Search

For systemic risk, the impact of the financial network's characteristics remains imperfectly understood at best, even if the view that network structure is closely related to systemic risk has become a broad consensus. By choosing S&P 500 constituents as the research sample, we investigate the structural characteristics of the Engle-Granger networks and explore the impact of network centrality on one-quarter-ahead systemic risk. We find that a firm's network centrality is positively related to both dimensions of its systemic risk (i.e. the firm's vulnerability to, and contribution to, system-wide downturns). The results remain robust after we consider the potential endogeneity and various sensitivity checks. An examination of potential channels reveals that centrally located firms in the network have a high extent of co-movement with the market, and are likely to trigger systemic market failures caused by stock price crashes in clusters once they fall into a downturn. We further show that the positive relation between network centrality and future systemic risk is more salient for financial firms and more pronounced during recessions. [ABSTRACT FROM AUTHOR]

Objective To explore the clinical application value of deep learning-based artificial intelligence (AI) in the detection and related quantitative measurement of pulmonary nodules. Methods We collected 250 cases of chest CT scan and compared the misdiagnosis rate, missed diagnosis rate, sensitivity, positive predictive value and average diagnosis time of pulmonary nodules among group A (hospitalized), group B (AI) and group C (hospitalized+AI). Meanwhile, AI quantization parameters of solid nodules and ground glass nodules (GGN) were compared, and ROC curve analysis was performed for the parameters with statistical difference. Results A total of 2230 nodules were identified. The misdiagnosis rate of group B was significantly higher than those of group A and C, and the positive predictive value was significantly lower than those of group A and C (P < 0.05). The rate of missed diagnosis in group A was significantly higher than those in group B and C, and the sensitivity was significantly lower than those in group B and C (P < 0.05). There were statistically significant differences in the long diameter, maximum area, volume, minimum CT value and malignant probability between solid benign and malignant nodules (P < 0.05). The indexes of area under the ROC curve (AUC) greater than 0.7 were: long diameter, maximum area, volume and malignant probability. There were statistically significant differences in the length, maximum area, volume, average CT value, maximum CT value and malignant probability between GGN benign and malignant nodules (P < 0.05). ROC curve analysis of all parameters showed that AUC was greater than 0.7. Conclusion AI-assisted film reading could significantly improve work efficiency and sensitivity of pulmonary nodules detection and reduce the rates of misdiagnosis and missed diagnosis. Meanwhile, it has certain reference value for the prediction of benign and malignant pulmonary nodules.