TY  - JOUR
AU  - Wyton, Ryan
AU  - Hasan, Raza
PY  - 2026
DA  - 2026/09/27
TI  - Non-Invasive Parkinson's Disease Screening from Vocal Biomarkers: A Subject-Level Machine Learning Framework with Bayesian Optimisation and Explainable AI
JO  - ICCK Transactions on Sensing, Communication, and Control
T2  - ICCK Transactions on Sensing, Communication, and Control
JF  - ICCK Transactions on Sensing, Communication, and Control
VL  - 3
IS  - 3
SP  - 197
EP  - 209
DO  - 10.62762/TSCC.2026.365257
UR  - https://www.icck.org/article/abs/TSCC.2026.365257
KW  - acoustic sensing
KW  - vocal biomarkers
KW  - multi-statistic feature fusion
KW  - remote health monitoring
KW  - intelligent decision support
KW  - explainable artificial intelligence
KW  - Bayesian optimisation
KW  - Parkinson's disease
AB  - An estimated ten million people live with Parkinson's, yet timely diagnosis remains constrained by specialist assessment, costly imaging, and unequal care access. Phonation analysis offers a low-cost alternative: dopaminergic degeneration causes vocal impairments years before motor symptoms emerge, enabling community screening. However, existing ML approaches are limited by recording-level data leakage and opacity. This paper presents a four-phase acoustic framework-spanning signal acquisition, feature processing, and interpretable decision support-applied to a multi-type Parkinson's speech dataset (40 training, 28 blind-test). The framework enforces subject-level partitioning via GroupKFold to eliminate leakage, derives 104 participant-level acoustic features through multi-statistic aggregation (mean, median, standard deviation, interquartile range), and employs Bayesian optimisation using Optuna. The optimised SVM achieved 90.00% cross-validated accuracy, a 12.50 percentage-point improvement over the 77.50% benchmark. Cross-validated sensitivity was 95.00%, specificity 85.00%, Youden Index \(J=0.80\), Brier Score \(\mathrm{BS}=0.1049\). External validation on a held-out vowel-only cohort achieved 85.71% sensitivity, surpassing the benchmark. A permutation test (\(p=0.349\)) indicated no significant linear mapping between acoustic features and motor severity, motivating binary classification. Shapley Additive Explanations identified interquartile range of degree-of-voice-breaks and shimmer amplitude variability as primary diagnostic drivers, aligning with neuroacoustic evidence for aperiodic phonation. These results show that leakage prevention, variance-sensitive engineering, and transparent attribution together constitute a viable foundation for remote, telehealth-linked pre-screening instruments in resource-constrained settings.
SN  - 3068-9287
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Wyton2026NonInvasiv,
  author = {Ryan Wyton and Raza Hasan},
  title = {Non-Invasive Parkinson's Disease Screening from Vocal Biomarkers: A Subject-Level Machine Learning Framework with Bayesian Optimisation and Explainable AI},
  journal = {ICCK Transactions on Sensing, Communication, and Control},
  year = {2026},
  volume = {3},
  number = {3},
  pages = {197-209},
  doi = {10.62762/TSCC.2026.365257},
  url = {https://www.icck.org/article/abs/TSCC.2026.365257},
  abstract = {An estimated ten million people live with Parkinson's, yet timely diagnosis remains constrained by specialist assessment, costly imaging, and unequal care access. Phonation analysis offers a low-cost alternative: dopaminergic degeneration causes vocal impairments years before motor symptoms emerge, enabling community screening. However, existing ML approaches are limited by recording-level data leakage and opacity. This paper presents a four-phase acoustic framework-spanning signal acquisition, feature processing, and interpretable decision support-applied to a multi-type Parkinson's speech dataset (40 training, 28 blind-test). The framework enforces subject-level partitioning via GroupKFold to eliminate leakage, derives 104 participant-level acoustic features through multi-statistic aggregation (mean, median, standard deviation, interquartile range), and employs Bayesian optimisation using Optuna. The optimised SVM achieved 90.00\% cross-validated accuracy, a 12.50 percentage-point improvement over the 77.50\% benchmark. Cross-validated sensitivity was 95.00\%, specificity 85.00\%, Youden Index \(J=0.80\), Brier Score \(\mathrm{BS}=0.1049\). External validation on a held-out vowel-only cohort achieved 85.71\% sensitivity, surpassing the benchmark. A permutation test (\(p=0.349\)) indicated no significant linear mapping between acoustic features and motor severity, motivating binary classification. Shapley Additive Explanations identified interquartile range of degree-of-voice-breaks and shimmer amplitude variability as primary diagnostic drivers, aligning with neuroacoustic evidence for aperiodic phonation. These results show that leakage prevention, variance-sensitive engineering, and transparent attribution together constitute a viable foundation for remote, telehealth-linked pre-screening instruments in resource-constrained settings.},
  keywords = {acoustic sensing, vocal biomarkers, multi-statistic feature fusion, remote health monitoring, intelligent decision support, explainable artificial intelligence, Bayesian optimisation, Parkinson's disease},
  issn = {3068-9287},
  publisher = {Institute of Central Computation and Knowledge}
}