Background Frontal lobe epilepsy (FLE) is the second most common type of focal epilepsy, however, imaging studies of FLE have been far less than Temporal lobe epilepsy (TLE) and the structural findings were not consistent in previous literature. Object Investigate the changes in cortical thickness in patients with FLE and the alteration of the structural covariance networks (SCNs) of cortical thickness with graph-theory. Method Thirty patients with FLE (18 males/12 females; 28.33 ± 11.81 years) and 27 demographically matched controls (15 males/12 females; 29.22 ± 9.73 years) were included in this study with high-resolution structural brain MRI scans. The cortical thickness was calculated, and structural covariance network (SCN) of cortical thickness were reconstructed using 68 × 68 matrix and analyzed with graph-theory approach. Result Cortical thickness was not significantly different between two groups, but path length and node betweenness were significantly increased in patients with FLE, and the regional network alterations were significantly changed in right precentral gyrus and right temporal pole (FDR corrected, p < 0.05). Comparing to HC group, network hubs were decreased and shifted away from frontal lobe. Conclusion The topological properties of cortical thickness covariance network were significantly altered in patients with FLE, even without obvious surface-based morphological damage. Graph-theory based SCN analysis may provide sensitive neuroanatomical biomarkers for FLE.

To develop and validate a machine learning model for the prediction of adverse outcomes in hospitalized patients with COVID-19. We included 424 patients with non-severe COVID-19 on admission from January 17, 2020, to February 17, 2020, in the primary cohort of this retrospective multicenter study. The extent of lung involvement was quantified on chest CT images by a deep learning–based framework. The composite endpoint was the occurrence of severe or critical COVID-19 or death during hospitalization. The optimal machine learning classifier and feature subset were selected for model construction. The performance was further tested in an external validation cohort consisting of 98 patients. There was no significant difference in the prevalence of adverse outcomes (8.7% vs. 8.2%, p = 0.858) between the primary and validation cohorts. The machine learning method extreme gradient boosting (XGBoost) and optimal feature subset including lactic dehydrogenase (LDH), presence of comorbidity, CT lesion ratio (lesion%), and hypersensitive cardiac troponin I (hs-cTnI) were selected for model construction. The XGBoost classifier based on the optimal feature subset performed well for the prediction of developing adverse outcomes in the primary and validation cohorts, with AUCs of 0.959 (95% confidence interval [CI]: 0.936–0.976) and 0.953 (95% CI: 0.891–0.986), respectively. Furthermore, the XGBoost classifier also showed clinical usefulness. We presented a machine learning model that could be effectively used as a predictor of adverse outcomes in hospitalized patients with COVID-19, opening up the possibility for patient stratification and treatment allocation. • Developing an individually prognostic model for COVID-19 has the potential to allow efficient allocation of medical resources. • We proposed a deep learning–based framework for accurate lung involvement quantification on chest CT images. • Machine learning based on clinical and CT variables can facilitate the prediction of adverse outcomes of COVID-19.

The outbreak of coronavirus disease 2019 (COVID-19) has rapidly spread to become a worldwide emergency. Early identification of patients at risk of progression may facilitate more individually aligned treatment plans and optimized utilization of medical resource. Here we conducted a multicenter retrospective study involving patients with moderate COVID-19 pneumonia to investigate the utility of chest computed tomography (CT) and clinical characteristics to risk-stratify the patients. Our results show that CT severity score is associated with inflammatory levels and that older age, higher neutrophil-to-lymphocyte ratio (NLR), and CT severity score on admission are independent risk factors for short-term progression. The nomogram based on these risk factors shows good calibration and discrimination in the derivation and validation cohorts. These findings have implications for predicting the progression risk of COVID-19 pneumonia patients at the time of admission. CT examination may help risk-stratification and guide the timing of admission.

AbstractObjectiveTo determine the predictive value of CT and clinical characteristics for short-term disease progression in patients with 2019 novel coronavirus pneumonia (NCP).Materials and Methods224 patients with confirmed 2019 novel coronavirus (COVID-19) infection outside Wuhan who had chest CT examinations were retrospectively screened. Clinical data were obtained from electronic medical records. CT images were reviewed and scored for lesion distribution, lobe and segment involvement, ground-glass opacities, consolidation, and interstitial thickening. All included patients with moderate NCP were observed for at least 14 days from admission to determine whether they exacerbated to severe NCP (progressive group) or not (stable group). CT and clinical characteristics between the two groups were compared, and multivariate logistic regression and sensitivity analyses were performed to identify the risk factors for developing severe NCP.ResultsA total of 141 patients with moderate NCP were included, of which 15 (10.6%) patients developed severe NCP during hospitalization and assigned to the progressive group. Multivariate logistic regression analysis showed that higher neutrophil-to-lymphocyte ratio (NLR) (odds ratio [OR] and 95% confidence interval [CI], 1.26 [1.04-1.53];P= 0.018) and CT severity score (OR and 95% CI, 1.25 [1.08-1.46];P= 0.004) on admission were independent predictors for progression to severe NCP, and sensitivity analysis confirmed the consistent results in nonimported patients but not in imported patients. However, no significant difference in lung involvement was found on CT between imported and nonimported patients (allP> 0.05). Patients who were admitted more than 4 days from symptom onset tended to have more severe lung involvement. Spearman correlation analysis showed the close association between CT severity score and inflammatory indexes (r= 0.17∼0.47, allP< 0.05).ConclusionCT severity score was associated with inflammatory levels and higher NLR and CT severity score on admission were independent risk factors for short-term progression in patients with NCP outside Wuhan. Furthermore, early admission and surveillance by CT should be recommended to improve clinical outcomes.

Background: Early identification of patients at risk of developing severe illness and timely interventions to prevent deterioration have the potential to improve outcomes in coronavirus disease 2019 (COVID-19) patients. Herein, we conducted this study to develop and validate a machine learning model for the early prediction of progression risk in COVID-19 patients. Methods: 424 patients with non-severe COVID-19 who underwent chest computed tomography (CT) scan on admission were included in the primary cohort of this retrospective multicentre study. The candidate feature set included 43 baseline clinical variables and CT quantitative measurements determined using deep learning-based segmentation and unsupervised texture clustering. The endpoint was the development of severe COVID-19. The optimal machine learning classifier and feature subset were selected for model construction. The performance of the prediction model was evaluated and tested in an external validation cohort consisting of 98 patients. The performance and clinical usefulness of the model were compared with the CALL score. Findings: There were no significant differences in the clinical characteristics between the primary and validation cohorts, including the prevalence of severe illness (8.7% vs. 8.2%, P = 0.858). The machine learning method extreme gradient boosting (XGBoost) and optimal feature subset including lactic dehydrogenase (LDH), comorbidities, CT lesion ratio (lesion%), and hypersensitive cardiac troponin I (hs-cTnI) were selected for model construction. COVID-19 patients who developed severe illness could be identified by our simplified XGBoost classifier with AUC of 0.956 and accuracy of 0.892 in the primary cohort. The test on the validation cohort showed comparable performance with that of the primary cohort. Compared with the CALL score, the simplified XGBoost classifier showed improved performance (AUC: 0.954 vs. 0.831, P < 0.001) and was clinically useful. Interpretation: A machine learning classifier incorporating LDH, comorbidities, CT lesion%, and hs-cTnI could be conveniently and effectively used as an early predictor of progression risk in patients with COVID-19, opening up the possibility for patient stratification and treatment allocation.Funding Statement: This study was supported by National Natural Science Foundation of China (81771827, 81471715 to Rong).Declaration of Interests: The authors declare no competing financial interests.Ethics Approval Statement: The Institutional Review Board of Third Xiangya Hospital approved our study and waived the informed consent of patients for the retrospective nature of this study.