Classification of alcoholic EEG signals using wavelet scattering transform-based features.

Following the research question and the relevant dataset, feature extraction is the most important component of machine learning and data science pipelines. The wavelet scattering transform (WST) is a recently developed knowledge-based feature extraction technique and is structurally like a convolutional neural network (CNN). It preserves information in high-frequency, is insensitive to signal deformations, and generates low variance features of real-valued signals generally required in classification tasks. With data from a publicly-available UCI database, we investigated the ability of WST-based features extracted from multichannel electroencephalogram (EEG) signals to discriminate 1.0-s EEG records of 20 male subjects with alcoholism and 20 male healthy subjects. Using record-wise 10-fold cross-validation, we found that WST-based features, inputted to a support vector machine (SVM) classifier, were able to correctly classify all alcoholic and normal EEG records. Similar performances were achieved with 1D CNN. In contrast, the highest independent-subject-wise mean 10-fold cross-validation performance was achieved with WST-based features fed to a linear discriminant (LDA) classifier. The results achieved with two 10-fold cross-validation approaches suggest that the WST together with a conventional classifier is an alternative to CNN for classification of alcoholic and normal EEGs. WST-based features from occipital and parietal regions were the most informative at discriminating between alcoholic and normal EEG records.

Laser induced breakdown spectroscopy with machine learning reveals lithium-induced electrolyte imbalance in the kidneys.

Lithium is a major psychiatric medication, especially as long-term maintenance medication for Bipolar Disorder. Despite its effectiveness, lithium has side-effects, such as on renal function. In this study, lithium was administered to adult rats. This animal model of renal function was validated by measuring blood lithium, urea nitrogen (BUN), and thyroxine (T4) using inductively-coupled plasma mass spectrometry and enzyme-linked immunosorbent assay. The kidneys were analyzed by laser induced breakdown spectroscopy (LIBS) with 1064 nm ablation and 300-900 nm detection. Principal components analysis (PCA), radial visualization, and random forest classification were performed on the LIBS spectra for multi-element prediction and classification. Lithium at 0.34 mmol/L was detected in the blood of lithium treated subjects only. BUN was increased (6.6 vs. 5.3 mmol/L) and T4 decreased (58.12 vs. 51.4 mmol/L) in the blood of lithium subjects compared with controls, indicating renal abnormalities. LIBS detected lithium at 2.3 mmol/kg in the kidneys of lithium subjects only. Calcium was also observed to be reduced in lithium subjects, compared with controls. Subsequent PCA observed a change in the balance of sodium and potassium in the kidneys. These are key electrolytes in the body. Importantly, partial least squares regression showed that standard clinical measurements, such as the blood tests, can be used to predict kidney electrolyte measurements, which typically cannot be performed in humans. Overall, lithium accumulates in the kidneys and adversely affects renal function. The effects are likely related to electrolyte imbalance. LIBS with machine learning analysis has potential to improve clinical management of renal side-effects in patients on lithium medication.