Pith. sign in

REVIEW 4 major objections 6 minor 42 references

Development of a Multiprocessing Interface Genetic Algorithm for Optimising a Multilayer Perceptron for Disease Prediction

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A multiprocessing genetic algorithm tunes a multilayer perceptron to 99.12% accuracy on breast cancer, 94.87% on Parkinson's, and 100% on chronic kidney disease, and cuts tuning time by about 60%.

desk verdict Routine GA-MLP tuning paper with a circular test-set fitness that invalidates its headline accuracies and a speedup table that doesn't add up. read the letter →

arxiv 2506.15694 v1 pith:ROWMUXDG submitted 2025-05-27 cs.LG

Iliyas Ibrahim Iliyas , Souley Boukari , Abdulsalam Yau Gital This is my paper
classification cs.LG
keywords multilayerperceptronmultiprocessinggeneticalgorithmhyperparameteroptimizationkernelprincipalcomponentanalysisparallelprocessingbreastcancerParkinson'sdiseasechronickidney
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that a disease-prediction pipeline can be made both more accurate and faster to tune by combining kernel PCA for nonlinear feature reduction with a multilayer perceptron whose hyperparameters are chosen by a modified multiprocessing genetic algorithm (MIGA). The MIGA evaluates candidate hyperparameter configurations in parallel across CPU threads, which the authors report cuts tuning time by roughly 60% relative to a single-threaded genetic algorithm. On three standard medical datasets the tuned MLP reaches 99.12% accuracy on breast cancer, 94.87% on Parkinson's disease, and 100% on chronic kidney disease, outperforming grid search, random search, and Bayesian optimization in the paper's comparisons. If true, this would give clinicians and researchers a largely automatic, much cheaper route to high-accuracy tabular disease classifiers.

What carries the argument

The load-bearing mechanism is the multiprocessing interface genetic algorithm (MIGA), a genetic algorithm whose fitness-evaluation step—training and scoring an MLPClassifier on the test set for each candidate hyperparameter configuration—is executed in parallel across multiple threads instead of sequentially. Each chromosome is the tuple [hidden-layer size, activation function, learning rate, solver]; fitness is the accuracy score on the test set; the top 50% of the population is kept by elitist selection, recombined by uniform crossover, and mutated with probability 0.1. The parallel evaluation is what the paper credits for the roughly 60% reduction in tuning time, while the preceding kernel PCA stage supplies the nonlinear features that the authors say improve classification relative to a standard GA.

What would settle it

Re-run the MIGA pipeline on all three datasets with the test set locked away until after tuning: optimize with a separate validation split as fitness, then measure accuracy once on the untouched test set. If held-out accuracy falls materially below 99.12% (breast cancer), 94.87% (Parkinson's), and 100% (CKD), the reported numbers were an artifact of selecting on the test set. A cheaper check is to repeat the 80/20 split with different random seeds and record how much the MIGA-selected configuration's accuracy varies across folds.

Watch

Extended reading notes

Core claim

The paper's central claim is that its proposed framework—kernel PCA with a radial basis function kernel keeping 95% of the variance, a multilayer perceptron classifier, and a multiprocessing genetic algorithm (MIGA) that parallelizes hyperparameter search—delivers high disease-classification accuracy at sharply reduced tuning cost. Each candidate MLP configuration is encoded as a chromosome (hidden-layer size, activation, learning rate, solver), and a population of 10 is evolved over 10 generations with elitist selection, uniform crossover, and a 10% mutation rate, while fitness evaluations run in parallel through a thread pool. The reported best accuracies are 99.12% for the Wisconsin Diagnostic Breast Cancer dataset, 94.87% for the Parkinson's telemonitoring dataset, and 100% for the chronic kidney disease dataset, with tuning time reduced by approximately 60% compared with a standard single-threaded GA. The per-generation logs show the best configuration typically appearing within the first two generations and stabilizing afterward, which the authors present as evidence of efficient convergence.

Load-bearing premise

The load-bearing premise is that using test-set accuracy as the genetic algorithm's fitness function—stated in Table 2—still gives an unbiased estimate of final performance, since the same test split both steers the search and is later reported as the result.

Editorial extensions

If this is right

  • Evolutionary hyperparameter search becomes practical in compute- or time-constrained settings: the same search quality costs about 40% of the wall-clock time, so facilities with limited hardware could run GA tuning where it was previously too slow.
  • The bundled graphical interface means the full pipeline—loading a CSV, choosing the target, reducing dimensions, tuning, and predicting—can be operated without writing code, which is the paper's stated route to clinical adoption.
  • The per-generation tables indicate the best configuration appears by generation 2 on these datasets, so a short evolutionary search captures most of the accuracy gain and longer runs mainly confirm the plateau.
  • Kernel PCA contributes to the result as well as the GA: the paper reports that the nonlinear reduction improved classification where a standard GA without it did not, so both components—feature extraction and optimizer—carry the performance.
  • On all three datasets the MIGA-tuned MLP matches or beats the cited published baselines, including Bayesian-optimized SVM, PSO-tuned CNN, and GA-tuned neural networks, which is the paper's evidence that the framework is competitive with established tuning methods.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Not explored by the authors but a direct consequence: the MIGA wrapper only needs a fitness function, so the same parallel-tuning interface should transfer to SVM, logistic regression, or gradient-boosted tree classifiers on the same datasets, where per-evaluation cost is lower than an MLP's.
  • The reported ~60% speedup is measured against a single-threaded GA; a matched-budget comparison against parallelized Bayesian optimization or random search with early stopping would be a fairer practical yardstick and is not reported.
  • Because the best configuration emerges by generation 2, the time saved by parallelization could be reinvested in larger populations or more generations to search harder datasets—an option the paper leaves implicit.
  • The paper's own tables do not fully agree—breast-cancer accuracy appears as both 99.12% and 99.00%, and the CKD optimal configuration differs between the text and Table 6—so a reader benchmarking this framework should reconcile the reported figures first.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This manuscript presents MIGA, a genetic algorithm with parallel fitness evaluation for tuning an MLP classifier, combined with RBF kernel PCA for dimensionality reduction. The framework is evaluated on three public medical datasets (Wisconsin Diagnostic Breast Cancer, Parkinson's Telemonitoring, and chronic kidney disease), reporting best accuracies of 99.12%, 94.87%, and 100%, a tuning-time reduction of about 60%, and a GUI for clinicians. The paper claims these results outperform grid search, random search, and Bayesian optimization.

Significance. The proposed pipeline is clearly described and the use of public datasets makes the application concrete; a low-code GUI for hyperparameter tuning would be practically useful if the underlying claims held. However, the central quantitative claims are supported only by a circular evaluation protocol and contain internal inconsistencies, and no code or reproducibility package is provided. As a result, the current evidence does not establish the claimed superiority or speedup.

major comments (4)
  1. [Table 2; Algorithm 1, Section 3.5.1] The GA fitness function is defined as accuracy on the test set (Table 2), and Algorithm 1 uses that same fitness to track the best individual. The best accuracies in Tables 3–5 are therefore the maximum of the test-set accuracy over 10 generations, i.e., the optimized objective on the reported split. This makes the headline accuracies selection artifacts, not independent estimates of generalization. Please re-run with fitness evaluated on a validation split and report final performance on a held-out test set (or use nested cross-validation), and ensure all compared optimizers use the same protocol.
  2. [Section 4, text near Table 6 and Table 6] The optimal configurations reported in the text do not match Table 6: the text gives 50 hidden units for breast cancer while Table 6 gives 100; the text gives 50 hidden units for CKD while Table 6 gives 100; and the Parkinson's accuracy in Table 6 is 95.00% although Table 3 reports a best of 0.9487. These inconsistencies mean the exact configuration behind each headline accuracy is not uniquely specified, so the results are not reproducible as written.
  3. [Section 3.5; Table 2; Table 7 (timing)] The claimed 59–61% time reduction relies on 'multithreading' via ThreadPoolExecutor, which under CPython does not provide true parallelism for CPU-bound Python-level MLP fitting because of the GIL. No profiling, hardware details, per-generation timings, or repeated runs are provided, so the speedup is not supported. The authors should either use process-based parallelism (ProcessPoolExecutor) or demonstrate with profiling and core-utilization data that the threaded execution actually parallelizes the workload.
  4. [Abstract; Section 4, second Table 7] The abstract claims that MIGA outperforms grid search, random search, and Bayesian optimization, but no controlled comparison with these methods under identical splits appears in the manuscript. The second Table 7 lists accuracies from unrelated studies that use different datasets, preprocessing, and evaluation protocols. A direct comparison on the same train/test splits with identical metrics is required to support the superiority claim.
minor comments (6)
  1. [Declarations / Reference [20]] Reference [20] for the Parkinson's dataset points to the UCI chronic kidney disease page rather than the Parkinson's Telemonitoring dataset.
  2. [Section 4] Table 7 is used twice (timing logs and comparison with other models); the second table should be renumbered.
  3. [Section 3.5] The algorithm is called 'Multi-Processing' but the implementation described in the text and Table 2 uses multithreading; the terminology should be reconciled.
  4. [Section 3.1.2] The Parkinson's Telemonitoring dataset is originally a regression dataset; please state how the binary status target was derived and report class balance, since this affects accuracy interpretation.
  5. [Throughout] There are frequent typos, including 'Parkison'/'Parkision' for Parkinson, 'stage of the art' for state of the art, and 'treating breast cancer' in a figure caption; a careful proofread is needed.
  6. [Section 3.2/3.3] Please clarify whether kernel PCA was fitted on the training split only; if fitted on the entire dataset before splitting, the reported results would be optimistically biased.

Circularity Check

1 steps flagged · score 6.0 of 10

Reported accuracies are the GA fitness values on the same test split, so the headline numbers are selection maxima rather than independent predictions.

  1. fitted input called prediction [Section 3.5.1, Table 2 (Genetic Algorithm Settings); Algorithm 1; Section 4, Tables 3–5]
    "The fitness function was defined as the accuracy score on the test dataset, and evaluations were executed in parallel via ThreadPoolExecutor to accelerate computation. Each chromosome encodes a complete model configuration represented by the tuple [hidden layer size, activation function, learning rate, solver]. ... Track the best solution: If any θ ∈ Pg+1 has better fitness than θ∗, update θ∗ ← θ."

    The GA's fitness is test-set accuracy, and Algorithm 1 returns the individual with the best such fitness. Tables 3–5 then report that same best fitness as the final accuracy (0.9487 for Parkinson's, 0.9912 for breast cancer, 1.0 for CKD). The headline accuracies are therefore the maximum of the exact objective being optimized on the exact split later reported; they are selection artifacts, not independent holdout estimates. The comparison against grid search, random search, and Bayesian optimization is also uninterpretable unless all methods used a common held-out protocol, which the paper does not document.

full rationale

The central accuracy claims reduce by construction to the GA's fitness function: Table 2 states 'Fitness Function Accuracy score on test set', and Algorithm 1 tracks 'the best solution' by that fitness. The 'best' values in Tables 3–5 are the same statistic, so reporting them as model performance is not an independent evaluation; it is the optimized objective on the same split. This affects all three headline accuracies and the claimed superiority over other optimizers. The runtime-reduction claim (Table 7, ~60% speed-up) is separate and not circular, and the GUI/engineering contribution is independent. Internal inconsistencies (e.g., Parkinson 95.00% vs 0.9487, hidden units 50 vs 100) are additional reliability concerns but are not circularity. Because the primary reported predictions are identical to the fitted objective, the circularity score is 6.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The empirical claims depend on a handful of hand-selected settings and on two questionable assumptions: test-set fitness and thread-based parallelism. No code, repeat runs, or sensitivity tests are provided to check these choices.

free parameters (5)
  • KPCA variance retention threshold = 95%
    The number of principal components is data-dependent and set to retain 95% of variance; exact component counts per dataset are not reported.
  • GA population size = 10
    Hand-selected; no sensitivity analysis is provided.
  • GA number of generations = 10
    Hand-selected; convergence claims are based on this single setting.
  • GA mutation rate = 0.1
    Hand-selected; no sensitivity analysis.
  • MLP max iterations = 500
    Hand-selected training budget; may limit convergence for some configurations.
assumptions (4)
  • ad hoc to paper Using test-set accuracy as the GA fitness function yields unbiased final performance.
    Table 2 lists fitness as accuracy on test set; no separate validation split exists, so final accuracies are the optimization target rather than independent estimates.
  • ad hoc to paper ThreadPoolExecutor parallelizes CPU-bound MLP training, supporting the 60% speedup.
    Section 3.5 claims parallel fitness evaluation via multithreading; under CPython's GIL, threads do not provide full CPU-bound parallelism and no profiling is given.
  • domain assumption RBF kernel PCA retaining 95% variance preserves the discriminative information needed for classification.
    Section 3.3 sets this threshold without ablation against linear PCA or other component counts.
  • domain assumption The public datasets are correctly labeled and preprocessing (KNN imputation, one-hot encoding, random 80/20 split) introduces no leakage.
    Section 3.2 describes preprocessing but no leakage analysis; imputation before splitting can leak information.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Development of a Multiprocessing Interface Genetic Algorithm for Optimising a Multilayer Perceptron for Disease Prediction." pith.science (2026). https://pith.science/paper/ROWMUXDG

@misc{pith2026250615694,
  author       = {Pith},
  title        = {Pith review of: Development of a Multiprocessing Interface Genetic Algorithm for Optimising a Multilayer Perceptron for Disease Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ROWMUXDG}},
  note         = {Machine review of arXiv:2506.15694}
}
read the original abstract

This study introduces a framework that integrates nonlinear feature extraction, classification, and efficient optimization. First, kernel principal component analysis with a radial basis function kernel reduces dimensionality while preserving 95% of the variance. Second, a multilayer perceptron (MLP) learns to predict disease status. Finally, a modified multiprocessing genetic algorithm (MIGA) optimizes MLP hyperparameters in parallel over ten generations. We evaluated this approach on three datasets: the Wisconsin Diagnostic Breast Cancer dataset, the Parkinson's Telemonitoring dataset, and the chronic kidney disease dataset. The MLP tuned by the MIGA achieved the best accuracy of 99.12% for breast cancer, 94.87% for Parkinson's disease, and 100% for chronic kidney disease. These results outperform those of other methods, such as grid search, random search, and Bayesian optimization. Compared with a standard genetic algorithm, kernel PCA revealed nonlinear relationships that improved classification, and the MIGA's parallel fitness evaluations reduced the tuning time by approximately 60%. The genetic algorithm incurs high computational cost from sequential fitness evaluations, but our multiprocessing interface GA (MIGA) parallelizes this step, slashing the tuning time and steering the MLP toward the best accuracy score of 99.12%, 94.87%, and 100% for breast cancer, Parkinson's disease, and CKD, respectively.

Figures

Figures reproduced from arXiv: 2506.15694 by the authors.

Figure 1
Figure 1. Proposed Framework 3.1 Dataset 3.1.1 Chronic kidney disease (CKD) Chronic kidney disease (CKD) is also called chronic renal failure (CRF). CKD is a clinical syndrome characterized by permanent structural and functional damage to the kidneys that lasts for at least three months and reduces their ability to filter metabolic waste, regulate fluid-electrolyte balance, and control blood pressure (Iliyas et al., 2025). It… view at source ↗
Figure 2
Figure 2. The proposed method interface on the CKD dataset [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 4
Figure 4. Screenshot of the proposed method interface for Parkinson’s disease [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Screenshot of the proposed method interface for treating breast cancer 5.0 Discussion The proposed MIGA overcomes the computational cost of the genetic algorithm by parallelizing fitness evaluations across multiple CPU cores, which reduces the tuning time by approximat…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

42 extracted references · 38 canonical work pages

  1. [1]

    classification

    Introduction Disease diagnosis is essential for effective patient management and improved clinical outcomes. Clinicians traditionally rely on manual interpretation of laboratory tests and diagnostic imaging to detect diseases. However, this approach is time-consuming and often incapable of capturing the full complexity of high-dimensional biomedical data ...

    work page 2025

  2. [2]

    Initialize population

    work page

  3. [3]

    Generate an initial population 𝒫0 = {𝜃1, 𝜃2, … , 𝜃𝑃} with random values from𝑆

    work page

  4. [4]

    Parallel Fitness Evaluation: • Compute 𝐹(𝜃𝑖) for each 𝜃𝑖 ∈ 𝒫𝑔 by multithreading b

    For each generation,𝑔 = 1 to 𝐺: a. Parallel Fitness Evaluation: • Compute 𝐹(𝜃𝑖) for each 𝜃𝑖 ∈ 𝒫𝑔 by multithreading b. Selection: Rank population by fitness Select the top 50% as parents𝒫elite ⊂ 𝒫𝑔 c. Crossover: While|𝒫𝑔+1| < 𝑃: i. Randomly select two parents𝜃𝑝 (1), 𝜃𝑝 (2) ∈ 𝒫elite ii. Generate child 𝜃𝑐 by randomly selecting each gene from either parent ii...

    work page

  5. [5]

    End 4.0 Results The study was able to propose a modified GA for optimizing the MLP

    Return Final best hyperparameter configuration 𝜃∗ and its fitness score 𝐹(𝜃∗). End 4.0 Results The study was able to propose a modified GA for optimizing the MLP. The optimized model was used to predict diseases, and the framework was transformed into an app that incorporates the proposed hyperparameter tuning technique to achieve enhanced performan ce. T...

    work page

  6. [6]

    Advanced Agrochem Understanding the potential applications of Arti fi cial Intelligence in Agriculture Sector,

    M. Javaid, A. Haleem, I. Haleem, and R. Suman, “Advanced Agrochem Understanding the potential applications of Arti fi cial Intelligence in Agriculture Sector,” Adv. Agrochem, vol. 2, no. 1, pp. 15–30, 2023, doi: 10.1016/j.aac.2022.10.001. 11

    work page doi:10.1016/j.aac.2022.10.001 2023

  7. [7]

    Genetic algorithm to optimize the SVM and K -means algorithms for mapping of mineral prospectivity,

    R. Ghezelbash, A. Maghsoudi, M. Shamekhi, B. Pradhan, and M. Daviran, “Genetic algorithm to optimize the SVM and K -means algorithms for mapping of mineral prospectivity,” Neural Comput. Appl., vol. 5, 2022, doi: 10.1007/s00521-022-07766-5

    work page doi:10.1007/s00521-022-07766-5 2022

  8. [8]

    Optimizing a Deep Residual Neural Network with Genetic Algorithm for Acute Lymphoblastic Leukemia Classification,

    L. F. Rodrigues, A. R. Backes, B. A. N. Travençolo, and G. M. B. de Oliveira, “Optimizing a Deep Residual Neural Network with Genetic Algorithm for Acute Lymphoblastic Leukemia Classification,” J. Digit. Imaging, pp. 623–637, 2022, doi: 10.1007/s10278-022-00600-3

    doi:10.1007/s10278-022-00600-3 2022

Show all 42 references
  1. [9]

    MLP PSO 92.76% Proposed Method MLP MIGA 100% Parkison [29]) SVM Bayesian Optimization 92.30%

  2. [10]

    Random forest grid fault prediction based on genetic algorithm optimization,

    K. Liu, Y. Gu, L. Tang, Y. Du, C. Zhang, and J. Zhu, “Random forest grid fault prediction based on genetic algorithm optimization,” no. March, pp. 1–10, 2025, doi: 10.3389/fphy.2025.1480749

    2025

  3. [11]

    A hyper-parameter tuning approach for cost-sensitive support vector machine classifiers,

    R. Guido, M. Carmela, and G. Domenico, “A hyper-parameter tuning approach for cost-sensitive support vector machine classifiers,” Soft Comput., vol. 27, no. 18, pp. 12863–12881, 2023, doi: 10.1007/s00500- 022-06768-8

    2023 doi

  4. [12]

    Chronic kidney disease prediction using boosting techniques based on clinical parameters,

    S. M. Ganie, P. K. D. Pramanik, S. Mallik, and Z. Zhao, “Chronic kidney disease prediction using boosting techniques based on clinical parameters,” PLoS ONEs one, pp. 1–21, 2023, doi: 10.1371/journal.pone.0295234

    2023 doi

  5. [13]

    Explainable Machine Learning Model for Chronic Kidney Disease Prediction,

    M. S. Arif, A. U. Rehman, and D. Asif, “Explainable Machine Learning Model for Chronic Kidney Disease Prediction,” pp. 1–17, 2024, [Online]. Available: https://www.proquest.com/openview/089e2d2fb3412464ac18a2157af05a5b/1?cbl=2032439&pq - origsite=gscholar

    2024

  6. [14]

    A Proposed Multilayer Perceptron Model and Kernel Principal Analysis Component for the Prediction of Chronic Kidney Disease,

    I. I. Iliyas, S. Boukari, and A. Ya, “A Proposed Multilayer Perceptron Model and Kernel Principal Analysis Component for the Prediction of Chronic Kidney Disease,” Int. J. Artif. Intell., pp. 99–113, 2024, doi: 10.36079/lamintang.ijai-01102.783

    2024 doi

  7. [15]

    Hyperparameter optimization : Foundations , algorithms , best practices , and open challenges,

    B. Bischl, J. Richter, M. Becker, M. Binder, and T. Pielok, “Hyperparameter optimization : Foundations , algorithms , best practices , and open challenges,” no. August 2022, pp. 1 –43, 2023, doi: 10.1002/widm.1484

    2022 doi

  8. [16]

    Hernández-Morales, I

    A. Hernández-Morales, I. Van Nieuwenhuyse, and R. S. Gonzalez, A Survey On Multi Objective Hyperparameter Optimization Algorithms For Machine Learning, vol. 56, no. 8. Springer Netherlands,

  9. [17]

    MLP Quantum Particle Swarm Optimization (QPSO) 93.00%

  10. [18]

    OPEN SRADHO : statistical reduction approach with deep hyper optimization for disease classification using artificial intelligence,

    G. S. Kumar, E. Suganya, S. Sountharrajan, and B. Balusamy, “OPEN SRADHO : statistical reduction approach with deep hyper optimization for disease classification using artificial intelligence,” scien, pp. 1– 27, 2025, [Online]. Available: https://www.nature.com/articles/s41598...

    2025

  11. [19]

    NN GA 95.00% 8 Proposed Method MLP MIGA 95.00% The study developed a GUI that performs hyperparameter tuning in a unified window: the top panel includes a select file button to load any CSV file; a dropdown to select the target variable; text fields for user hidden -layer size...

    2022

  12. [20]

    Parkinson’s Disease Data Set,

    V. Ukani, “Parkinson’s Disease Data Set,” 2020. [Online]. Available: https://archive.ics.uci.edu/ml/datasets/Chronic_Kidney_Disease

    2020

  13. [21]

    B. P. Kaur, H. S. Id, R. Hans, S. K. Sharma, C. Sharma, and M. H. Id, A Genetic algorithm aided hyper parameter optimization based ensemble model for respiratory disease prediction with Explainable AI . 2024. doi: 10.1371/journal.pone.0308015

    2024 doi

  14. [22]

    CNN PSO 98.23% (DDSM), 97.98% (MIAS)

  15. [23]

    Wisconsin Diagnostic Breast Cancer (WDBC) Data Set,

    W. H. Wolberg and O. L. Mangasarian, “Wisconsin Diagnostic Breast Cancer (WDBC) Data Set,” 1995. [Online]. Available: https://archive.ics.uci.edu/ml/datasets/Breast+Cancer+Wisconsin+(Diagnostic)

    1995

  16. [24]

    A New Optimization Model for MLP Hyperparameter Tuning: Modeling and Resolution by Real-Coded Genetic Algorithm,

    F. Z. El-Hassani, M. Amri, N. E. Joudar, and K. Haddouch, “A New Optimization Model for MLP Hyperparameter Tuning: Modeling and Resolution by Real-Coded Genetic Algorithm,” Neural Process. Lett., vol. 56, no. 2, pp. 1–31, 2024, doi: 10.1007/s11063-024-11578-0

    2024 doi

  17. [25]

    Optimizing Chronic Kidney Disease Diagnosis in Uddanam : A Smart Fusion of GA-MLP Hybrid and PCA Dimensionality Reduction,

    P. Ranga, V. Terlapu, D. Jayaram, S. Rakesh, and M. V. Gopalachari, “Optimizing Chronic Kidney Disease Diagnosis in Uddanam : A Smart Fusion of GA-MLP Hybrid and PCA Dimensionality Reduction,” Procedia Comput. Sci., vol. 230, no. 2023, pp. 522–531, 2023, doi: 10.1016/j.procs.2...

    2023 doi

  18. [26]

    Recent trends in prediction of chronic kidney disease using different learning approaches : a systematic literature review,

    I. I. Iliyas, S. Boukari, and A. Ya, “Recent trends in prediction of chronic kidney disease using different learning approaches : a systematic literature review,” J. Med. Artif. Intell., no. July 2024, pp. 0–2, 2025, doi: 10.21037/jmai-24-256

    2024 doi

  19. [27]

    LightGBM PSO 99.0% Proposed Method MLP MIGA 99.00% CKD [13] MLP GA 98.54%

  20. [28]

    SVM Grid Search 99.33%

  21. [29]

    Enhancing the Early Detection of Chronic Kidney Disease : A Robust Machine Learning Model,

    S. M. Arif, A. Mukheimer, and D. Asif, “Enhancing the Early Detection of Chronic Kidney Disease : A Robust Machine Learning Model,” Big Data Cogn. Comput., vol. 7, 144, 2023, doi: https://doi.org/10.3390/bdcc7030144

    2023 doi

  22. [30]

    Chronic Kidney Disease dataset

    Mansoordaku, “Chronic Kidney Disease dataset.” [Online]. Available: https://archive.ics.uci.edu/dataset/336/chronic+kidney+disease

  23. [31]

    Early and High-Accuracy Diagnosis of Parkinson ’ s Disease : Outcomes of a New Model,

    S. A. Doumari, K. Berahmand, and M. J. Ebadi, “Early and High-Accuracy Diagnosis of Parkinson ’ s Disease : Outcomes of a New Model,” Comput. Math. Methods Med., vol. 2023, 2023, [Online]. Available: https://onlinelibrary.wiley.com/doi/pdf/10.1155/2023/1493676

    2023 doi

  24. [32]

    Automated Detection of Parkinson ’ s Disease Based on Multiple Types of Sustained Phonations Using Linear Discriminant Analysis and Genetically Optimized Neural Network,

    L. Ali, C. E. Zhu, and Z. Zhang, “Automated Detection of Parkinson ’ s Disease Based on Multiple Types of Sustained Phonations Using Linear Discriminant Analysis and Genetically Optimized Neural Network,” IEEE J. Transl. Eng. Heal. Med., vol. 7, no. October, pp. 1–10, 2019, do...

    2019

  25. [33]

    An Optimized Framework for Breast Cancer Classification Using Machine Learning,

    E. Michael, H. Ma, H. Li, and S. Qi, “An Optimized Framework for Breast Cancer Classification Using Machine Learning,” BioMed Res. Int., vol. 2022, 2022, [Online]. Available: https://onlinelibrary.wiley.com/doi/pdf/10.1155/2022/8482022

    2022 doi

  26. [34]

    A CNN Hyperparameters Optimization Based on Particle Swarm Optimization for Mammography Breast Cancer Classification,

    K. Aguerchi, Y. Jabrane, M. Habba, A. Hajjam, and E. Hassani, “A CNN Hyperparameters Optimization Based on Particle Swarm Optimization for Mammography Breast Cancer Classification,” J. imaging, vol. 10, 2024, [Online]. Available: https://www.mdpi.com/2313-433X/10/2/30/pdf

    2024

  27. [36]

    R. J. Woodman and A. A. Mangoni, A comprehensive review of machine learning algorithms and their application in geriatric medicine : present and future, vol. 35, no. 11. Springer International Publishing,

  28. [37]

    doi: 10.1007/s40520-023-02552-2

    doi

  29. [38]

    Genetic -algorithm-optimized neural networks for gravitational wave classification,

    D. S. Deighan, S. E. Field, C. D. Capano, G. Khanna, and S. E. Field, “Genetic -algorithm-optimized neural networks for gravitational wave classification,” Neural Comput. Appl. Manuscr., 2021, [Online]. Available: https://archive.ics.uci.edu/dataset/336/chronic+kidney+disease

    2021

  30. [39]

    Assessing ranking and effectiveness of evolutionary algorithm hyperparameters using global sensitivity analysis methodologies,

    V. Ojha, J. Timmis, and G. Nicosia, “Assessing ranking and effectiveness of evolutionary algorithm hyperparameters using global sensitivity analysis methodologies,” Swarm Evol. Comput., vol. 74, no. October 2021, p. 101130, 2022, doi: 10.1016/j.swevo.2022.101130. 12

    2021

  31. [40]

    An integrated approach of feature selection and machine learning for early detection of breast cancer,

    J. Zhu, Z. Zhao, B. Yin, C. Wu, C. Yin, and R. Chen, “An integrated approach of feature selection and machine learning for early detection of breast cancer,” pp. 1–12, 2025, [Online]. Available: https://www.nature.com/articles/s41598-025-97685-x.pdf

    2025

  32. [41]

    A Robust Chronic Kidney Disease Classifier Using Machine Learning,

    D. Swain et al., “A Robust Chronic Kidney Disease Classifier Using Machine Learning,” Electronics, vol. 12, no. 212, pp. 1–13, 2023, doi: https://doi.org/ 10.3390/electronics12010212

    2023 doi

  33. [42]

    Bayesian Optimization with Support Vector Machine Model for Parkinson Disease Classification,

    A. M. Elshewey, M. Y. Shams, N. El-Rashidy, A. M. Elhady, S. M. Shohieb, and Z. Tarek, “Bayesian Optimization with Support Vector Machine Model for Parkinson Disease Classification,” Sensor, pp. 1–21, 2023, [Online]. Available: https://www.mdpi.com/1424-8220/23/4/2085/pdf

    2023

  34. [2023]

    doi: 10.1007/s10462-022-10359-2

    doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.