AI and Neurological Rare & Neuroimmune Diseases ── Parkinson's ── 2026-11-05
Parkinson's disease and parkinsonism: treatment, trials and AI
- Abstract
OBJECTIVE: To apply an integrated causal-inference and welfare-optimization framework to Iranian Parkinson's disease (PD) inpatient insurance claims, estimating how resource-allocation decisions relate to recovery, discharge risk, cost and financial burden; testing whether these relationships vary across patient subgroups; and assessing whether budget-neutral reallocation could improve efficiency and equity simultaneously. SETTING: Six provinces of Iran, using de-identified administrative inpatient claims obtained from the Iran Health Insurance Organization for admissions with a primary or secondary diagnosis of PD during the 2023-2025 observation period. DESIGN: Retrospective claims-based cohort study. Transaction-level records (n = 21,121) were aggregated into 1,813 hospitalization episodes and analyzed through a six-stage pipeline: gradient-boosted predictive models; double machine learning and augmented inverse-probability weighting for average treatment effects, restricted to an admission-time-only, pre-treatment severity adjustment; causal forests for subgroup heterogeneity; concentration-index decomposition for socioeconomic inequality; constrained integer-linear-programming optimization for reallocation scenarios; and bootstrap, E-value and design-effect sensitivity analyses. RESULTS: Predictive models showed acceptable-to-strong discrimination (recovery, AUC-ROC 0.709; unsuccessful-discharge proxy, AUC-ROC 0.870) and explained 81.3% of variance in total cost. Prolonged length of stay, treated as an exploratory, within-episode exposure rather than a causal one because it is realized during the hospitalization itself, was associated with a 6.4-percentage-point lower discharge proxy and 200% higher cost (E-value limit 1.93-4.04). Higher insurance-coverage generosity, the study's pre-treatment, policy-assignable exposure, was associated with an approximately 40% reduction in cost (E-value limit 2.19) and no significant change in the discharge proxy; effects were consistent across age, severity, hospital sector, rurality and socioeconomic rank, with an attenuated estimate in a small private-hospital subgroup. Recovery was concentrated among higher-socioeconomic episodes, while cost, out-of-pocket burden and length of stay were concentrated among lower-socioeconomic episodes; age and hospital ownership, not disease severity (which contributed only -0.8% to +6.2% of the observed gradients), were the leading correlates. A coverage-based, budget-neutral reallocation scenario reached 35.8% of the cohort within the observed budget. Ninety-eight percent of significant associations remained robust to plausible within-patient clustering, the principal limitation arising from the absence of a patient identifier in these claims data. CONCLUSION: Insurance-coverage generosity showed the strongest evidence of a causal, cost-reducing effect, while socioeconomic inequity in PD hospitalization outcomes was driven primarily by administrative rather than clinical-severity factors. The reallocation scenarios illustrate model-based, budget-neutral efficiency-equity gains and should be read as decision-support for phased piloting, not as empirically demonstrated policy effects. Patient-level claims linkage and validated clinical severity measures are needed to extend this framework toward genuine longitudinal outcomes.
Journal IF-equivalent: 3.5 (OpenAlex 2-year mean citedness, value as of 2026-11-03, retrieved 2026-11-05; not the official Clarivate IF)Reference: Momahhed SS, Haghighathoseini A. Resource allocation under uncertainty and equity constraints in Parkinson’s disease care: An econometrically grounded machine learning framework. PLoS One. 2026 Nov 3;21(11):e0360944. doi:10.1371/journal.pone.0360944. PMID: 42976191.Checked: Abstract only - Abstract
Class imbalance is a common problem in speech-based severity detection, which can result in biased models and poor recognition of minority severity classes. Generative Adversarial Networks (GANs) have been widely used for data augmentation, but very few studies have comparatively evaluated GAN-based augmentation at different stages of the speech-processing pipeline for assessing Parkinson's hypokinetic dysarthria severity. This paper compares GAN-generated speech (GGS) and GAN-generated image (GGI) augmentation methods for addressing class imbalance in low-frequency cepstrogram-based severity detection using the PC-GITA dataset. GAN-based augmentation was used to generate balanced training datasets using two different augmentation pipelines. The original speech and synthetic speech generated using GGS were converted into low-frequency cepstrogram representations, whereas GGI directly generated low-frequency cepstrogram images. These representations were used as inputs to a CNN trained and evaluated under Speaker-independent (SI) and Speaker-dependent (SD) conditions. In the speaker-independent evaluation, repeated five-fold speaker-stratified cross-validation with five independent random seeds yielded a mean accuracy of 56.47% ± 3.40% for GGS compared to 55.47% ± 3.35% for GGI and 53.13% ± 3.26% for the original-dataset baseline. GGS showed higher mean accuracy, precision, and specificity, whereas GGI showed higher mean recall and F1-score. In the speaker-dependent evaluation, GGI achieved the highest mean accuracy of 76.48% ± 4.09% compared to 73.04% ± 3.63% for GGS and 73.04% ± 3.25% for the original-dataset baseline. However, the speaker-dependent results represent an optimistic upper-bound evaluation because speaker identity was shared between the training and evaluation partitions. In general, the two GAN-based augmentation pipelines showed potential to address class imbalance under the conditions studied, while their relative performance depended on the evaluation protocol and the classification measure.
Journal IF-equivalent: 6.0 (OpenAlex 2-year mean citedness, value as of 2026-10-09, retrieved 2026-10-10; not the official Clarivate IF)Reference: S A, Ramachandran P, M RK. Comparative analysis of GAN-based augmentation pipelines for Parkinson's hypokinetic dysarthria severity detection using deep learning. Front Artif Intell. 2026 Oct 20;9:1951063. doi:10.3389/frai.2026.1951063; PMCID: PMC13727737.Checked: Full text checked