The aim of this study was to determine which supervised machine learning (ML) algorithm can most accurately classify people with Parkinson’s disease (pwPD) from speed-matched healthy subjects (HS) based on a selected minimum set of IMU-derived gait features. Twenty-two gait features were extrapolated from the trunk acceleration patterns of 81 pwPD and 80 HS, including spatiotemporal, pelvic kinematics, and acceleration-derived gait stability indexes. After a three-level feature selection procedure, seven gait features were considered for implementing five ML algorithms: support vector machine (SVM), artificial neural network, decision trees (DT), random forest (RF), and K-nearest neighbors. Accuracy, precision, recall, and F1 score were calculated. SVM, DT, and RF showed the best classification performances, with prediction accuracy higher than 80% on the test set. The conceptual model of approaching ML that we proposed could reduce the risk of overrepresenting multicollinear gait features in the model, reducing the risk of overfitting in the test performances while fostering the explainability of the results.

The interpretability of gait analysis studies in people with rare diseases, such as those with primary hereditary cerebellar ataxia (pwCA), is frequently limited by the small sample sizes and unbalanced datasets. The purpose of this study was to assess the effectiveness of data balancing and generative artificial intelligence (AI) algorithms in generating synthetic data reflecting the actual gait abnormalities of pwCA. Gait data of 30 pwCA (age: 51.6 ± 12.2 years; 13 females, 17 males) and 100 healthy subjects (age: 57.1 ± 10.4; 60 females, 40 males) were collected at the lumbar level with an inertial measurement unit. Subsampling, oversampling, synthetic minority oversampling, generative adversarial networks, and conditional tabular generative adversarial networks (ctGAN) were applied to generate datasets to be input to a random forest classifier. Consistency and explainability metrics were also calculated to assess the coherence of the generated dataset with known gait abnormalities of pwCA. ctGAN significantly improved the classification performance compared with the original dataset and traditional data augmentation methods. ctGAN are effective methods for balancing tabular datasets from populations with rare diseases, owing to their ability to improve diagnostic models with consistent explainability.

The interictal deficit of habituation to repetitive visual stimuli in migraine patients could be due to deficient intracortical inhibition and/or to low cortical pre-activation levels. Which of these abnormalities contributes more to the habituation deficit cannot be determined with the common methods used to record transient visual responses. We investigated lateral inhibition in the visual cortex during the migraine cycle and in healthy subjects by using differential temporal modulations of radial windmill-dartboard (WD) or partial-windmill (PW) visual patterns. Transient (TR-VEP) and steady-state visual-evoked potentials (SS-VEP) were recorded in 65 migraine patients (21 without and 22 with aura between attacks; 22 patients during an attack) and in 21 healthy volunteers (HV). Three stimulations were used in each subject: classic checkerboard pattern (contrast-reversion 3.1Hz), WD and PW (contrast-reversion ~4Hz). For each randomly presented stimulation protocol, 600 sweeps were acquired and off-line partitioned in 6 blocks of 100. Fourier analysis allowed data to extract in SS-VEP the fundamental (1H) and the second harmonic (2H) components that reflect respectively short-(WD) and long- range lateral inhibition (attenuation of 2H in WD compared to PW). Compared to HV, migraineurs recorded interictally had significantly less habituation of the N1-P1 TR-VEP component over subsequent blocks and they tended to have a smaller 1^st block amplitude. 1H amplitude in the 1^st block of WD SS-VEP was significantly greater than in HV and habituated in successive blocks, contrasting with an amplitude increase in HV. Both the interictal TR-VEP and SS-VEP abnormalities normalized during an attack. There was no significant between group difference in the PW 2H amplitude and its attenuation. When data of HV and migraine patients were combined, the habituation slope of WD-VEP 1H was negatively correlated with that of TR-VEP N1-P1 and with number of days since the last migraine attack. These results are in favour of a migraine cycle-dependent imbalance between excitation and inhibition in the visual cortex. We hypothesize that an interictal hypoactivity of monaminergic pathways may cause a functional disconnection of the thalamus in migraine leading to an abnormal intracortical short-range lateral inhibition that could contribute to the habituation deficit observed during stimulus repetition.