Abstract： Diseases prognosis is often influenced by multiple factors, and some intricate non-linear relationships exist among those factors. Artificial neural network （ANN）, an artificial intelligence model, simulates the work mode of biological neurons and has a b capability to analyze multi-factor non-linear relationships. In recent years, ANN is increasingly applied in clinical medical fields, especially for the prediction of disease prognosis. This article focuses on the basic principles of ANN and its application in disease prognosis research.
The continuous left ventricle blood pressure prediction based on selected heart sound features was realized in this study. The experiments were carried out on three beagle dogs and the variations of cardiac hemodynamics were induced by various dose of epinephrine. The phonocardiogram, electrocardiogram and blood pressures in left ventricle were synchronously acquired. We obtained 28 valid recordings in this study. An artificial neural network was trained with the selected feature to predict left ventricular blood pressure and this trained network made a good performance. The results showed that the absolute average error was 7.3 mm Hg even though the blood pressures had a large range of fluctuation. The average correlation coefficient between the predicted and the measured blood pressure was 0.92. These results showed that the method in this paper was helpful to monitor left ventricular hemodynamics non-invasively and continuously.
Existing near-infrared non-invasive blood glucose detection modelings mostly detect multi-spectral signals with different wavelength, which is not conducive to the popularization of non-invasive glucose meter at home and does not consider the physiological glucose dynamics of individuals. In order to solve these problems, this study presented a non-invasive blood glucose detection model combining particle swarm optimization (PSO) and artificial neural network (ANN) by using the 1 550 nm near-infrared absorbance as the independent variable and the concentration of blood glucose as the dependent variable, named as PSO-2ANN. The PSO-2ANN model was based on two sub-modules of neural networks with certain structures and arguments, and was built up after optimizing the weight coefficients of the two networks by particle swarm optimization. The results of 10 volunteers were predicted by PSO-2ANN. It was indicated that the relative error of 9 volunteers was less than 20%; 98.28% of the predictions of blood glucose by PSO-2ANN were distributed in the regions A and B of Clarke error grid, which confirmed that PSO-2ANN could offer higher prediction accuracy and better robustness by comparison with ANN. Additionally, even the physiological glucose dynamics of individuals may be different due to the influence of environment, temper, mental state and so on, PSO-2ANN can correct this difference only by adjusting one argument. The PSO-2ANN model provided us a new prospect to overcome individual differences in blood glucose prediction.
General anesthesia is an essential part of surgery to ensure the safety of patients. Electroencephalogram (EEG) has been widely used in anesthesia depth monitoring for abundant information and the ability of reflecting the brain activity. The paper proposes a method which combines wavelet transform and artificial neural network (ANN) to assess the depth of anesthesia. Discrete wavelet transform was used to decompose the EEG signal, and the approximation coefficients and detail coefficients were used to calculate the 9 characteristic parameters. Kruskal-Wallis statistical test was made to these characteristic parameters, and the test showed that the parameters were statistically significant for the differences of the four levels of anesthesia: awake, light anesthesia, moderate anesthesia and deep anesthesia (P < 0.001). The 9 characteristic parameters were used as the input of ANN, the bispectral index (BIS) was used as the reference output, and the method was evaluated by the data of 8 patients during general anesthesia. The accuracy of the method in the classification of the four anesthesia levels of the test set in the 7:3 set-out method was 85.98%, and the correlation coefficient with the BIS was 0.977 0. The results show that this method can better distinguish four different anesthesia levels and has broad application prospects for monitoring the depth of anesthesia.
ObjectiveTo establish and validate the diagnostic model of ferroptosis genes for acute myocardial infarction (AMI) based on bioinformatics. MethodsFive AMI gene expression data were obtained from Gene Expression Omnibus (GEO), namely GSE66360, GSE48060, GSE60993, GSE83500, GSE34198. Among them, GSE66360 was used as the training set to perform differential analysis, and intersection of differential genes and ferroptosis genes was taken to obtain differentially expressed ferroptosis genes in AMI. GO and KEGG enrichment analysis was performed using Metascape website. Subsequently, random forest (RF) algorithm was used to screen out key genes with high classification performance according to the Keeny coefficient score, and artificial neural network (ANN) diagnostic model of AMI ferroptosis feature gene was constructed by model group GSE83500. The area under the receiver operating characteristic curve (AUC) of 10-fold cross-validation was used to evaluate the performance and generalization ability of the model, and 3 external independent datasets were used to verify the diagnostic performance of this model. The single sample gene setenrichment analysis was used to explore the difference in immune cell infiltration between infarcted myocardium and normal myocardium after AMI. In addition, correlation analysis between immune cells and key genes was also conducted. Finally, potential drugs that would prevent and treat AMI by regulating ferroptosis were screened out from the Coremin Medical platform. ResultsA total of 16 differentially expressed ferroptosis genes were obtained in the training set, GO enrichment analysis showed that they mainly participated in biological functions such as cellular response to biological stimuli and chemical stress, regulation of interleukin 17, etc. KEGG enrichment analysis showed that these genes were significantly enriched in NOD-like receptor signaling pathway, programmed cell necrosis, Leishmaniasis and other pathways. Four genes with good classification performance were screened out using RF algorithm, namely EPAS1, SLC7A5, FTH1, and ZFP36. The results of 10-fold cross-validation showed that the minimum AUC value was 0.746, the maximum value was 0.906, and the average value was 0.805. The AUC of the ANN model was 0.859, and the AUC values of the three independent validation sets were 0.763 (GSE48060), 0.673 (GSE60993), 0.698 (GSE34198). Immune cell infiltration found that macrophages, mast cells and monocytes were significantly active after AMI. Correlation analysis found that there were positive correlations between 4 key genes and activated dendritic cells, eosinophils and γδT cells. A total of 20 potential western medicines were predicted which could prevent and treat AMI by regulating ferroptosis, and the predicted potential Chinese medicine was mainly heat-clearing and detoxifying and blood-activating and removing blood stasis drugs. ConclusionThe identified AMI ferroptosis genes by bioinformatics method have certain diagnostic significance, which provides a reference for disease diagnosis and treatment.
ObjectiveThe aim of this study was to investigate the value of Artificial Neural Networks (ANNs) in predicting the occurrence of Venous Thromboembolism (VTE) in patients with Obstructive Sleep Apnea (OSA), and to compare it with traditional Logistic regression models to assess its predictive efficacy, providing theoretical basis for the prediction of VTE risk in OSA patients. MethodsA retrospective analysis was conducted on patients diagnosed with OSA and hospitalized in the Department of Respiratory and Critical Care Medicine, Second Affiliated Hospital of Kunming Medical University, from January 2018 to August 2023. Patients were divided into OSA combined with VTE group (n=128) and pure OSA control group (n=680). The dataset was randomly divided into a training set (n=646) and an independent validation set (n=162). The Synthetic Minority Oversampling Technique (SMOTE) was employed to address the issue of data imbalance. Artificial Neural Networks and Logistic regression models were then built on training sets with and without SMOTE. Finally, the performance of each model was evaluated using accuracy, sensitivity, specificity, Youden's index, and Area Under the Receiver Operating Characteristic Curve (AUC). Results When oversampling was conducted using SMOTE on the training set, both the Artificial Neural Network and Logistic regression models showed improved AUC. The Artificial Neural Network model with SMOTE performed the best with an AUC value of 0.935 (95% CI: 0.898-0.961), achieving an accuracy of 90.15%, specificity of 87.32%, sensitivity of 93.44%, and Youden's index of 0.808 at the optimal cutoff point. The Logistic regression model with SMOTE yielded an AUC value of 0.817 (95% CI: 0.765-0.861), with an accuracy of 77.27%, specificity of 83.80%, sensitivity of 69.67%, and Youden's index of 0.535. The difference in AUC between the Artificial Neural Network model and Logistic regression model was statistically significant after employing SMOTE (P<0.05). Conclusions The Artificial Neural Network model demonstrates high effectiveness in predicting VTE formation in OSA patients, particularly with the further improvement in predictive performance when utilizing SMOTE oversampling technique, rendering it more accurate and stable compared to the traditional Logistic regression model.