition, the MABC optimization technique for feature selection and the RB-SVM classification in our proposed framework result in decreased classification errors, resulting in a classification error of 11.9 for our recommended method.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The proposed MABC-RB-SVM method employs the hybrid of the MABC optimizer and RB-SVM classifier for optimal feature selection and classification at HSP clients, and FedMA at the global orchestrator permits the overall model to accomplish better accuracy in a lesser number of communication rounds than existing FL algorithms.

    “Since the proposed MABC-RB-SVM method employs the hybrid of the MABC optimizer and RB-SVM classifier for optimal feature selection and classification at HSP clients, and for the HSP global orchestrator, we deploy the FedMA which permits our overall model to accomplish better accuracy in a lesser number of communication rounds than the existing FL algorithms.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    Training multiple models independently on every HSP client and averaging their predictions to produce a final prediction can lead to better performance by reducing overfitting and increasing the diversity of the models.

    “Our idea is to train multiple models independently on every HSP client and then average their predictions to produce a final prediction.This can lead to a better performance by reducing overfitting and increasing the diversity of the models.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    The combined dataset used comprises over eleven common features from the Cleveland, Stalog, Hungary, Long Beach, and Switzerland heart disease datasets and contains 918 entries.

    “Utilizing the combined dataset of five heart illness datasets, this dataset combines over eleven common features from the datasets of Cleveland, Stalog, Hungary, Long Beach, and Switzerland [43].”
    • Also stated in: § 5.2. Dataset Description · plain
      “There are 918 entries in the combined dataset collection, along with 76 characteristics in each dataset.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [43] Fedesoriano (2022, November 28). Heart Failure Prediction Dataset. Available online: https://www.kaggle.com/fedesoriano/heart-failure-prediction.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    The suggested inference approach is not reliant on parameters.

    “The total neurons in the federated model can gradually increase in accordance to the sizes of the HSP client models because our suggested inference approach is not reliant on parameters.”
    Quote in the paperNot a substantial claim
16 of 16

  • The paper relies on this without support.

    In the FL-based HSP environment, there are few HSP client devices and the data on each device are insufficient for independent model training.

    “In the FL-based HSP environment, we have a small number of HSP client devices and the data for independent model training on each device are not sufficient; an FedMA approach is more suitable.”

  • The paper relies on this without support.

    The sharing and collection of biomedical data have a significant network communication cost and lead to delayed heart disease prediction.

    “Furthermore, the sharing and collection of biomedical data have a significant network communication cost and lead to delayed heart disease prediction.”

  • The paper relies on this without support.

    Biomedical data collected by HSPs contain private information about the patient and are subject to general privacy concerns.

    “The biomedical data collected by health service providers (HSPs) contain private information about the patient and are subject to general privacy concerns, and the sharing of the data is restricted under global privacy laws.”

  • The paper relies on this without support.

    The hybrid M-ABC with firefly algorithm has global optimization abilities and shows better results than the original ABC or firefly algorithm alone.

    “This hybrid algorithm is known for its global optimization abilities and has been applied to various optimization problems, showing better results than the original ABC or firefly algorithm alone.”

  • The paper relies on this without support.

    SVM is less prone to overfitting and is successful in high-dimensional space.

    “In IoMT-based healthcare solutions, wearable devices are DL, SVM is less prone to overfitting and is known to be successful in high-dimensional space.”

  • The paper relies on this without support.

    The M-ABC is particularly effective for solving complex optimization issues that are hard to resolve with traditional optimization methods.

    “This method is particularly effective for solving complex optimization issues that are hard to resolve with traditional optimization methods.”

  • The paper relies on this without support.

    It is hard to exchange user data in healthcare systems due to privacy and security concerns.

    “Due to privacy and security concerns, it is hard to exchange user data when it comes to healthcare systems.”

  • The paper relies on this without support.

    Predicting survival in patients with cardiac disease is challenging.

    “It is challenging to predict survival in patients with cardiac disease.”
8 of 8

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

  • Classification error conflicts with reported accuracy

    Major

    The paper defines classification error as the proportion of mistaken predictions. An accuracy of 93.8% therefore implies a classification error of 6.2%, not 11.9%.

    Reported 11.9 · recomputed 6.2

1 of 1

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Yaqoob, M. M., Nazir, M., Khan, M. A., Qureshi, S., & Al-Rasheed, A.
Affiliations
  • HITEC University (PK)
  • COMSATS University Islamabad (PK)
  • ILMA University (PK)
  • Department of Information Systems, College of Computer and Information Sciences, Princess Nourah bint Abdulrahman University, Riyadh 11671, Saudi Arabia
Corresponding author
  • Yaqoob, Muhammad Mateen
  • Khan, Muhammad Amir
ORCID
Venue
Applied Sciences
Publisher
MDPI AG
Published
2023-02-01
Volume
13
Issue
3
Pages
1911
ISSN
2076-3417 (ISSN-L)
DOI
10.3390/app13031911
OpenAlex
W4319023769
Semantic Scholar
bb2f79498320fa0e16526900200a293185d96dc5
Cited by
96
References
40
Citations by year
2026: 14 · 2025: 40 · 2024: 24 · 2023: 18
FWCI
8.50
Type
journal article
Language
en
Open access
Licence
Locations
  • Applied Sciences (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia. (PNURSP2023R235)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
  • Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
  • Physical Sciences › Computer Science › Information Systems › Internet of Things and AI
Keywords
federated learning, heart disease prediction, cardiovascular disease, artificial bee colony algorithm, support vector machine, feature selection, privacy preservation, healthcare workers, communication overhead, training latency

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review