A Robust TabNet-Based Multi-Classification Algorithm for Infrared Spectral Data of Chinese Herbal Medicine with High-Dimensional Small Samples.

Robust classification algorithms for high-dimensional, small-sample datasets are valuable in practical applications. Faced with the infrared spectroscopic dataset with 568 samples and 3448 wavelengths (features) to identify the origins of Chinese medicinal materials, this paper proposed a novel embedded multiclassification algorithm, ITabNet, derived from the framework of TabNet. Firstly, a refined data pre-processing (DP) mechanism was designed to efficiently find the best adaptive one among 50 DP methods with the help of Support Vector Machine (SVM). Following this, an innovative focal loss function was designed and joined with a cross-validation experiment strategy to mitigate the impact of sample imbalance on algorithm. Detailed investigations on ITabNet were conducted, including comparisons of ITabNet with SVM for the conditions of DP and Non-DP, GPU and CPU computer settings, as well as ITabNet against XGBT (Extreme Gradient Boosting). The numerical results demonstrate that ITabNet can significantly improve the effectiveness of prediction. The best accuracy score is 1.0000, and the best Area Under the Curve (AUC) score is 1.0000. Suggestions on how to use models effectively were given. Furthermore, ITabNet shows the potential to apply the analysis of medicinal efficacy and chemical composition of medicinal materials. The paper also provides ideas for multi-classification modeling data with small sample size and high-dimensional feature.

Segmental abnormalities of superior longitudinal fasciculus microstructure in patients with schizophrenia, bipolar disorder, and attention-deficit/hyperactivity disorder: An automated fiber quantification tractography study.

Introduction: Superior longitudinal fasciculus (SLF) is a white matter (WM) tract that connects the frontal, parietal and temporal lobes. SLF integrity has been widely assessed in neuroimaging studies of psychiatric disorders, such as schizophrenia (SZ), bipolar disorder (BD), and attention-deficit/hyperactivity disorder (ADHD). However, prior studies have revealed inconsistent findings and comparisons across disorders have not been fully examined. Methods: Here, we obtained data for 113 patients (38 patients with SZ, 40 with BD, 35 with ADHD) and 94 healthy controls from the UCLA Consortium for Neuropsychiatric Phenomic LA5c dataset. We assessed the integrity of 20 major WM tracts with a novel segmentation method by automating fiber tract quantification (AFQ). The AFQ divides each tract into 100 equal parts along the direction of travel, with fractional anisotropy (FA) of each part taken as a characteristic. Differences in FA among the four groups were examined. Results: Compared to healthy controls, patients with SZ showed significantly lower FA in the second half (51-100 parts) of the SLF. No differences were found between BD and healthy controls, nor between ADHD and healthy controls. Results also demonstrated that patients with SZ showed FA reduction in the second half of the SLF relative to patients with BP. Moreover, greater FA in patients in SLF was positively correlated with the manic-hostility score of the Brief Psychiatry Rating scale. Discussion: These findings indicated that differences in focal changes in SLF might be a key neurobiological abnormality contributing to characterization of these psychiatric disorders.