Morbidity After Symptomatic Hemorrhage of Cerebral Cavernous Malformation: A Nomogram Approach to Risk Assessment.

BACKGROUND AND PURPOSE: Symptomatic hemorrhage contributes to an increased risk of repeated bleeding and morbidity in cerebral cavernous malformation (CCM). A better understanding of morbidity after CCM hemorrhage would be helpful to identify patients of higher risk for unfavorable outcome and tailor individualized management. METHODS: We identified 282 consecutive patients who referred to our institute from 2014 to 2018 for CCM with symptomatic hemorrhage and had an untreated follow-up period over 6 months after the first hemorrhage. The morbidity after hemorrhage was described in CCM of different features. Nomogram to predict morbidity was formulated based on the multivariable model of risk factors. The predictive accuracy and discriminative ability of nomogram were determined with concordance index (C-index) and calibration curve, and further validated in an independent CCM cohort of a prospective multicenter study from 2019 to 2020. RESULTS: The overall morbidity of CCM was 26.2% after a mean follow-up of 1.9 years (range 0.5-3.5 years) since the first hemorrhage. The morbidity during untreated follow-up was associated with hemorrhage ictus (adjusted odds ratio per ictus increase, 4.17 [95% CI, 1.86-9.33]), modified Rankin Scale score at initial hemorrhage (adjusted odds ratio per point increase, 2.57 [95% CI, 1.82-3.63]), brainstem location (adjusted odds ratio, 2.93 [95% CI, 1.28-6.68]), and associated developmental venous anomaly (adjusted odds ratio, 2.21 [95% CI, 1.01-4.83]). Subgroup analysis revealed similar findings in brainstem and non-brainstem CCM. Nomogram was contracted based on these features. The calibration curve showed good agreement between nomogram prediction and actual observation. The C-index of nomogram predicting morbidity was 0.83 (95% CI, 0.77-0.88). In validation cohort, the nomogram maintained the discriminative ability (C-index, 0.87 [95% CI, 0.78-0.96]). CONCLUSIONS: Multiple symptomatic hemorrhages, initial neurological function after hemorrhage, brainstem location, and associated developmental venous anomaly were associated with morbidity of CCM hemorrhage. The nomogram represented a practical approach to provide individualized risk assessment for CCM patients. Registration: URL: https://www.clinicaltrials.gov. Unique identifier: NCT04076449.

Cost-based multi-parameter logistics routing path optimization algorithm.

The traditional path optimization problem is to consider the shortest path of the vehicle, but the shortest path does not effectively reduce the logistics cost. On the contrary, in the case of one-sided pursuit of the shortest path, it may cause some negative effects. This paper constructs a more realistic path optimization model on the path of traditional logistics distribution, and designs a path model based on simulated annealing algorithm which taking fuel consumption, cost, road gradient and condition of vehicle into account. The algorithm model of load capacity and other problems is used to verify the algorithm of the model through a simulation case of multiple distribution points. The experimental results show that the path optimization strategy considering the gradient of the road reduces the cost of the vehicle path, indicating the correctness of considering the vehicle load and road gradient factors in logistics transportation.

Prediction of crime tendency of high-risk personnel using C5.0 decision tree empowered by particle swarm optimization.

The research on the big data in the security and protection industry has been increasingly recognized as the hotspot in case of the rapid development of the big data. This paper mainly focuses on addressing the problem that predicts the criminal tendency of the high-risk personnel based on the recorded behavior data of the high-risk personnel. Therefore, we propose a novel predictive model that is the crime tendency of high-risk personnel using C5.0 based on particle swarm optimization. In this model, the C5.0 decision tree algorithm is first used as a classifier, in which repeated tenfold cross-validation is used and then continuously tuned according to the custom fitness function based on particle swarm optimization. In addition, the classification accuracy, the reduced number of feature subset, specificity and sensitivity under different algorithms are compared. Finally, the proposed model has higher accuracy through the optimal value of the particle position, the error rate of the cost under different iterations and the trend and the concavity and convexity of ROC curve. The experimental results show that the proposed model has a good effect on the predictive classification, which may provide guidance for predicting crime tendency of high-risk personnel.