Grouped-sampling technique to deal with unbalance in Raman spectral data modeling.

Due to limitations in disease prevalence and hospital specificity, spectral data are often collected with unbalanced sample size. To solve this problem, a new sampling method - grouped-sampling was proposed in this research, which is shown to be effective for unbalanced data. It avoids over-fitting of over-sampling and overcomes under-sampling utilization of under-sampling. In this study, we applied grouped-sampling to two unbalanced datasets where the sample proportions are 199:40 and 75:225. And then verified from two classic models: PCA-SVM (Principal Component Analysis-Support Vector Machine) and the deep learning algorithm GoogLeNet. The accuracy of these two datasets were 85.11% and 96.15% in PCA-SVM and 85.10% and 84.61% on GoogLeNet. Also, the F1-score were evaluated to measure the classification balance of sampling method, and result shows that F1-score of grouped-sampling is always the highest compared to over-sampling and under-sampling. In summary, compared to traditional sampling methods, grouped-sampling performs better on prediction for classes with smaller sample size, which means grouped-sampling can improve the balance of classification results and the potential of practical application. Therefore, we develop a group sampling method that distinguishes between under- and over-sampling, which greatly improves the accuracy and balance of predictions for unbalanced samples.

Application of serum mid-infrared spectroscopy combined with an ensemble learning method in rapid diagnosis of gliomas.

The diffuse growth of glioma cells leads to gliomatosis, which has less cure rate and high mortality. As the severity deepens, the treatment difficulty and mortality of glioma patients gradually increase. Therefore, a rapid and non-invasive diagnostic technique is very important for glioma patients. The target of this study is to classify contract subjects and glioma patients by serum mid-infrared spectroscopy combined with an ensemble learning method. The spectra were normalized and smoothed, and principal component analysis (PCA) was utilized for dimensionality reduction. Particle swarm optimization-support vector machine (PSO-SVM), decision tree (DT), logistic regression (LR) as well as random forest (RF) were used as base classifiers, and AdaBoost integrated learning was introduced. AdaBoost-SVM, AdaBoost-LR, AdaBoost-RF and AdaBoost-DT models were established to discriminate glioma patients. The single classification accuracy of the four models for the test set was 87.14%, 90.00%, 92.00% and 90.86%, respectively. For the purpose of further improving the prediction accuracy, the four models were fused at decision level, and the final classification accuracy of the test set reached 94.29%. Experiments show that serum infrared spectroscopy combined with the ensemble learning method algorithm shows wonderful potential in non-invasive, fast and precise identification of glioma patients, and can also be used for reference in intelligent diagnosis of other diseases.

Human serum mid-infrared spectroscopy combined with machine learning algorithms for rapid detection of gliomas.

Glioma has a low cure rate and a high mortality rate. Therefore, correct diagnosis and treatment are essential for patients. This research aims to use mid-infrared spectroscopy combined with machine learning algorithms to identify patients with glioma. The glioma infrared spectra and the control group serum are smoothed and normalized, then the principal component analysis (PCA) algorithm is used to reduce the data dimensionality, and finally, the particle swarm optimization-support vector machine (PSO-SVM), backpropagation (BP) neural network and decision tree (DT) model are established. The classification accuracy of the three models was 92.00 %, 91.83 %, 87.20 %, and the AUC values were 0.919, 0.945, and 0.866, respectively. The results show that PCA-PSO-SVM has a better classification effect. This study shows that mid-infrared spectroscopy combined with machine learning algorithms has great potential in the application of non-invasive, rapid and accurate identification of glioma patients.