Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Stroke is one of the most common diseases that cause mortality. Detecting the risk of stroke for individuals is critical yet challenging because of a large number of risk factors for stroke. Objective: To address the limitation of ineffective feature selection in existing research on stroke risk detection, we propose a new feature selection method called Weighting- and Ranking-based Hybrid Feature Selection (WRHFS) to select important risk factors for detecting ischemic stroke. Methods: WRHFS integrates the strengths of various filter algorithms by following the principle of a wrapper approach. We employed a variety of filter-based feature selection models as the candidate set, including standard deviation, Pearson correlation coefficient, Fisher score, information gain, RELIEF, and Chi-squared, and used sensitivity, specificity, accuracy, and Youden’s index as performance metrics to evaluate the proposed method. Results: This study chooses 792 samples from the electronic records of 13,421 patients in a community hospital. Each sample includes 28 features (the 24 blood test features and the 4 demographic features). The results of evaluation show that the proposed method selects 9 important features out of the original 28 features and significantly outperforms baseline methods. Their cumulative contribution was 0.51. The WRHFS method achieves sensitivity of 82.7% (329/398), specificity of 80.4% (317/394), the classification accuracy of 81.5% (645/792), and Youden’s Index of 0.63 using only the top 9 features. We also present a chart for visualizing the risk of having ischemic strokes. Conclusions: This research proposes, develops, and evaluates a new feature selection method for identifying most important features for building effective and parsimonious models for stroke risk detection. The findings of this research provide several novel research contributions and practical implications.

In this paper, we propose time series prediction methods for depth-averaged current velocities (DACVs) of underwater gliders. Based on historical DACV data, these methods can predict the DACVs of future profiles with good performance. Regarding DACVs as time series, we use backpropagation neural network and least squares support vector machine (LSSVM) methods to predict the DACVs. To obtain better prediction performance, the features of DACVs are considered, and we use empirical mode decomposition (EMD) to decompose the time series into several sub-series. Then, the two methods are reused to predict each sub-series, and the results of all the sub-series with each method are added. Based on the real-time DACVs obtained from the simulation environment and the DACVs obtained from sea trials, we test and verify the four methods. The results demonstrate that all the methods exhibit a good prediction performance for conditions in which ocean currents are relatively regular; whereas in other cases, EMD-LSSVM shows inherent robustness and superiority compared with the other three methods.

In this paper, the data feature of depth-averaged current velocities (DACVs) derived from underwater gliders is analyzed for the first time. Two features of DACVs have been proposed: one is the complex ingredients and small samples, and the other is the stationarity that occurs as the length of a DACV sequence increases. With these features in mind, a set of methods combining statistical analysis and machine learning are proposed to realize the prediction of DACVs. Four groups of DACV data of different gliders from sea trials in the South China Sea are used to verify the prediction method. Based on three general error criteria, the prediction performance of the proposed model is demonstrated. The persistence method is used as a comparison model. The results show that the prediction methods proposed in this paper are effective.

Inkjet printing is regarded as a new generation of green intelligent manufacturing technology. However, precise control of the state of droplets in inkjet printing is critical and costly. In this study, a voltage-driven signal-based ink droplet state prediction model suitable for a small sample data environment is proposed. Seven public ink datasets and self-collected experimental data are grouped together, and the decision tree, AdaBoost.R2, and two-stage TrAdaBoost.R2 algorithms are used, respectively. The two-stage TrAdaBoost.R2 model shows excellent prediction performance in the case of insufficient data by transferring knowledge of different inks. In particular, it achieves mean square errors of 0.00219 for droplet volume predictions and 0.0645 for velocity predictions. Similarly, the mean absolute percentage errors are 0.91% for droplet volume and 1.37% for velocity. Furthermore, the two-stage TrAdaBoost.R2 model demonstrates strong generalization ability and adaptability to maintain high prediction accuracy under different conditions. The results indicate that the two-stage TrAdaBoost.R2 effectively mitigates the prediction performance issues caused by data scarcity, paving the way for technological advancements in the field of inkjet printing.