Verified Rigor RecordRigor Reviewed

Rigor-Reviewed Manuscript

An effective PO-RSNN and FZCIS based diabetes prediction and stroke analysis in the metaverse environment · 10.1038/s41598-025-96541-2

Name: Verified Rigor Record — An effective PO-RSNN and FZCIS based diabetes prediction and stroke analysis in the metaverse environment · 10.1038/s41598-025-96541-2
Item: An effective PO-RSNN and FZCIS based diabetes prediction and stroke analysis in the metaverse environment · 10.1038/s41598-025-96541-2
Rating: 62
Author: Αlpha¹

Paper source

Read full text

Αlpha¹ Editorial

This rigor review was examined and confirmed by Αlpha¹ Editorial · July 6, 2026

Αlpha¹ Rigor Review

Rigor Reviewed

WARN

Warnings to address · Computational Study

Eight dimensions — click a segment for details

Critical

High-priority

Model-reported confidence: moderateReviewer agreement: lowEngine: 7.6.0Source: full text

Not verified in this report

No reported statistical tests were found to recompute.

This Αlpha¹ Rigor Review uses AI Rigor Reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.

The paper presents a novel computational framework for diabetes and stroke prediction, with a well-articulated scientific premise and detailed mathematical methods. However, the study has several methodological and reporting weaknesses: extremely high accuracy claims that are not tempered, missing inferential statistics, no code sharing, unsupported claims about communication support for disabled persons, and a retracted citation. These issues reduce confidence in the robustness of the findings.

All three reviewers classified the study as computational. The evaluation covers all eight dimensions; ethical approvals and biological variables (in part) were marked not applicable due to the use of public de-identified data. The reviewers diverged on study design, biological variables, key resources, statistical analysis, and reporting transparency, but the synthesized judgments reflect the weighted evidence.

12 major claims checked against the paper's own evidence: 3 not fully backed by the presented evidence (unsupported or overstated).

12 integrity concerns flagged (1 high).

Rigor

Eight Dimensions1 pass · 4 warn · 2 fail

12 major claims3 unsupported7 partial2 supported

The introduction and literature survey cite 15+ prior studies on diabetes/stroke prediction, IoT, and metaverse healthcare (e.g., Hennebelle et al., R. Huang et al., Prabhakar et al.). The premise — that existing models lack support for disabled persons, have security/generalizability/feature-selection issues — is clearly stated and linked to the study's objectives. The paper's contributions explicitly address these limitations (communication support, MJCC security, PO-RSNN, A-PCA, SMOTE). However, the discussion of limitations of prior work is more descriptive than analytical (each cited study's drawback is stated in one sentence without deep critique).

Sub-criteria

prior work citedADEQUATE

premise rationaleADEQUATE

limitations addressedADEQUATE

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

The paper cites prior work, establishes a clear rationale, and addresses limitations of existing models.

Evidence

direct quote[Introduction and Literature survey sections (pages 2-6)]

“Existing models were not regularized well enough to provide communication support for disabled persons in ME-centric CD monitoring. In Ref. , the severity level of diabetes was calculated according to the unstable score values, which affected the model performance.”

direct quote[Introduction, paragraph 5]

“To overcome these limitations, a novel probabilistic-centric optimized recurrent sechelliott neural network (PO-RSNN)-based diabetes prediction (DP) and Fuzzy Z-log-clipping inference system (FZCIS)-based severity level estimation in ME is carried out.”

paraphrase[Introduction, 'The novelty of the proposed system' paragraph]

“The paper claims its novelties (PO-RSNN, FZCIS, MJCC, A-PCA, SMOTE) overcome the specific issues of vanishing gradient, security, dimensionality, and data imbalance mentioned in prior works.”

PASSReviewer 2· 85% conf

The paper cites prior work, establishes a clear rationale for the study, and addresses limitations of existing models.

Evidence

direct quote[Introduction and Literature Survey sections]

“With the evolution of ME-based healthcare, the existing works have utilized wearable IoT sensors and AI-based models for continuous monitoring.”

direct quote[Introduction, paragraph 6]

“Some limitations of the traditional models are given below: Existing models were not regularized well enough to provide communication support for disabled persons in ME-centric CD monitoring.”

direct quote[Introduction, paragraph 7]

“The proposed work is proficiently designed to provide communication support for disabled persons in the ME.”

PASSReviewer 3· 85% conf

The paper adequately establishes its scientific premise by citing relevant prior work and providing a clear rationale linking limitations of existing models to the proposed approach.

Evidence

paraphrase[Introduction, paragraphs 1-7; Literature survey section; Table 1]

“The authors review 15 prior studies in the literature survey, each with stated drawbacks (e.g., 'had security and privacy issues', 'increased misclassification rate', 'data was not balanced').”

direct quote[Introduction, paragraph 4]

“Some limitations of the traditional models are given below: ... Existing models were not regularized well enough to provide communication support for disabled persons ... Most of the prevailing works obtained limited outcomes and high complexity.”

direct quote[Introduction, paragraph 5]

“The proposed work is proficiently designed to provide communication support for disabled persons in the ME. An effective FZCIS is introduced to estimate the severity level of diabetes according to the risk score.”

The study is computational, using two public datasets (DPD and SPD). Randomization, blinding, power analysis, and independent replication are typically not applicable for a pure machine-learning paper, so those are marked n/a. Inclusion/exclusion is not explicitly stated but the datasets are described. Outlier handling is not reported. Controls are not applicable (no wet-lab). The paper describes a train/test split (80/20) and tenfold cross-validation, which is adequate. However, the paper does not mention any pre-registration or protocol for the analysis plan.

Sub-criteria

controlsN/A

power analysisN/A

blinding levelsN/A

outlier handlingNOT REPORTED

randomization unitN/A

inclusion exclusionNOT REPORTED

randomization methodN/A

replicate distinctionN/A

independent replicationN/A

Per reviewer (3) — the votes behind the verdict

WARNReviewer 1· 75% conf

Several design aspects are reported, but key transparency elements (blinding, replication, pre-registration of analysis) are absent or not applicable for this computational study.

Evidence

absence[Results and Discussion, Dataset description]

No inclusion/exclusion criteria are stated for the datasets.

absence[Methods, Pre-processing section]

Outlier handling is not described.

paraphrase[Results and Discussion, Performance analysis]

“The paper states an 80/20 train/test split and uses tenfold cross-validation.”

FAILReviewer 2· 70% conf

The study design lacks key methodological details for a computational study, such as pre-specified inclusion/exclusion criteria, outlier handling, and power analysis for dataset selection.

Evidence

direct quote[Results and Discussion, Comparative analysis]

“The performance of the proposed PO-RSNN is analogized with prevailing approaches, such as RNN, Deep Belief Network (DBN), Restricted Boltzmann Machine (RBM), DNN, ANN, and SVM.”

absence[Methods]

No power analysis or sample-size justification is provided.

absence[Dataset description section]

No explicit inclusion/exclusion criteria for the datasets are stated.

WARNReviewer 3· 70% conf

The study is a computational evaluation using public datasets and does not report several design elements common in experimental or clinical work; applicable sub-criteria show gaps in dataset preprocessing and validation.

Evidence

absence[Proposed methodology, Diabetes dataset and Stroke dataset sections]

No explicit inclusion/exclusion criteria for dataset subjects are reported.

absence[Pre-processing sections (3.5.2 and 4.2)]

Outlier handling is not described.

direct quote[Generalizability of the model paragraph and Fig. 8]

“the tenfold cross-validation strategy is applied... The model exhibited high stability across different folds.”

The paper describes the features in the datasets (gender, age, BMI, etc.) in Table 2, but does not report the actual distribution (e.g., percentage of females, mean age) for the sample used in the study. Sex is listed as a feature but the proportion is not given. Age, weight/health metrics are only mentioned as features, not reported with summary statistics. Demographics (multiple variables) are not summarized. For a computational study using human subject data, this is insufficient.

Sub-criteria

demographicsINADEQUATE

sex reportedINADEQUATE

sex justifiedN/A

age weight healthINADEQUATE

housing conditionsN/A

species strain sourceN/A

Per reviewer (3) — the votes behind the verdict

N/AReviewer 1· 95% conf

The study uses two public, anonymized datasets (DPD and SPD) with demographic and clinical features; no new biological subjects are involved, and the variables are treated as input features.

Evidence

direct quote[Methods, Dataset description]

“The datasets are collected from the publically available resources that are mentioned under the reference section.”

FAILReviewer 2· 70% conf

The paper lists features (sex, age, BMI, etc.) but does not provide summary statistics or distributions for the study sample, leaving biological variable reporting inadequate.

Evidence

direct quote[Table 2, Features description]

“Gender Male/female Used for gender-based risk differences in diabetes prediction”

direct quote[Table 2]

“Age Patient's age in years Older individuals have a higher risk of diabetes, so it is considered for the analysis”

absence[Dataset description section]

No summary statistics for BMI, blood glucose, or other health metrics are provided for the study sample.

PASSReviewer 3· 85% conf

The paper reports biological variables relevant to the computational study through the feature sets of the two public datasets, which include age, sex, BMI, hypertension, heart disease, and other clinical variables.

Evidence

direct quote[Proposed methodology, Table 2 (Feature description)]

“DPD: Gender Male/female ... Age Patient's age in years ... SPD: Gender Male/female ... Age Patient's age in years ... BMI Body mass index (BMI) (kg/m^2)”

paraphrase[Table 2 and sample sizes in Table 3/4]

“The DPD and SPD feature descriptions include 'Gender' as a feature with values Male/female, indicating both sexes are present.”

The paper explicitly states that the data are from public Kaggle datasets. No new human or animal data were collected. Therefore, ethical approvals, consent, and regulatory compliance are not applicable. The paper does not include any ethics statement, which is appropriate for this type of secondary data analysis.

Sub-criteria

iacuc statementN/A

informed consentN/A

irb ethics statementN/A

regulatory complianceN/A

Per reviewer (3) — the votes behind the verdict

N/AReviewer 1· 90% conf

The study uses only publicly available, fully de-identified datasets from Kaggle, so no ethics approval or consent is required.

Evidence

direct quote[Results and Discussion, Dataset description]

“https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset , https://www.kaggle.com/datasets/iammustafatz/diabetes-prediction-dataset”

N/AReviewer 2· 90% conf

The study uses only public, fully de-identified datasets from Kaggle, making ethics approval not applicable per the guidelines.

Evidence

direct quote[Data availability and Dataset description]

“The data used to support the findings of this study are available from the corresponding author upon request. https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset”

N/AReviewer 3· 90% conf

The study uses only public, fully de-identified datasets from Kaggle and does not collect new human data, conduct animal research, or use controlled-access subject-level data, so ethics approval is not applicable.

Evidence

direct quote[Data availability section and Dataset description paragraph]

“The datasets are collected from the publically available resources that are mentioned under the reference section.”

direct quote[Data availability section]

“https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset https://www.kaggle.com/datasets/iammustafatz/diabetes-prediction-dataset”

No antibodies, cell lines, organisms, or traditional reagents are used; these are not applicable. The paper identifies Python 3.10, PyCharm, and several libraries (Tkinter, Matplotlib, Keras, TensorFlow, etc.) but only gives version numbers for Python and PyCharm. Custom code is not shared in a public repository. Datasets are identified with URLs but are not covered under the checklist items. Therefore, the sole applicable sub-criterion (software_tools_identified) is inadequate.

Sub-criteria

mycoplasma testingN/A

reagents identifiedN/A

organisms identifiedN/A

antibodies identifiedN/A

cell line authenticationN/A

software tools identifiedINADEQUATE

Per reviewer (3) — the votes behind the verdict

WARNReviewer 1· 80% conf

Software and datasets are identified with versions and sources, but the Python libraries are listed without version numbers and custom code is not shared in a repository.

Evidence

direct quote[Results and Discussion, first paragraph]

“The proposed system is developed using Python 3.10... PyCharm 2022.3.3 is utilized... libraries: Tkinter, Matplotlib, Keras and TensorFlow, NumPy, CSV, imblearn, Scikit-learn (sklearn), cryptography.”

absence[Methods and Data Availability sections]

No public code repository is provided for custom code.

FAILReviewer 2· 80% conf

Only software tools are applicable, and their identification is incomplete (library versions not fully stated, no code repository).

Evidence

direct quote[Results and Discussion, first paragraph]

“The proposed system is developed using Python 3.10, which is a widely used programming language for DL applications. The PyCharm 2022.3.3 is utilized for coding, debugging, and executing the model efficiently on a Windows 10 (64-bit) platform.”

absence[Data availability section]

No code repository is provided.

PASSReviewer 3· 75% conf

The paper adequately identifies the software tools and datasets used. The study does not use antibodies, cell lines, or other wet-lab reagents, and the primary resources are software packages and public datasets.

Evidence

direct quote[Results and discussion, first paragraph]

“The proposed system is developed using Python 3.10... PyCharm 2022.3.3 is utilized for coding, debugging ... Keras and TensorFlow serve as the primary frameworks”

direct quote[Data availability section]

“https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset https://www.kaggle.com/datasets/iammustafatz/diabetes-prediction-dataset”

The paper names the metrics used (accuracy, precision, recall, F-measure, FPR, FNR, training time) and compares against baselines, but does not report any p-values, confidence intervals, or effect sizes. The statistical tests for model comparison are not named (e.g., no McNemar's test or paired t-test). Assumptions (normality, variance) are not verified. Data presentation shows bar charts and tables with means, but individual data points are not shown. Error bars are not defined. The training times are reported in milliseconds, which is reasonable for this type of work. No internal consistency issues were found with the reported metrics as they are all percentages that could plausibly sum correctly.

Sub-criteria

tests namedNOT REPORTED

exact p valuesNOT REPORTED

effect sizes ciNOT REPORTED

data presentationINADEQUATE

software identifiedADEQUATE

assumptions verifiedNOT REPORTED

mathematical plausibilityN/A

Per reviewer (3) — the votes behind the verdict

WARNReviewer 1· 70% conf

Statistical tests are named and software is identified, but exact p-values, effect sizes with CIs, and assumption verification are not reported. Data presentation uses standard metrics but lacks individual data points or explicit error bar definitions.

Evidence

absence[Results and Discussion, Performance analysis]

No statistical tests (e.g., for comparing models) are named.

absence[Results and Discussion]

No exact p-values or confidence intervals are reported.

direct quote[Results and Discussion, first paragraph]

“The proposed system is developed using Python 3.10... PyCharm 2022.3.3... Scikit-learn (sklearn) is utilized for model selection and training.”

FAILReviewer 2· 80% conf

The paper reports performance metrics without any inferential statistics, hypothesis tests, p-values, or effect sizes; software identification and data presentation are inadequate.

Evidence

paraphrase[Results and Discussion]

“The paper reports accuracy, precision, recall, FPR, FNR, etc., without any hypothesis tests or p-values.”

absence[Figures 3-7]

Data presentation lacks error bars on performance metrics.

paraphrase[Results, first paragraph]

“Scikit-learn (sklearn) is mentioned but no version is given.”

WARNReviewer 3· 65% conf

The paper reports standard classification metrics and compares models, but lacks exact p-values, confidence intervals, effect sizes, and verification of statistical assumptions; mathematical plausibility checks are not feasible on the reported aggregate data.

Evidence

absence[Performance analysis sections (5.4-5.7)]

No p-values or inferential statistics are reported for model comparisons.

direct quote[Results and discussion, first paragraph]

“The proposed system is developed using Python 3.10... Scikit-learn (sklearn) is utilized for model selection and training.”

paraphrase[Tables 3 and 4 report sample sizes; Figures 3-11 show bar charts]

“Tables 3 and 4 provide training/testing splits with counts (e.g. DPD training 7776, testing 1946). Performance metrics are presented as bar charts without individual data points or error bars.”

The paper provides a data availability statement: 'The data used to support the findings of this study are available from the corresponding author upon request.' It then lists two Kaggle URLs for the datasets. The statement is vague because 'upon request' is used alongside the URLs; the URLs themselves are concrete but the paper does not state that the data is deposited in a persistent repository with a DOI or accession. Custom code (the proposed models) is not shared. For a computational study, sharing code is a key reproducibility criterion.

Sub-criteria

code sharingNOT REPORTED

accession numbersNOT REPORTED

repository depositINADEQUATE

data availability statementINADEQUATE

Per reviewer (3) — the votes behind the verdict

WARNReviewer 1· 85% conf

A data availability statement is present that names public repositories, but it is vague (no DOI/accession numbers) and code is not shared.

Evidence

direct quote[Data Availability section, end of paper]

“The data used to support the findings of this study are available from the corresponding author upon request.”

direct quote[Data Availability section]

“https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset , https://www.kaggle.com/datasets/iammustafatz/diabetes-prediction-dataset”

absence[Full paper text]

No custom code repository is provided.

WARNReviewer 2· 85% conf

A data availability statement is present with direct Kaggle URLs, but no code repository is provided and deposit in a formal repository with persistent identifier is missing.

Evidence

direct quote[Data availability section]

“The data used to support the findings of this study are available from the corresponding author upon request. https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset”

absence[Data availability section]

No custom code repository is provided.

WARNReviewer 3· 80% conf

The paper provides URLs to the public datasets and a vague data availability statement, but does not share custom code or provide persistent identifiers for the deposited data.

Evidence

direct quote[Data availability section]

“The data used to support the findings of this study are available from the corresponding author upon request.”

direct quote[Data availability section]

“https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset, https://www.kaggle.com/datasets/iammustafatz/diabetes-prediction-dataset”

absence[Entire manuscript]

Custom code is not shared in any public repository.

The methods section is extensive, providing mathematical formulations for all proposed components. The limitations section acknowledges that other chronic diseases (hypertension, cancer, heart disease) were not considered. Funding and conflicts of interest are not explicitly stated in the main text, but the author affiliations and journal information suggest institutional support, and the paper was published in Scientific Reports, which requires a COI statement. The paper does not reference any reporting checklist (e.g., STROBE, TRIPOD). All outcomes appear to be reported (accuracy, precision, etc.) but there is no explicit statement about pre-specification. Conclusions generally match the results (98.97% accuracy claims), though such extremely high accuracy on a medical prediction task could be considered overstated.

Sub-criteria

funding coiADEQUATE

trial registrationN/A

reporting guidelineNOT REPORTED

methods completenessREPORTED_ADEQUATE

all outcomes reportedADEQUATE

limitations discussedADEQUATE

conclusions proportionalINADEQUATE

Per reviewer (3) — the votes behind the verdict

WARNReviewer 1· 80% conf

Methods are detailed, limitations are discussed, and funding/COI are present, but the paper lacks a reporting guideline checklist, trial registration (not applicable), and it is not clear if all outcomes are pre-specified.

Evidence

direct quote[Limitations and future scope]

“Although the proposed work effectively predicted diabetes and analyzed stroke in a patient, other CDs, such as hypertension, cancer, and heart disease, were not considered.”

absence[Methods]

No reporting guideline checklist is mentioned.

direct quote[Abstract]

“The experimental validation demonstrated the superior performance of the proposed system by achieving 98.97% accuracy in DP and 98.89% accuracy in stroke analysis, outperforming existing classifiers.”

FAILReviewer 2· 80% conf

The paper lacks several key transparency elements: no funding/COI statement, no reporting guideline adherence, incomplete methods, and no evidence that all outcomes were reported.

Evidence

absence[Proposed methodology sections]

Methods lack specific training hyperparameters.

absence[End of paper (Declarations section)]

No funding or conflict of interest statement is provided.

direct quote[Conclusion, Limitations and future scope]

“Although the proposed work effectively predicted diabetes and analyzed stroke in a patient, other CDs, such as hypertension, cancer, and heart disease, were not considered.”

WARNReviewer 3· 75% conf

The paper describes methods adequately for a computational study and discusses some limitations, but lacks a reporting guideline reference, trial registration (not applicable), and does not report all pre-specified outcomes with transparent handling of null results; however, conclusions are proportional.

Evidence

paraphrase[Proposed methodology sections (3.1-3.10)]

“The paper provides equations for key steps: MJCC encryption (Eq. 2-7), A-PCA (Eq. 14-18), PO-RSNN (Eq. 19-27), FZCIS (Eq. 28-33), plus pseudocode algorithms.”

direct quote[Limitations and future scope]

“the proposed system only focused on communication support for disabled persons in ME and limited CD prediction.”

absence[Declarations section]

No funding or conflict-of-interest statement is provided.

30 copyedit issues flagged (19 major): mostly typo, other, grammar.

Checked 41 references: 40 verified — 1 retracted.

Reproducibility

2 data/code links checked; 0 live, 2 dead.

Registration stated in text, but no registry ID was detected. No reporting guideline cited.

Αlpha¹ Rigor Action ItemsFinal pass before journal submission

This paper is already published, but it would benefit from a correction or erratum to address the retracted citation, the unsupported claims, the internal contradictions in table labels, and the missing code and funding statements. An informed reader should weigh the extremely high accuracy claims cautiously, given the lack of inferential statistics and the potential for overfitting or data leakage.

1.
HIGHother
Remove or replace the retracted reference: 'A novel diabetes healthcare disease prediction framework using machine learning techniques' (DOI 10.1155/2023/9872970) with a non-retracted source.
Citing a retracted paper undermines the study's credibility and could mislead readers.
2.
HIGHrigor
Provide evidence for the claim that the system provides communication support for disabled persons (e.g., user study, implementation details, evaluation metrics) or remove the claim from the abstract and conclusions.
The claim is unsupported by any experimental validation, which is a serious integrity issue.
3.
HIGHrigor
Correct the O(1) training time complexity claim for the proposed PO-RSNN; a neural network cannot have O(1) complexity – provide a correct complexity analysis or remove the claim.
The claim is mathematically implausible and indicates a misunderstanding of computational complexity.
4.
HIGHreporting
Correct the swapped class labels in Tables 3 and 4: Table 3 (DPD) should show 'Diabetes' and 'Non-diabetes', not 'Stroke' and 'Non-Stroke'; Table 4 (SPD) should show 'Stroke' and 'Non-Stroke', not 'Diabetes' and 'Non-diabetes'.
This internal contradiction undermines the credibility of the data description and suggests a copy-paste error.
5.
HIGHdata code
Upload the custom code (PO-RSNN, FZCIS, MJCC, A-PCA, P-DBO) to a public repository (e.g., GitHub, Zenodo) with a persistent DOI and add a Code Availability statement to the paper.
Without code, the study is not reproducible, which is a major limitation for a computational paper.
6.
HIGHreporting
Add a funding statement (including grant numbers if applicable) and a conflict-of-interest disclosure in the Declarations section.
Missing this information is a transparency gap that journals typically require.
7.
HIGHstatistics
Report exact p-values or confidence intervals for the key performance comparisons between the proposed model and baselines, or use bootstrapping to generate confidence intervals for accuracy.
Without inferential statistics, readers cannot assess whether the reported improvements are statistically significant.
8.
HIGHreporting
Add a limitations section that discusses potential overfitting, the generalizability of the extremely high accuracy claims, and the lack of external validation.
The current limitations paragraph is too brief and does not address the most critical threats to validity.
9.
HIGHreporting
Provide summary statistics (mean, SD, range) for the biological variables (age, BMI, blood glucose, etc.) in the sample used, either in a table or in the text.
This allows readers to assess the representativeness of the study sample and potential biases.
10.
HIGHreporting
Reference a relevant reporting guideline (e.g., TRIPOD for prediction model studies) and note adherence in the Methods section.
Following reporting guidelines improves transparency and reproducibility.
11.
HIGHrigor
Specify the inclusion/exclusion criteria for the datasets (e.g., any filtering of subjects based on missing data, extreme values) and describe how outliers were handled during preprocessing.
These details are essential for understanding the data preparation and potential biases.
12.
MEDIUMdata code
Add version numbers for all Python libraries (e.g., TensorFlow, scikit-learn, Keras) to the software description in the Results section.
Version numbers are critical for reproducibility of the computational environment.
13.
MEDIUMdata code
Replace the vague 'available upon request' data availability statement with a clear statement that the datasets are publicly available at the provided Kaggle URLs, and note that code is available upon request or has been shared.
The current statement is contradictory and may confuse readers about data access.
14.
MEDIUMcopyedit
Replace the term 'deaf and dumb' with 'deaf or hard of hearing and speech-impaired' or another respectful term throughout the paper.
The term 'dumb' is outdated and considered offensive.
15.
MEDIUMcopyedit
Fix subject-verb agreement errors, e.g., change 'Chronic disease... impacts' to 'Chronic diseases... impact' and 'CD... is' to 'CDs... are'.
Multiple grammatical errors reduce readability and professionalism.
16.
MEDIUMcopyedit
Consistently hyphenate compound modifiers: e.g., 'PO-RSNN- and FZCIS-based' instead of 'PO-RSNN and FZCIS based'.
Missing hyphens cause ambiguity in compound adjectives.
17.
LOWcopyedit
Capitalize 'Sechelliott' consistently as a proper name (e.g., 'Sechelliott neural network') throughout the paper.
Inconsistent capitalization of a proposed term may confuse readers.
18.
LOWcopyedit
Add periods after author initials in the Author Contributions section (e.g., 'S.S., K.M., A.M., and K.V.').
Missing periods after initials is a formatting inconsistency.
19.
LOWcopyedit
Remove extra spaces before periods throughout the manuscript.
This pattern occurs in multiple locations and should be corrected for professional formatting.

Full text (65,090 chars)Analyzed: Jul 6, 2026

Αlpha¹ assesses methodological rigor, not the importance of the findings. See how we calibrate →