Prediction of stock price movement using an improved NSGA-II-RF algorithm with a three-stage feature engineering process.

Prediction of stock price has been a hot topic in artificial intelligence field. Computational intelligent methods such as machine learning or deep learning are explored in the prediction system in recent years. However, making accurate predictions of stock price direction is still a big challenge because stock prices are affected by nonlinear, nonstationary, and high dimensional features. In previous works, feature engineering was overlooked. How to select the optimal feature sets that affect stock price is a prominent solution. Hence, our motivation for this article is to propose an improved many-objective optimization algorithm integrating random forest (I-NSGA-II-RF) algorithm with a three-stage feature engineering process in order to decrease the computational complexity and improve the accuracy of prediction system. Maximizing accuracy and minimizing the optimal solution set are the optimization directions of the model in this study. The integrated information initialization population of two filtered feature selection methods is used to optimize the I-NSGA-II algorithm, using multiple chromosome hybrid coding to synchronously select features and optimize model parameters. Finally, the selected feature subset and parameters are input to the RF for training, prediction, and iterative optimization. Experimental results show that the I-NSGA-II-RF algorithm has the highest average accuracy, the smallest optimal solution set, and the shortest running time compared to the unmodified multi-objective feature selection algorithm and the single target feature selection algorithm. Compared to the deep learning model, this model has interpretability, higher accuracy, and less running time.

The factors of fiducial marker motions and individual margin assessment in postoperative breast cancer radiotherapy.

Background: As a surrogate for the breast tumor bed, individual fiducial markers frequently move during radiotherapy. This study aimed to classify the motions and calculate the individualized target margin. Methods: The mammary basal diameters (D) and heights (H) were measured to represent breast sizes for 15 patients after breast-conserving surgery. The clinical target volume (CTV) was divided into 4 quadrants by a coordinate system with the center of mass of the tumor bed as the origin. The lateral, anteroposterior, and craniocaudal motions of markers were calculated (MLR, MAP, MSI) based on the difference of the setup errors between the spine matching and the fiducial markers matching. The distances between markers and the innermost, foremost, and uppermost borders of CTV (DSLR, DSAP, DSSI) were recorded. Results: In the first quadrant, MAP was strongly correlated with D×H (r>0.80) when D×H <99.89 cm2. Both MLR and MAP were positively linearly related to DSLR, DSAP, DSSI (r>0.85, R2>0.75). MSI was also positively linearly correlated with DSAP and DSLR (r>0.90, R2>0.80). In the fourth quadrant with D×H <90.71 cm2, only MLR and DSLR showed a linear positive correlation (r>0.90, R2>0.75), whereas the others showed linear negative correlations (r>-0.90, R2>0.80). The planning target volume (PTV) margin varied significantly between the first and fourth quadrant (P<0.05), and the largest margin was 12.4 mm in the craniocaudal direction of the first quadrant with D×H ≥99.89 cm2. Conclusions: Fiducial motion is susceptible to breast size and fiducial position, and the individualized PTV margins should take the above factors into account.

Revaluation of clopidogrel: let the data speak for themselves.

Clopidogrel was believed to be superior to aspirin by the well-known CAPRIE trial. However, no other large clinical trials demonstrated the same results, but all focused on the combination use of clopidogrel with aspirin, and combination therapy in CREDO was called the "Emperor's New Clothes". However, no one overturned the results of these clinical trials by quantitatively analyzing them. We reviewed ten large-scale clinical trials about clopidogrel. On the basis of results of CAPRIE, CREDO and CHARISMA trials, we re-estimated their minimal sample sizes and their powers by three well-established statistical methodologies. From the results of CAPRIE, we inferred that the minimal sample size should be 85 086 or 84 968 but its power was only 30.70%. A huge gap existed. The same was also true of CREDO and CHARISMA trials. Moreover, in CAPRIE trial, 0 was included in the 95% confidence interval and 1 was included in the 95% confidence interval for the relative risk. There were some paradoxical data in CAPRIE trial. We are led to conclude that the results in CAPRIE, CREDO, and from the subgroup analysis in CHARISMA trials were questionable. These results failed to demonstrate that clopidogrel was superior to aspirin or that clopidogrel used in combination with aspirin was better than aspirin alone. The cost-effectiveness analyses by some previous studies were not reliable.