Community-Acquired Pneumonia and Hospital-Acquired Pneumonia in Adult Patients with Idiopathic Inflammatory Myopathy: Outcome and Antibiotic Therapy

An explainable machine learning-based model to predict intensive care unit admission among patients with community-acquired pneumonia and connective tissue disease

BACKGROUND

There is no individualized prediction model for intensive care unit (ICU) admission on patients with community-acquired pneumonia (CAP) and connective tissue disease (CTD) so far. In this study, we aimed to establish a machine learning-based model for predicting the need for ICU admission among those patients.

METHODS

This was a retrospective study on patients admitted into a University Hospital in China between November 2008 and November 2021. Patients were included if they were diagnosed with CAP and CTD during admission and hospitalization. Data related to demographics, CTD types, comorbidities, vital signs and laboratory results during the first 24 h of hospitalization were collected. The baseline variables were screened to identify potential predictors via three methods, including univariate analysis, least absolute shrinkage and selection operator (Lasso) regression and Boruta algorithm. Nine supervised machine learning algorithms were used to build prediction models. We evaluated the performances of differentiation, calibration, and clinical utility of all models to determine the optimal model. The Shapley Additive Explanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) techniques were performed to interpret the optimal model.

RESULTS

The included patients were randomly divided into the training set (1070 patients) and the testing set (459 patients) at a ratio of 70:30. The intersection results of three feature selection approaches yielded 16 predictors. The eXtreme gradient boosting (XGBoost) model achieved the highest area under the receiver operating characteristic curve (AUC) (0.941) and accuracy (0.913) among various models. The calibration curve and decision curve analysis (DCA) both suggested that the XGBoost model outperformed other models. The SHAP summary plots illustrated the top 6 features with the greatest importance, including higher N-terminal pro-B-type natriuretic peptide (NT-proBNP) and C-reactive protein (CRP), lower level of CD4 + T cell, lymphocyte and serum sodium, and positive serum (1,3)-β-D-glucan test (G test).

CONCLUSION

We successfully developed, evaluated and explained a machine learning-based model for predicting ICU admission in patients with CAP and CTD. The XGBoost model could be clinical referenced after external validation and improvement.

Elevated spleen FDG uptake predicts unfavorable outcome in adult idiopathic-inflammatory-myopathy patients: a crisis beyond muscles

BACKGROUND

Idiopathic inflammatory myopathy (IIM) is a group of autoimmune diseases that frequently leads to unfavorable outcome. This study aimed at identifying the clinical value of PET/CT scan in predicting the outcome of adult IIM patients.

METHODS

Adult IIM patients who were admitted to the four divisions of the First Affiliated Hospital, Zhejiang University School of Medicine (FAHZJU), from January 1, 2017, to December 31, 2020, were retrospectively reviewed. PET/CT scan and other factors of IIM patients who met the inclusion and exclusion criteria were collected and analyzed.

RESULTS

A total of 69 adult IIM patients were finally enrolled into this study. Thirty cases (43.5%) of all the patients enrolled died in follow-up, and the medium follow-up time was 11.90 (4.00, 23.80) months. In particular, 14 patients died within 3 months. The univariate Cox proportional hazards regression analyses revealed pulmonary bacterial infection (P < 0.001), rapidly progressive interstitial lung disease (RP-ILD, P = 0.018), maximum standard uptake value of spleen (spleen SUVmax, P = 0.002), and positivity of anti-MDA5 antibody (P = 0.041) were significantly related to survival in follow-up. The following multivariate Cox proportional hazards regression analysis identified pulmonary bacterial infection (P = 0.003) and spleen SUVmax (P = 0.032) as factors significantly associated with survival of IIM-ILD patients. The subsequent receiver operating characteristic curve analysis showed SUVmax was comparably effective in predicting death within 3 months.

CONCLUSION

Spleen SUVmax and complication of pulmonary bacterial infection were significantly associated with survival of IIM patients. In addition, elevated spleen SUVmax was efficient in predicting unfavorable outcome of adult IIM patients. Key Points • IIM is a group of autoimmune diseases that frequently leads to unfavorable outcome • Complications of splenic SUVmax and pulmonary bacterial infection were significantly associated with survival in IIM patients.

The lungs were on fire: a pilot study of 18F-FDG PET/CT in idiopathic-inflammatory-myopathy-related interstitial lung disease

Background

Interstitial lung disease (ILD) and its rapid progression (RP) are the main contributors to unfavourable outcomes of patients with idiopathic inflammatory myopathy (IIM). This study aimed to identify the clinical value of PET/CT scans in IIM-ILD patients and to construct a predictive model for RP-ILD.

Methods

Adult IIM-ILD patients who were hospitalized at four divisions of the First Affiliated Hospital, Zhejiang University School of Medicine (FAHZJU), from 1 January 2017 to 31 December 2020 were reviewed. PET/CT scans and other characteristics of patients who met the inclusion and exclusion criteria were collected and analysed.

Results

A total of 61 IIM-ILD patients were enrolled in this study. Twenty-one patients (34.4%) developed RP-ILD, and 24 patients (39.3%) died during follow-up. After false discovery rate (FDR) correction, the percent-predicted diffusing capacity of the lung for carbon monoxide (DLCO%, P = 0.014), bilateral lung mean standard uptake value (SUVmean, P = 0.014) and abnormal mediastinal lymph node (P = 0.045) were significantly different between the RP-ILD and non-RP-ILD groups. The subsequent univariate and multivariate logistic regression analyses verified our findings. A “DLM” model was established by including the above three values to predict RP-ILD with a cut-off value of ≥ 2 and an area under the curve (AUC) of 0.905. Higher bilateral lung SUVmean (P = 0.019) and spleen SUVmean (P = 0.011) were observed in IIM-ILD patients who died within 3 months, and a moderate correlation was recognized between the two values.

Conclusions

Elevated bilateral lung SUVmean, abnormal mediastinal lymph nodes and decreased DLCO% were significantly associated with RP-ILD in IIM-ILD patients. The “DLM” model was valuable in predicting RP-ILD and requires further validation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13075-021-02578-9.