RESUMO
Overall survival (OS) in cancer is crucial for cancer treatment. Many machine learning methods have been applied to predict OS, but there are still the challenges of dealing with multiview data and overfitting. To overcome these problems, we propose a multiview deep forest (MVDF) in this paper. MVDF can learn the features of each view and fuse them with integrated learning and multiple kernel learning. Then, a gradient boost forest based on the information bottleneck theory is proposed to reduce redundant information and avoid overfitting. In addition, a pruning strategy for a cascaded forest is used to limit the impact of outlier data. Comprehensive experiments have been carried out on a data set from West China Hospital of Sichuan University and two public data sets. Results have demonstrated that our method outperforms the compared methods in predicting overall survival.
Assuntos
Neoplasias , Humanos , Aprendizado de Máquina , China/epidemiologiaRESUMO
OBJECTRIVE: To compare the differences in risk factors for low birth weight (LBW) between Han and Uygur full-term infants and to provide a basis for the prevention of LBW in newborn infants. METHODS: Eighty-seven full-term LBW infants (38 Hans and 49 Uygurs) between March 2013 and June 2014 were selected as the case group, and 186 full-term normal birth weight infants (92 Hans and 94 Uygurs) were selected as the control group. A questionnaire survey was performed to investigate the related factors for LBW. Multivariate logistic regression analysis was carried out to determine the risk factors for LBW. RESULTS: The birth weights in Uyghur LBW infants were lower than in Han ones (P<0.05). Multivariate logistic regression analysis showed that drinking (OR=2.472, P=0.015) and smoking (OR=2.323, P=0.007) by the father, pregnancy complications (OR=14.377, P<0.001), and times of pregnancy (OR=2.995, P=0.001) were the risk factors for LBW in Han infants, while drinking by the father (OR=1.968, P=0.007), times of pregnancy (OR=1.953, P=0.005), pregnancy complications (OR=10.283, P=0.002), and poor indoor environment (OR=1.367, P=0.027) were the risk factors for LBW in Uyghur infants. CONCLUSIONS: There are differences in physical growth between Han and Uygur LBW infants. Han and Uygur infants share the same traditional risk factors for LBW, such as father's harmful behaviors like drinking, times of pregnancy, and pregnancy complications, however, the indoor environment also plays a role in the occurrence of LBW in Uygur infants.