Predictive Modeling For Real-Time Personalized Health Monitoring in Muscular Dystrophy Management

Mohammed Akkaoui

REVIEW 4 major objections 6 minor 26 references

Predictive Modeling For Real-Time Personalized Health Monitoring in Muscular Dystrophy Management

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper proposes an IoT-based system that streams muscular dystrophy biomarkers into personalized real-time risk alerts.

desk verdict A clear conceptual outline of an IoT-plus-ML monitoring system for muscular dystrophy, but the one quantitative piece, the RiskScore, is internally inconsistent and the paper has no data or implementation to back its claims. read the letter →

arxiv 2411.14923 v1 pith:OINYTB6A submitted 2024-11-22 cs.LG

Mohammed Akkaoui This is my paper · ORCID

classification cs.LG

keywords musculardystrophyIoThealthmonitoringwearablesensorspredictivemodelingRiskScoreremotepatientpoint-of-caretestingmachinelearninghealthcare

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 conceptual paper proposes an Internet-of-Things care-management system for muscular dystrophy (MD), a group of progressive muscle disorders. It argues that wearable sensors for muscle, cardiac, respiratory, and environmental data, combined with point-of-care blood tests for CPK and ALT/AST, can feed cloud machine-learning models that issue real-time, personalized alerts before health crises develop. The intended payoff is a shift from periodic clinic visits to continuous home monitoring, giving clinicians earlier warning of muscle damage and letting patients adjust activity, hydration, and medication day by day. The paper presents the architecture, data flows, and a composite RiskScore formula, but does not report a working implementation or clinical validation.

What carries the argument

The load-bearing object is the RiskScore equation, a weighted composite of four inputs CPK, ALT, AST, and EMG amplitude designed so that CPK contributes half the weight. The system pairs this score with a set of fixed clinical thresholds (CPK above 1000 U/L, ALT above 140 U/L, AST above 100 U/L, SpO2 below 90%, heart rate above 120 bpm, HRV below 20 ms, temperature above 30°C, humidity above 70%) and with machine-learning models assigned by data type: ARIMA and LSTM for CPK and other time-series trends, Random Forest and SVM for risk classification. A gateway performs local preprocessing, including a Fast Fourier Transform on EMG, before encrypted upload to the cloud, and that local filtering is what makes the real-time alert loop plausible. The RiskScore is the single number that decides whether caregiver and provider are notified at moderate or high risk.

What would settle it

Calculate the RiskScore for a patient with CPK 1200 U/L, ALT 120 U/L, AST 120 U/L, and EMG amplitude 0.3 mV; the formula gives 1.23, while the text says scores above 6/10 are high risk, so the comparison only makes sense if the thresholds are read as 0.6 on a unit scale—which the paper never states. A clinician testing the system would need a rule for converting the raw weighted sum to the 0-to-10 scale before the moderate and high alerts can fire.

Watch

Extended reading notes

Core claim

The paper's central claim is that a complete real-time picture of an MD patient's health can be assembled from a body-worn sensor network plus frequent finger-prick biomarker measurements, and that machine-learning models (ARIMA, LSTM, Random Forest, SVM) can turn this stream into predictive warnings. The proposed RiskScore is the mechanism that condenses the stream into an actionable number: $RiskScore = 0.5 \times \frac{CPK}{1000} + 0.25 \times \frac{ALT}{100} + 0.25 \times \frac{AST}{100} + 0.1 \times EMG\ amplitude\ (mV)$, with moderate-risk alerts above 3/10 and high-risk alerts above 6/10. The paper argues that CPK, as the most sensitive muscle-damage marker, should dominate the score, and that thresholds such as CPK above 1000 U/L, SpO2 below 90%, and EMG amplitude below 0.5 mV should trigger provider notifications. If the design works as described, clinicians would see deteriorating trends such as a multi-day CPK climb before an acute episode, enabling earlier changes to medication or physiotherapy.

Load-bearing premise

The entire alert system rests on hand-picked RiskScore weights and clinical thresholds that the paper states but never derives from patient data or published calibration; if those numbers misrepresent real muscular dystrophy physiology, the system will either cry wolf or miss the deterioration it promises to catch.

Editorial extensions

If this is right

Care could move from snapshot clinic visits to continuous home surveillance, with alerts triggered by trends rather than by scheduled tests.
A clinician could be warned when predicted CPK is about to cross 200 U/L or 1000 U/L, creating a window for medication or therapy changes before muscle damage accumulates.
Patients would receive day-specific activity, hydration, and exercise guidance based on the current RiskScore and environmental conditions.
The same architecture of wearables, cloud analytics, risk score, and alert thresholds could be reused for other progressive chronic diseases.

Reading between the lines

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

Before deployment, the RiskScore's 0-to-10 alert thresholds would need to be reconciled with a formula whose weights sum to 1.1, since as written the numeric scales do not match.
Because CPK rises after any muscle injury or exertion, the system would need a personal baseline and activity context; otherwise a benign post-exercise spike could be scored as moderate risk.
A testable extension would be to run the RiskScore on retrospective MD patient data with known adverse events and measure whether alert timing precedes hospitalizations.

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.

Desk Editor's Note

read the letter

Colleague —

This is a conceptual paper, and it should be judged as one. The authors say so up front, and they do a reasonable job of assembling the standard IoT-plus-wearables-plus-ML architecture for remote monitoring and applying it to MD. To their credit, they recognize that wearables cannot measure CPK/ALT/AST and build in point-of-care testing for those biomarkers. The device/transmission table is sensible, and they cite relevant clinical guidelines for thresholds like SpO2 below 90% and heart rate above 120 bpm. As a broad system sketch, it is readable and organized.

But the actual contribution is thin. There is no data, no implementation, no validation, and no new algorithm. The only novel-looking item is the RiskScore in Eq. (1), and it does not work as written. The weights sum to 1.1 rather than 1.0; the stated thresholds (3/10 and 6/10) do not correspond to the formula's output range; and plugging in the paper's own normal values (CPK=200, ALT/AST=40, EMG=0.5 mV) gives a score above 0.3, so the moderate-risk alert would fire on normal data. The stress-test note is right: this is load-bearing, because the entire early-warning claim depends on these thresholds being clinically correct, and no calibration data or derivation is provided. Saying the weights are "calibrated" and "evidence-based" without showing a calibration procedure is not enough.

The conclusion overstates what has been shown. The system "will" do many things, but nothing has been built or tested. The reference list has duplicate entries and some weak items, though that is secondary to the missing evidence.

I would not send this to a full peer review as is. A desk reject with detailed feedback is appropriate: the authors would need at least a fixed and justified risk score, a pilot dataset or simulation, and a validation plan before this becomes refereeable. The current audience is someone looking for a broad architecture overview, not a clinician or researcher needing evidence.

Referee Report

4 major / 6 minor

Summary. This conceptual paper proposes an IoT-based remote monitoring system for managing Muscular Dystrophy (MD). The envisioned system combines wearable sensors (EMG, ECG/PPG, accelerometers), point-of-care testing for CPK/ALT/AST, environmental sensors, and cloud-based analytics using ARIMA, LSTM, Random Forest, and SVM models. The only concrete quantitative mechanism is a RiskScore formula in Section III-D, intended to trigger moderate-risk and high-risk alerts. The paper claims this will enable real-time, personalized interventions and improve outcomes, but it presents no implementation, no data, no experimental validation, and no clinical evidence for the risk thresholds.

Significance. If the system were realized and validated, the multi-modal monitoring concept could be clinically useful for a chronic, progressive condition like MD. The paper correctly identifies gaps in current remote monitoring and names relevant technologies. However, the paper's sole quantitative contribution, the RiskScore, is internally inconsistent, and the central claim that the system improves outcomes is entirely unsupported by data. The manuscript does not provide reproducible code, a prototype, a dataset, or falsifiable predictions beyond a hand-specified formula. As it stands, the paper is a system-design sketch rather than a validated predictive modeling study.

major comments (4)

[Section III-D, Eq. (1)] The RiskScore formula has a basic internal inconsistency: the weights 0.5, 0.25, 0.25, and 0.1 sum to 1.1, so the score is not a normalized 0-1 risk measure. The thresholds are written as 'RiskScore > 3/10' and 'RiskScore > 6/10', which would be 0.3 and 0.6 on a normalized scale, but the formula has no such range. Using the paper's own normal upper values (CPK = 200 U/L, ALT = 40 U/L, AST = 40 U/L, EMG = 0.5 mV), the score equals 0.5*(0.2) + 0.25*(0.4) + 0.25*(0.4) + 0.1*(0.5) = 0.35, which already exceeds the stated moderate-risk threshold of 0.3. The alert logic would therefore fire on entirely normal readings, undermining the system's primary early-warning function.
[Section III-D] The paper states that 'These weights and risk thresholds are calibrated in line with a clinically aligned approach', but no calibration procedure, calibration dataset, or external source is provided. The weights and thresholds are free parameters chosen by the authors, and the same RiskScore is used both to define patient risk and to justify alert thresholds, creating a circular justification. No evidence is given that these particular weights or thresholds have any predictive or clinical validity, so the claim of 'calibration' is unsupported.
[Sections IV and V] The central claim that the proposed system 'would significantly improve clinical outcomes and the quality of life for the patients' (Section V) has no supporting evidence. Section IV describes model training and validation in general terms (e.g., cross-validation, accuracy, MAE/RMSE), but the manuscript reports no dataset, no experiments, no performance metrics, and no comparison against baseline monitoring. The abstract itself labels the paper as 'conceptual', yet the conclusion makes empirical causal claims that cannot be supported without implementation and evaluation.
[Section III-D, 'Biochemical Markers' and threshold lists] The specific alert thresholds (e.g., CPK rise > 500 U/L in 24 hours, ALT > 140 U/L, AST > 100 U/L, heart rate variability < 20 ms, SpO2 < 90%) are presented as derived from 'evidence-based guidelines from AANEM, ACR, ESC, ATS, and CDC', but no specific guideline or reference is cited for any of these cutoffs. References [7]-[11] are general society guidelines and do not contain these particular numerical thresholds. Since the system's entire value proposition is timely, reliable alerts, the provenance of these thresholds is load-bearing and currently unsubstantiated.

minor comments (6)

[Section IV, 'Model Selection'] There is a typo in the paragraph on interpretability: 'which biomarkers—Thai is, CPK levels' should read 'that is, CPK levels'.
[References] References [17] and [18] are identical duplicates (Brown, T., & Green, M., 2019).
[Figure 3] The caption reads 'Fig 3: This table summarizes...' but the object is a table, not a figure; the caption should be adjusted for consistency with IEEE style.
[Section I] The paper uses both 'Creatine Kinase (CK)' and 'Creatine Phosphokinase (CPK)' interchangeably; the terminology should be unified throughout.
[Section III-D] The sentence 'The higher its value is, the more severe the injury to muscles, all the way to 1,000 U/L and above' is ambiguous: it is unclear whether values above 1,000 U/L are considered a separate alert threshold or an extension of the normal-to-damage continuum.
[References] Some references are incomplete or provisional, such as [4] 'unpublished', [5] 'in press', and [12] lacking a year and venue; these should be completed before submission.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is a conceptual system design with no fitted predictions or self-cited load-bearing results.

full rationale

The paper does not claim to derive empirical predictions from data; it proposes an IoT-based monitoring architecture. The RiskScore formula in Section III-D is introduced explicitly as a proposed scoring scheme ('The RiskScore employs a new formula...'), with hand-selected weights emphasizing CPK. The thresholds (RiskScore > 3/10 and > 6/10) are presented as design settings, not as fitted or validated cutoffs. Because the paper never claims that these weights were estimated from clinical data or that the score was validated against an independent outcome, the 'risk' is the score by definition and this is not disguised as a derived result. There is no self-citation chain, no imported uniqueness theorem, and no known result renamed as new. The internal inconsistencies in thresholds (weights summing to 1.1; the stated 3/10 and 6/10 thresholds not matching the formula's natural range) are substantive correctness concerns, but they are not circularity: they do not involve a prediction reducing to its own input by construction. The cited guidelines (AANEM, ACR, ESC, ATS, CDC) are external clinical standards, not prior work by these authors. Accordingly, no circular step meets the standard of Eq. X = Eq. Y by construction or fitted parameter renamed as prediction, and the appropriate finding is no significant circularity.

Assumptions & free parameters 16 free parameters · 5 assumptions · 1 invented entities

The central claim rests on hand-picked RiskScore weights and thresholds, unvalidated assumptions about sensor and POCT accuracy, and the availability of training data. The iCare-1300 device is named without evidence.

free parameters (16)

RiskScore weight for CPK (w_CPK) = 0.5
Hand-chosen because the paper considers CPK the most sensitive muscle-damage marker; no calibration data shown.
RiskScore weight for ALT (w_ALT) = 0.25
Hand-chosen; the paper states ALT should not overshadow CPK.
RiskScore weight for AST (w_AST) = 0.25
Hand-chosen; same rationale as ALT.
RiskScore weight for EMG amplitude (w_EMG) = 0.1
Hand-chosen; EMG amplitude is in mV and the weight is small, with no derivation.
Moderate risk threshold = 3/10
Asserted without derivation; the paper references 'the clinical guideline cited' but gives no specific source.
High risk threshold = 6/10
Asserted; the paper says 'according to research on MD' without specifying the research or data.
CPK high alert threshold = 1000 U/L
Presented as a significant damage threshold with no derivation or citation.
CPK 24-hour rise threshold = 500 U/L
Hand-set emergency alert value, no source given.
CPK daily rise threshold = 200 U/L
Hand-set alert value, no source given.
ALT alert threshold = 140 U/L
Hand-set value, no source given.
AST alert threshold = 100 U/L
Hand-set value, no source given.
SpO2 critical threshold = 90%
Common clinical threshold but not cited in the paper.
Heart rate critical threshold = 120 bpm
Hand-set value, no source given.
Heart rate variability threshold = 20 ms
Hand-set value, no source given.
Temperature alert threshold = 30 °C
Hand-set value, no source given.
Humidity alert threshold = 70%
Hand-set value, no source given.

assumptions (5)

domain assumption The named clinical guidelines (AANEM, ACR, ESC, ATS, CDC) support the specific alert thresholds used in the system.
The paper cites these guidelines as a block in Section III-D but gives no specific quotes or threshold derivations.
domain assumption Wearable EMG, PPG, and accelerometer sensors provide accurate and reliable data for clinical decisions in home settings.
Assumed in Section III-B without supporting evidence.
domain assumption Point-of-care testing devices such as the iCare-1300 produce lab-quality CK, ALT, and AST measurements from finger-prick samples.
Assumed in Section III-B; no device validation or citation is provided.
domain assumption Sufficient labeled historical MD data exist to train ARIMA, LSTM, Random Forest, and SVM models.
Assumed in Section IV, Model Training and Validation; no dataset is identified.
ad hoc to paper A linear combination of normalized biomarkers with the specified weights represents patient risk.
Eq. (1) defines RiskScore; no theoretical or empirical justification is given.

invented entities (1)

iCare-1300 POCT device
purpose: Measures CK, ALT, and AST from small blood samples in the proposed system.
Named in the data transmission table (Fig 3) but never cited, specified, or validated; it may be a placeholder or nonexistent product.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Predictive Modeling For Real-Time Personalized Health Monitoring in Muscular Dystrophy Management." pith.science (2026). https://pith.science/paper/OINYTB6A

@misc{pith2026241114923,
  author       = {Pith},
  title        = {Pith review of: Predictive Modeling For Real-Time Personalized Health Monitoring in Muscular Dystrophy Management},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OINYTB6A}},
  note         = {Machine review of arXiv:2411.14923}
}

read the original abstract

Muscular Dystrophy is a group of genetic disorders that progressively affect the strength and functioning of muscles, thereby affecting millions of people worldwide. The lifetime nature of MD requires continuous follow-up care due to its progressive nature. This conceptual paper proposes an Internet of Things-based system to support the management of MD through remote, multi-dimensional monitoring of patients in order to provide real-time health status updates. Traditional methods have failed to give actionable data in real time, hence denying healthcare providers the opportunity to make evidence-based decisions. Technology-driven approaches are urgently needed to provide deep insights into disease progression and patient health. It aims to enhance treatment strategies, enabling patients to better manage their condition and giving healthcare professionals more confidence in their management decisions.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 26 canonical work pages

[7]

Guidelines on the Use of Electrodiagnostic Testing and Biomarkers in Neuromuscular Disorders

American Association of Neuromuscular & Electrodiagnostic Medicine (AANEM). (2020). "Guidelines on the Use of Electrodiagnostic Testing and Biomarkers in Neuromuscular Disorders."

work page 2020
[11]

Heat- Related Illness: Guidelines for Prevention and Response

Centers for Disease Control and Prevention (CDC). (2017). "Heat- Related Illness: Guidelines for Prevention and Response."

work page 2017
[1]

Muscular Dystrophy: The facts,

R. Emery, “Muscular Dystrophy: The facts,” Oxford University Press, 2020

work page 2020
[2]

Wearable sensor systems for chronic disease management: A review,

J. Liu, M. Chen, and S. Li, “Wearable sensor systems for chronic disease management: A review,” IEEE Reviews in Biomedical Engineering, vol. 14, pp. 300-315, 2021

work page 2021
[3]

The application of IoT in personalized healthcare for chronic diseases,

A. Singh, B. Patel, and C. Wong, “The application of IoT in personalized healthcare for chronic diseases,” Journal of Medical Internet Research, vol. 24, no. 7, pp. 50-67, 2022

work page 2022
[4]

Wearable technology in healthcare: New frontiers in patient monitoring,

S. Johnson, “Wearable technology in healthcare: New frontiers in patient monitoring,” unpublished

work page
[5]

IoT and smart healthcare systems: An overview,

M. Klein and P. Schwartz, “IoT and smart healthcare systems: An overview,” in press

work page
[6]

The role of cloud computing in remote healthcare solutions,

L. Santos, “The role of cloud computing in remote healthcare solutions,” IEEE Transactions on Cloud Computing, vol. 8, no. 2, pp. 230-245, 2020

work page 2020

Show all 26 references

[8]

Standards for Monitoring Liver Enzymes and Systemic Inflammation in Chronic Conditions

American College of Rheumatology (ACR). (2019). "Standards for Monitoring Liver Enzymes and Systemic Inflammation in Chronic Conditions."

2019
[9]

Heart Rate Variability: Standards of Measurement, Physiological Interpretation, and Clinical Use

European Society of Cardiology (ESC). (2016). "Heart Rate Variability: Standards of Measurement, Physiological Interpretation, and Clinical Use."

2016
[10]

Guidelines for the Six- Minute Walk Test and Oxygen Saturation Measurements

American Thoracic Society (ATS). (2012). "Guidelines for the Six- Minute Walk Test and Oxygen Saturation Measurements."

2012
[12]

Physical Activity Monitoring in Patients with Neurological Disorders: A Review of Novel Body-Worn Devices

L. Walsh, “Physical Activity Monitoring in Patients with Neurological Disorders: A Review of Novel Body-Worn Devices.”
[13]

Exploiting Smart E-Health Gateways at the Edge of Healthcare Internet-of-Things: A Fog Computing Approach

W. Ahmed, et al. “Exploiting Smart E-Health Gateways at the Edge of Healthcare Internet-of-Things: A Fog Computing Approach.”
[14]

A Systematic Review of Wearable Sensors for Monitoring Physical Activity

Y. Wang, et al. “A Systematic Review of Wearable Sensors for Monitoring Physical Activity.”
[15]

Wearable Technology in Health Care: A Review of the Literature,

“Wearable Technology in Health Care: A Review of the Literature,” PMC8945966
[16]

Wearable Devices and Health: Opportunities and Challenges

National Institute of Health, “Wearable Devices and Health: Opportunities and Challenges.”
[18]

Wearable Biosensors and IoT in Chronic Disease Management

Brown, T., & Green, M. (2019). "Wearable Biosensors and IoT in Chronic Disease Management." Journal of Medical Internet Research, 21(4), e12345

2019
[19]

Using Machine Learning for Real- Time Health Monitoring

Taylor, S., & Miller, D. (2021). "Using Machine Learning for Real- Time Health Monitoring." IEEE Journal of Biomedical and Health Informatics, 25(3), 789-798

2021
[20]

Dynamic Risk Scoring in Chronic Disease: An IoT Approach

Johnson, R., & Allen, M. (2020). "Dynamic Risk Scoring in Chronic Disease: An IoT Approach." Journal of Biomedical Research, 33(6), 1124-1135

2020
[21]

Multimodal Health Monitoring for Early Detection of Neurological Decline

Chen, L., & West, D. (2019). "Multimodal Health Monitoring for Early Detection of Neurological Decline." Journal of Neural Engineering, 16(3), 035002

2019
[22]

The Impact of Environmental Factors on Physical Performance in Neuromuscular Disorders

Greenfield, J., & Kim, H. (2018). "The Impact of Environmental Factors on Physical Performance in Neuromuscular Disorders." Neurorehabilitation and Neural Repair, 32(7), 612-620

2018
[23]

Falls Prevention and Injury Response in Neurological Conditions

Centers for Disease Control and Prevention (CDC). (2020). "Falls Prevention and Injury Response in Neurological Conditions."

2020
[24]

Monitoring and Managing Muscular Dystrophy: A Clinical Overview

Anderson, C., & Brown, S. (2020). "Monitoring and Managing Muscular Dystrophy: A Clinical Overview." Muscular Dystrophy Journal, 29(2), 101-115

2020
[25]

Toma, M., & Wei, O. C. (2023). Predictive modeling in medicine. Encyclopedia, 3(2), 590-601

2023
[26]

Large Language Models in Healthcare and Medical Domain: A Review, MDPI, August 7, 2024

2024
[27]

Recent Advances in Large Language Models for Healthcare, MDPI, April 16, 2024

2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [7]

Guidelines on the Use of Electrodiagnostic Testing and Biomarkers in Neuromuscular Disorders

American Association of Neuromuscular & Electrodiagnostic Medicine (AANEM). (2020). "Guidelines on the Use of Electrodiagnostic Testing and Biomarkers in Neuromuscular Disorders."

work page 2020

[2] [11]

Heat- Related Illness: Guidelines for Prevention and Response

Centers for Disease Control and Prevention (CDC). (2017). "Heat- Related Illness: Guidelines for Prevention and Response."

work page 2017

[3] [1]

Muscular Dystrophy: The facts,

R. Emery, “Muscular Dystrophy: The facts,” Oxford University Press, 2020

work page 2020

[4] [2]

Wearable sensor systems for chronic disease management: A review,

J. Liu, M. Chen, and S. Li, “Wearable sensor systems for chronic disease management: A review,” IEEE Reviews in Biomedical Engineering, vol. 14, pp. 300-315, 2021

work page 2021

[5] [3]

The application of IoT in personalized healthcare for chronic diseases,

A. Singh, B. Patel, and C. Wong, “The application of IoT in personalized healthcare for chronic diseases,” Journal of Medical Internet Research, vol. 24, no. 7, pp. 50-67, 2022

work page 2022

[6] [4]

Wearable technology in healthcare: New frontiers in patient monitoring,

S. Johnson, “Wearable technology in healthcare: New frontiers in patient monitoring,” unpublished

work page

[7] [5]

IoT and smart healthcare systems: An overview,

M. Klein and P. Schwartz, “IoT and smart healthcare systems: An overview,” in press

work page

[8] [6]

The role of cloud computing in remote healthcare solutions,

L. Santos, “The role of cloud computing in remote healthcare solutions,” IEEE Transactions on Cloud Computing, vol. 8, no. 2, pp. 230-245, 2020

work page 2020