Rolling bearings are an important part of rotating machinery, and are of great significance for fault diagnosis and life monitoring of rolling bearings. Analyzing fault signals, extracting effective degradation information and establishing corresponding models are the premise of residual life prediction of rolling bearings. In this paper, first, the time-domain features were extracted to form the eigenvector of the vibration signal, and then the index representing the bearing degradation was found. It was found that the time-domain index could effectively describe the degradation information of the bearing, and the multi-dimensional time-domain characteristic information could effectively describe the attenuation trend of the vibration signal of the rolling bearing. On this basis, appropriate feature vectors were selected to describe the degradation characteristics of bearings. Aiming at the problems of large amounts of data, large amounts of information redundancy and unclear performance index of multi-dimensional feature vectors, the dimensionality of multi-dimensional feature vectors was reduced with principal component analysis, thus, simplifying the multi-dimensional feature vectors and reducing the information redundancy. Finally, in view of the support vector machine (SVM)’s needs to determine kernel function parameters and penalty factors, the squirrel optimization algorithm (SOA) was used to adaptively select parameters and establish the state-life evaluation model of rolling bearings. In addition, mean absolute error (MAE) and root mean squared error (RMSE) were used to comprehensively evaluate SOA. The results showed that the SOA reduced the errors by 5.1% and 13.6%, respectively, compared with a genetic algorithm (GA). Compared with particle swarm optimization (PSO), the error of SOA was reduced by 7.6% and 15.9%, respectively. It showed that SOA-SVM effectively improved the adaptability and regression performance of SVM, thus, significantly improving the prediction accuracy.

Purpose: To create and evaluate the accuracy of an artificial intelligence platform capable of using only retinal fundus images to predict both an individual overall 10 year Cardiovascular Disease (CVD) risk and the relative contribution of the component risk factors that comprise this risk (CVD-AI). Methods: The UK Biobank and the US-based AREDS 1 datasets were obtained and used for this study. The UK Biobank data was used for training, validation and testing, while the AREDS 1 dataset was used as an external testing dataset. Overall, we used 110,272 fundus images from 55,118 patient visits. A series of models were trained to predict the risk of CVD against available labels in the UK Biobank dataset. Results: In both the UK Biobank testing dataset and the external validation dataset (AREDS 1), the 10-year CV risk scores generated by CVD-AI were significantly higher for patients who had suffered an actual CVD event when compared to patients who did not experience a CVD event. In the UK Biobank dataset the median 10-year CVD risk for those individuals who experienced a CVD was higher than those who did not (4.9% [ICR 2.9-8%] v 2.3% [IQR 4.3- 1.3%] P<0.01.]. Similar results were observed in the AREDS 1 dataset The median 10-year CVD risk for those individuals who experienced a CVD event was higher than those who did not (6.2% [ICR 3.2%-12.9%] v 2.2% [IQR 3.9- 1.3%] P<0.01 Conclusion: Retinal photography is inexpensive and as fully automated, inexpensive camera systems are now widely available, minimal training is required to acquire them. As such, AI enabled retinal image-based CVD risk algorithms like CVD-AI will make cardiovascular health screening more affordable and accessible. CVD-AI unique ability to assess the relative contribution of the components that comprise an individual overall risk could inform personalized treatment decisions based on the specific needs of an individual, thereby increasing the likelihood of positive health outcomes.

No abstract

Deep learning (DL) models have shown promise in detecting chronic kidney disease (CKD) from fundus photographs. However, previous studies have utilized a serum creatinine-only estimated glomerular rate (eGFR) equation to determine CKD despite the existence of more accurate methods. In this study, we used the UK Biobank as a test and validation dataset to demonstrate an incremental and statistically significant improvement in model performance for predicting CKD when using a creatinine and cystatin C eGFR equation over a creatinine-only equation. Attempts to directly compare our results with the results from existing DL models is complicated due to significant differences in the composition of the dataset, particularly in the incidence rate of confounding risk factors. We hypothesize that existing eGFR equations' limitations in accurately identifying CKD and the paucity of retinal features uniquely indicative of CKD may contribute to the observed differences in model performance, highlighting the need for more principled research to quantify the effects of dataset distribution on DL models' ability to predict CKD from fundus photographs.