Performance of Machine Learning Algorithms for Predicting Adverse Outcomes in Community-Acquired Pneumonia.

Background: The ability to assess adverse outcomes in patients with community-acquired pneumonia (CAP) could improve clinical decision-making to enhance clinical practice, but the studies remain insufficient, and similarly, few machine learning (ML) models have been developed. Objective: We aimed to explore the effectiveness of predicting adverse outcomes in CAP through ML models. Methods: A total of 2,302 adults with CAP who were prospectively recruited between January 2012 and March 2015 across three cities in South America were extracted from DryadData. After a 70:30 training set: test set split of the data, nine ML algorithms were executed and their diagnostic accuracy was measured mainly by the area under the curve (AUC). The nine ML algorithms included decision trees, random forests, extreme gradient boosting (XGBoost), support vector machines, Naïve Bayes, K-nearest neighbors, ridge regression, logistic regression without regularization, and neural networks. The adverse outcomes included hospital admission, mortality, ICU admission, and one-year post-enrollment status. Results: The XGBoost algorithm had the best performance in predicting hospital admission. Its AUC reached 0.921, and accuracy, precision, recall, and F1-score were better than those of other models. In the prediction of ICU admission, a model trained with the XGBoost algorithm showed the best performance with AUC 0.801. XGBoost algorithm also did a good job at predicting one-year post-enrollment status. The results of AUC, accuracy, precision, recall, and F1-score indicated the algorithm had high accuracy and precision. In addition, the best performance was seen by the neural network algorithm when predicting death (AUC 0.831). Conclusions: ML algorithms, particularly the XGBoost algorithm, were feasible and effective in predicting adverse outcomes of CAP patients. The ML models based on available common clinical features had great potential to guide individual treatment and subsequent clinical decisions.

Inflammation/Coagulopathy/Immunology Responsive Index Predicts Poor COVID-19 Prognosis.

In the early stage of coronavirus disease 2019 (COVID-19), most cases are identified as mild or moderate illnesses. Approximately 20% of hospitalised patients become severe or critical at the middle or late stage of the disease. The predictors and risk factors for prognosis in those with mild or moderate disease remain to be determined. Of 694 patients with COVID-19, 231 patients with mild or moderate disease, who were hospitalised at 10 hospitals in Wenzhou and nearby counties in China, were enrolled in this retrospective study from 17 January to 20 March 2020. The outcomes of these patients included progression from mild/moderate illness to severe or critical conditions. Among the 231 patients, 49 (21.2%) had a poor prognosis in the hospital. Multivariate logistic regression analysis showed that higher inflammation/coagulopathy/immunology responsive index (ICIRI=[c-reactive protein × fibrinogen × D-dimer]/CD8 T cell count) on admission (OR=345.151, 95% CI=23.014-5176.318) was associated with increased odds ratios for poor prognosis. The area under the receiver operating characteristic curve for ICIRI predicting severe and critical condition progression was 0.65 (95% CI=0.519-0.782) and 0.80 (95% CI=0.647-0.954), with cut-off values of 870.83 and 535.44, respectively. Conversely, age, sex, comorbidity, neutrophil/lymphocyte ratio, CD8 T cell count, and c-reactive protein, fibrinogen, and D-dimer levels alone at admission were not good predictors of poor prognosis in patients with mild or moderate COVID-19. At admission, a novel index, ICIRI, tends to be the most promising predictor of COVID-19 progression from mild or moderate illness to severe or critical conditions.

Periodontal disease increases the host susceptibility to COVID-19 and its severity: a Mendelian randomization study.

BACKGROUND: Emerging evidence shows that periodontal disease (PD) may increase the risk of Coronavirus disease 2019 (COVID-19) complications. Here, we undertook a two-sample Mendelian randomization (MR) study, and investigated for the first time the possible causal impact of PD on host susceptibility to COVID-19 and its severity. METHODS: Summary statistics of COVID-19 susceptibility and severity were retrieved from the COVID-19 Host Genetics Initiative and used as outcomes. Single nucleotide polymorphisms associated with PD in Genome-wide association study were included as exposure. Inverse-variance weighted (IVW) method was employed as the main approach to analyze the causal relationships between PD and COVID-19. Three additional methods were adopted, allowing the existence of horizontal pleiotropy, including MR-Egger regression, weighted median and weighted mode methods. Comprehensive sensitivity analyses were also conducted for estimating the robustness of the identified associations. RESULTS: The MR estimates showed that PD was significantly associated with significantly higher susceptibility to COVID-19 using IVW (OR = 1.024, P = 0.017, 95% CI 1.004-1.045) and weighted median method (OR = 1.029, P = 0.024, 95% CI 1.003-1.055). Furthermore, it revealed that PD was significantly linked to COVID-19 severity based on the comparison of hospitalization versus population controls (IVW, OR = 1.025, P = 0.039, 95% CI 1.001-1.049; weighted median, OR = 1.030, P = 0.027, 95% CI 1.003-1.058). No such association was observed in the cohort of highly severe cases confirmed versus those not hospitalized due to COVID-19. CONCLUSIONS: We provide evidence on the possible causality of PD accounting for the susceptibility and severity of COVID-19, highlighting the importance of oral/periodontal healthcare for general wellbeing during the pandemic and beyond.

Inflammation/coagulopathy/fibrinolysis: Dynamic indicators of COVID-19 progression in patients with moderate COVID-19 in Wenzhou, China.

BACKGROUND: The majority of the coronavirus disease 2019 (COVID-19) non-survivors meet the criteria for disseminated intravascular coagulation (DIC). Although timely monitoring of clotting hemorrhagic development during the natural course of COVID-19 is critical for understanding pathogenesis, diagnosis, and treatment of the disease, however, limited data are available on the dynamic processes of inflammation/coagulopathy/fibrinolysis (ICF). METHODS: We monitored the dynamic progression of ICF in patients with moderate COVID-19. Out of 694 COVID-19 inpatients from 10 hospitals in Wenzhou, China, we selected 293 adult patients without comorbidities. These patients were divided into different daily cohorts according to the COVID-19 onset-time. Furthermore, data of 223 COVID-19 patients with comorbidities and 22 critical cases were analyzed. Retrospective data were extracted from electronic medical records. RESULTS: The virus-induced damages to pre-hospitalization patients triggered two ICF fluctuations during the 14-day course of the disease. C-reactive protein (CRP), fibrinogen, and D-dimer levels increased and peaked at day 5 (D) 5 and D9 during the 1st and 2nd fluctuations, respectively. The ICF activities were higher during the 2nd fluctuation. Although 12-day medication returned high CRP concentrations to normal and blocked fibrinogen increase, the D-dimer levels remained high on days 17 ±â€¯2 and 23 ±â€¯2 days of the COVID-19 course. Notably, although the oxygenation index, prothrombin time and activated partial thromboplastin time were within the normal range in critical COVID-19 patients at administration, 86% of these patients had a D-dimer level > 500 µg/L. CONCLUSION: COVID-19 is linked with chronic DIC, which could be responsible for the progression of the disease. Understanding and monitoring ICF progression during COVID-19 can help clinicians in identifying the stage of the disease quickly and accurately and administering suitable treatment.

Identification of the RNase-binding site of SARS-CoV-2 RNA for anchor primer-PCR detection of viral loading in 306 COVID-19 patients.

The pandemic of coronavirus disease 2019 (COVID-19) urgently calls for more sensitive molecular diagnosis to improve sensitivity of current viral nuclear acid detection. We have developed an anchor primer (AP)-based assay to improve viral RNA stability by bioinformatics identification of RNase-binding site of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA and implementing AP dually targeting the N gene of SARS-CoV-2 RNA and RNase 1, 3, 6. The arbitrarily primed polymerase chain reaction (AP-PCR) improvement of viral RNA integrity was supported by (a) the AP increased resistance of the targeted gene (N gene) of SARS-CoV-2 RNA to RNase treatment; (b) the detection of SARS-CoV-2 RNA by AP-PCR with lower cycle threshold values (-2.7 cycles) compared to two commercially available assays; (c) improvement of the viral RNA stability of the ORF gene upon targeting of the N gene and RNase. Furthermore, the improved sensitivity by AP-PCR was demonstrated by detection of SARS-CoV-2 RNA in 70-80% of sputum, nasal, pharyngeal swabs and feces and 36% (4/11) of urine of the confirmed cases (n = 252), 7% convalescent cases (n = 54) and none of 300 negative cases. Lastly, AP-PCR analysis of 306 confirmed and convalescent cases revealed prolonged presence of viral loading for >20 days after the first positive diagnosis. Thus, the AP dually targeting SARS-CoV-2 RNA and RNase improves molecular detection by preserving SARS-CoV-2 RNA integrity and reveals the prolonged viral loading associated with older age and male gender in COVID-19 patients.

Potential Impact of Diabetes and Obesity on Alveolar Type 2 (AT2)-Lipofibroblast (LIF) Interactions After COVID-19 Infection.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a new emerging respiratory virus, caused evolving pneumonia outbreak around the world. In SARS-Cov-2 infected patients, diabetes mellitus (DM) and obesity are two metabolic diseases associated with higher severity of SARS-CoV-2 related complications, characterized by acute lung injury requiring assisted ventilation as well as fibrosis development in surviving patients. Different factors are potentially responsible for this exacerbated response to SARS-CoV-2 infection. In patients with DM, base-line increase in inflammation and oxidative stress represent preexisting risk factors for virus-induced damages. Such factors are also likely to be found in obese patients. In addition, it has been proposed that massive injury to the alveolar epithelial type 2 (AT2) cells, which express the SARS-CoV-2 receptor angiotensin-converting enzyme 2 (ACE2), leads to the activation of their stromal niches represented by the Lipofibroblasts (LIF). LIF are instrumental in maintaining the self-renewal of AT2 stem cells. LIF have been proposed to transdifferentiate into Myofibroblast (MYF) following injury to AT2 cells, thereby contributing to fibrosis. We hypothesized that LIF's activity could be impacted by DM or obesity in an age- and gender-dependent manner, rendering them more prone to transition toward the profibrotic MYF status in the context of severe COVID-19 pneumonia. Understanding the cumulative effects of DM and/or obesity in the context of SARS-CoV-2 infection at the cellular level will be crucial for efficient therapeutic solutions.

Metabolomics reveals sex-specific metabolic shifts and predicts the duration from positive to negative in non-severe COVID-19 patients during recovery process.

Metabolic profiling in COVID-19 patients has been associated with disease severity, but there is no report on sex-specific metabolic changes in discharged survivors. Herein we used an integrated approach of LC-MS-and GC-MS-based untargeted metabolomics to analyze plasma metabolic characteristics in men and women with non-severe COVID-19 at both acute period and 30 days after discharge. The results demonstrate that metabolic alterations in plasma of COVID-19 patients during the recovery and rehabilitation process were presented in a sex specific manner. Overall, the levels of most metabolites were increased in COVID-19 patients after the cure relative to acute period. The major plasma metabolic changes were identified including fatty acids in men and glycerophosphocholines and carbohydrates in women. In addition, we found that women had shorter length of hospitalization than men and metabolic characteristics may contribute to predict the duration from positive to negative in non-severe COVID-19 patients. Collectively, this study shed light on sex-specific metabolic shifts in non-severe COVID-19 patients during the recovery process, suggesting a sex bias in prognostic and therapeutic evaluations based on metabolic profiling.

Sex- and age-specific clinical and immunological features of coronavirus disease 2019.

To simultaneously determine clinical and immunological responses to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in young and old females and males, 681 coronavirus disease 2019 (COVID-19) patients and 369 normal controls (NCs) were analyzed based on age and sex classifications using multiple linear regression analysis. Compared to the age-matched NCs, both young and old male and female non-comorbid COVID-19 patients had lower lymphocyte counts and alanine aminotransferase (ALT) concentration, and only young male and female patients had lower neutrophil counts. Compared to young patients, both old males and females had significantly higher plasma ALT and AST concentrations. Compared to young and old females, age-matched males had higher plasma ALT and AST concentrations, but only young males had higher C-reactive protein (CRP) concentration. Compared to females, old males, but not young males, showed higher incidence of critical illness. Compared to young patients, old females had more leukocyte and neutrophil counts above the normal upper limit and B cell count below the normal lower limit (NLL), while old males had more lymphocyte and natural killer (NK) cell counts below the NLL. No sex or age associations with B cell and NK cell counts were observed. However, there were age-dependent decreases in CD8+ T-cell counts in both male and female COVID-19 patients. Age was negatively associated with CD8+ T cell counts but positively associated with neutrophil count, CRP, ALT, and AST concentrations, and sex (females) was negatively associated with neutrophil count, CRP, ALT, and AST concentrations. The present study suggests that SARS-CoV-2 infection mainly induced 1) beneficial sex (female)-related differences regarding reduced COVID-19 disease severity and negative associations with inflammatory responses and liver damage, and 2) harmful age-related differences relating to negative associations with CD8+ T cell count and positive associations with inflammatory responses and liver damage. Thus, sex and age are biological variables that should be considered in the prevention and treatment of COVID-19.

Age cohorts stratified according to age-distributions of COVID-19 morbidity statistics identify uniquely age-dependent CD3+CD8+ T-cell lymphocytopenia in COVID-19 patients without comorbidities on admission.

If age boundaries are arbitrarily or roughly defined, age-related analyses can result in questionable findings. Here, we aimed to delineate the uniquely age-dependent immune features of coronavirus disease 2019 (COVID-19) in a retrospective study of 447 patients, stratified according to age distributions of COVID-19 morbidity statistics into well-defined age-cohorts (2-25y, 26-38y, 39-57y, 58-68y, and 69-79y). Age-dependent susceptibilities and severities of the disease were observed in COVID-19 patients. A comparison of the lymphocyte counts among the five age-groups indicated that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection led to age-dependent lymphopenia. Among the lymphocyte subsets, the CD8+ T cell count alone was significantly and age-dependently decreased (520, 385, 320, 172, and 139 n/µl in the five age-groups, respectively). In contrast, the CD4+ T cell, B cell, and natural killer cell counts did not differ among age-cohorts. Age and CD8+ T cell counts (r=‒0.435, p<0.0001) were negatively correlated in COVID-19 patients. Moreover, SARS-CoV-2 infection age-dependently increased the plasma C-reactive protein concentrations (2.0, 5.0, 9.0, 11.6, and 36.1 mg/L in the five age-groups, respectively). These findings can be used to elucidate the role of CD8+ T cells in age-related pathogenesis and to help develop therapeutic strategies for COVID-19.

Thalidomide combined with short-term low-dose glucocorticoid therapy for the treatment of severe COVID-19: A case-series study.

OBJECTIVES: The aim was to evaluate the safety and effectiveness of thalidomide, an immunomodulatory agent, in combination with glucocorticoid, for the treatment of COVID-19 patients with life-threatening symptoms. METHODS: A nonrandomized comparative case series study was performed. Six patients received thalidomide 100 mg per day (with therapy lasting for ≥7 days) plus low-dose short-term dexamethasone, and 6 control patients matched with patients in the thalidomide group, received low-dose short-term treatment with dexamethasone alone. The main outcomes were: the duration of SARS-CoV-2 negative conversion from admission; length of hospital stay; and changes in inflammatory cytokine concentrations and lymphocyte subsets. RESULTS: The median thalidomide treatment time was 12.0 days. The median duration of SARS-CoV-2 negative conversion from admission and hospital stay length were briefer in the thalidomide group compared to the control group (respectively, 11.0 vs 23.0 days, P = 0.043; 18.5 vs 30.0 days, P = 0.043). The mean reduction rates at 7-10 days after treatment for serum interleukin-6 and interferon-γ concentrations were greater in the thalidomide group compared to the control group. Alterations in lymphocyte numbers in the subsets between the 2 groups were similar. CONCLUSIONS: Thalidomide plus short-term glucocorticoid therapy is an effective and safe regimen for the treatment of severely ill COVID-19 patients. The mechanism of action is most likely inhibition of inflammatory cytokine production.

New Measures for the Coronavirus Disease 2019 Response: A Lesson From the Wenzhou Experience.

As the outbreak of coronavirus disease 2019 (COVID-19) has spread globally, determining how to prevent the spread is of paramount importance. We reported the effectiveness of different responses of 4 affected cities in preventing the COVID-19 spread. We expect the Wenzhou anti-COVID-19 measures may provide information for cities around the world that are experiencing this epidemic.