OSCR

Self-Explaining Neural Networks for Transparent Parkinson's Disease Screening.

Overview

Authors: Mahmoud E Farfoura1, Ahmad A A Alkhatib1, Tee Connie2
  1. Cybersecurity Department, Al-Zaytoonah University of Jordan, Amman 11733, Jordan
  2. Faculty of Information Science & Technology, Multimedia University, Jalan Ayer Keroh Lama, Melaka 75450, Malaysia
Institutions: Al-Zaytoonah University of Jordan (Jordan); Multimedia University (Malaysia)
Journal: Sensors (Basel, Switzerland), volume 26, issue 9, article 2671
Dates: received 22 February 2026; accepted 21 April 2026; published online 25 April 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.3390/s26092671 · PMID 42122390 · PMCID PMC13165671 · OpenAlex W7156895585
Open access: gold, a free copy (OpenAlex)
Categories: other (modality), human (organism), Parkinson's (population), clinical / translational (subfield)
Methods: Preprocessing, Connectivity, Statistics, Machine learning
Keywords: Parkinson’s disease detection, self-explaining neural networks, explainable artificial intelligence, gait analysis, ground reaction force, intrinsic interpretability, residual CNN, concept learning, wearable sensors, clinical decision support
MeSH: Neural Networks, Computer*, Parkinson Disease*, Convolutional Neural Networks, Deep Learning, Gait, Humans, ROC Curve (* major topic)
Topic: Balance, Gait, and Falls Prevention (Physical Therapy, Sports Therapy and Rehabilitation, Health Professions), according to OpenAlex
Funding: Malaysia-Jordan Matching Fund (MMUI/240092)
Citations: not cited yet (Europe PMC); 29 references in the paper

Abstract

Transparent clinical decision-making remains a critical barrier to deploying deep learning in medical diagnosis. Post hoc explanation methods approximate model behaviour after training but cannot guarantee that explanations faithfully reflect the underlying reasoning. This study proposes a Self-Explaining Neural Network (SENN) for Parkinson’s Disease (PD) screening via Ground Reaction Force (GRF) gait analysis, enforcing intrinsic interpretability through learnable basis concepts and input-dependent relevance scores computed jointly with the prediction. The architecture combines a four-block residual CNN backbone with stochastic depth regularisation, a 16-concept encoder with diversity and stability constraints, and temperature-scaled probability calibration for reliable clinical operating points. Evaluated on the PhysioNet Gait in Parkinson’s Disease dataset (306 subjects, 16 GRF sensors per foot), SENN achieves a subject-level ROC-AUC of 0.916 [95% CI: 0.867–0.964], sensitivity of 0.913 [0.862–0.963], specificity of 0.671 [0.485–0.858], and Average Precision of 0.942 [0.918–0.967], reported across five independent random seeds. Comparative evaluation against four deep learning baselines—CNN-Residual, BiLSTM, CNN-LSTM, and CNN-Attention—confirms that the interpretability constraints impose no statistically significant reduction in discriminative performance, with all pairwise ROC-AUC confidence intervals overlapping. Concept-level analysis reveals that the three most discriminative concepts correspond to disrupted midfoot loading patterns, increased step-length variability, and bilateral cadence asymmetry—all established biomechanical hallmarks of parkinsonian gait—providing clinically grounded, patient-specific explanations without post hoc approximation. These findings demonstrate that rigorous intrinsic interpretability and competitive predictive accuracy are simultaneously achievable in deep gait analysis, supporting the clinical adoption of transparent diagnostic AI.

Reproduced under the paper's license (CC BY), from the paper cited above.

Code

The paper links to its data and to no code of its authors: see the Data section.

Tracing map

A tracing map links a paper to the code its authors published: this paper has none, so it has no map.

Data

Datasets cited

Data Availability Statement

You can access the PhysioNet GRF dataset using the following link: https://physionet.org/content/gaitpdb/1.0.0/ (accessed on 3 January 2026).

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 30 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 3 authors, 10 keywords, 7 MeSH terms, 1 funder, 25 references.

Cite

This paper

Farfoura, M. E., Alkhatib, A. A. A., & Connie, T. (2026). Self-Explaining Neural Networks for Transparent Parkinson's Disease Screening. Sensors (Basel, Switzerland), 26(9), 2671. https://doi.org/10.3390/s26092671

BibTeX

@article{farfoura2026self,
author = {Farfoura, Mahmoud E and Alkhatib, Ahmad A A and Connie, Tee},
title = {{Self-Explaining Neural Networks for Transparent Parkinson's Disease Screening}},
journal = {Sensors (Basel, Switzerland)},
year = {2026},
month = apr,
volume = {26},
number = {9},
pages = {2671},
publisher = {Multidisciplinary Digital Publishing Institute (MDPI)},
issn = {1424-8220},
doi = {10.3390/s26092671},
url = {https://doi.org/10.3390/s26092671},
pmid = {42122390},
pmcid = {PMC13165671}
}

RIS

TY - JOUR
AU - Farfoura, Mahmoud E
AU - Alkhatib, Ahmad A A
AU - Connie, Tee
TI - Self-Explaining Neural Networks for Transparent Parkinson's Disease Screening
T2 - Sensors (Basel, Switzerland)
J2 - Sensors (Basel)
PY - 2026
DA - 2026/04/25
VL - 26
IS - 9
SP - 2671
SN - 1424-8220
PB - Multidisciplinary Digital Publishing Institute (MDPI)
DO - 10.3390/s26092671
UR - https://doi.org/10.3390/s26092671
LA - en
ER -

CSL-JSON

{
"id": "10.3390/s26092671",
"type": "article-journal",
"title": "Self-Explaining Neural Networks for Transparent Parkinson's Disease Screening",
"container-title": "Sensors (Basel, Switzerland)",
"author": [
{
"family": "Farfoura",
"given": "Mahmoud E"
},
{
"family": "Alkhatib",
"given": "Ahmad A A"
},
{
"family": "Connie",
"given": "Tee"
}
],
"container-title-short": "Sensors (Basel)",
"volume": "26",
"issue": "9",
"page": "2671",
"DOI": "10.3390/s26092671",
"PMID": "42122390",
"PMCID": "PMC13165671",
"ISSN": "1424-8220",
"publisher": "Multidisciplinary Digital Publishing Institute (MDPI)",
"URL": "https://doi.org/10.3390/s26092671",
"language": "en",
"issued": {
"date-parts": [
[
2026,
4,
25
]
]
}
}

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.3390/s25227100
RecovGait: Occluded Parkinson's Disease Gait Reconstruction Using Unscented Tracking with Gated Initialization Technique.
Journal: Sensors (Basel, Switzerland)
In common: Parkinson's, other, 2 references, 2 authors
[2] doi:10.3390/diagnostics15162046
Parkinson's Disease Diagnosis and Severity Assessment from Gait Signals via Bayesian-Optimized Deep Learning.
Journal: Diagnostics (Basel, Switzerland)
In common: physionet.org/content/gaitpdb, Parkinson's, other, clinical / translational, 1 reference
[3] doi:10.1007/s10462-025-11347-y [code]
Applications of machine learning for computer-aided diagnosis of Parkinson's disease: progress and benchmark case study.
Journal: Artificial intelligence review
In common: physionet.org/content/gaitpdb, Parkinson's, clinical / translational, 1 reference
[4] doi:10.1371/journal.pone.0319826 [code]
Efficient quantification of Parkinson's disease severity using augmented time-series data.
Journal: PloS one
In common: physionet.org/content/gaitpdb, Parkinson's, 1 reference
[5] doi:10.1038/s41598-024-54680-y
Tuning attention based long-short term memory neural networks for Parkinson's disease detection using modified metaheuristics.
Journal: Scientific reports
In common: physionet.org/content/gaitpdb, Parkinson's, 1 reference
[6] doi:10.7717/peerj-cs.2031 [code]
Detecting Parkinson's disease from shoe-mounted accelerometer sensors using convolutional neural networks optimized with modified metaheuristics.
Journal: PeerJ. Computer science
In common: physionet.org/content/gaitpdb, Parkinson's, other
[7] doi:10.1038/s41598-025-07069-4
Enhanced EfficientNet-Extended Multimodal Parkinson's disease classification with Hybrid Particle Swarm and Grey Wolf Optimizer.
Journal: Scientific reports
In common: physionet.org/content/gaitpdb, Parkinson's
[8] doi:10.1038/s41598-026-47769-z [code]
A multimodal explainable artificial intelligence framework for interpretable Parkinson's disease prediction.
Journal: Scientific reports
In common: Parkinson's, 2 references
[9] doi:10.3390/s23094426
Context Recognition Algorithms for Energy-Efficient Freezing-of-Gait Detection in Parkinson's Disease.
Journal: Sensors (Basel, Switzerland)
In common: Parkinson's, other, 1 reference
[10] doi:10.2196/71560
Digital Biomarkers for Parkinson Disease: Bibliometric Analysis and a Scoping Review of Deep Learning for Freezing of Gait.
Journal: Journal of medical Internet research
In common: Parkinson's, clinical / translational, 1 reference

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.

Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.