Deep Learning Models to Predict Fatal Pneumonia Using Chest X-Ray Images.

Background and Aims: Chest X-ray (CXR) is indispensable to the assessment of severity, diagnosis, and management of pneumonia. Deep learning is an artificial intelligence (AI) technology that has been applied to the interpretation of medical images. This study investigated the feasibility of classifying fatal pneumonia based on CXR images using deep learning models on publicly available platforms. Methods: CXR images of patients with pneumonia at diagnosis were labeled as fatal or nonfatal based on medical records. We applied CXR images from 1031 patients with nonfatal pneumonia and 243 patients with fatal pneumonia for training and self-evaluation of the deep learning models. All labeled CXR images were randomly allocated to the training, validation, and test datasets of deep learning models. Data augmentation techniques were not used in this study. We created two deep learning models using two publicly available platforms. Results: The first model showed an area under the precision-recall curve of 0.929 with a sensitivity of 50.0% and a specificity of 92.4% for classifying fatal pneumonia. We evaluated the performance of our deep learning models using sensitivity, specificity, PPV, negative predictive value (NPV), accuracy, and F1 score. Using the external validation test dataset of 100 CXR images, the sensitivity, specificity, accuracy, and F1 score were 68.0%, 86.0%, 77.0%, and 74.7%, respectively. In the original dataset, the performance of the second model showed a sensitivity, specificity, and accuracy of 39.6%, 92.8%, and 82.7%, respectively, while external validation showed values of 38.0%, 92.0%, and 65.0%, respectively. The F1 score was 52.1%. These results were comparable to those obtained by respiratory physicians and residents. Conclusions: The deep learning models yielded good accuracy in classifying fatal pneumonia. By further improving the performance, AI could assist physicians in the severity assessment of patients with pneumonia.

Splenic volume in pneumococcal pneumonia patients is associated with disease severity and mortality.

Splenectomy is a risk factor for serious pneumococcal disease like overwhelming post-splenectomy infection (OPSI). In healthy individuals with small spleen, fulminant pneumococcal infection similar to OPSI has been reported. Furthermore, it is reported that small spleen was associated with severe pneumococcal infection patients treated in an intensive care unit. However, the association between the small spleen and pneumococcal pneumonia was not investigated enough. We retrospectively analyzed patients with pneumococcal pneumonia who underwent computed tomography examination with measurement of the splenic volume at Harasanshin Hospital between 2004 and 2019. Data on their background characteristics, laboratory findings, and clinical courses were collected. 413 patients were included in the final analysis. The splenic volume was significantly lower in the moderate (P < 0.001), severe (P < 0.00005), and extremely severe (P < 0.001) pneumonia groups compared with the mild pneumonia group. Furthermore, the splenic volume was significantly lower in patients died within 30 days of pneumonia treatment (median of 73.49 versus 110.77 cm3, P < 0.005) or during hospitalization (median of 71.69 versus 111.01 cm3, P < 0.0005). Splenic volume <40 cm3 was significantly associated with mortality within 30 days and total hospital mortality as a risk factor in univariate analysis. Splenic volume <40 cm3 was an independent risk factor for mortality within 30 days (odds ratio: 5.0, 95% confidence interval: 1.2-21.1, P < 0.05) and total hospital mortality (odds ratio: 7.4, 95% confidence interval: 1.8-30.6, P < 0.01) in multivariate logistic regression analysis. These results suggest that small spleen is a risk factor for severity and mortality of pneumococcal pneumonia.

Retrospective cohort study on the safety and efficacy of docetaxel in Japanese non-small cell lung cancer patients with nondialysis chronic kidney disease stage 3b or higher.

BACKGROUND: It has been reported that 20% of lung cancer patients have renal impairment caused by chronic kidney disease (CKD). Since docetaxel is predominantly excreted by the hepatobiliary system, it is administered to non-small cell lung cancer (NSCLC) patients with renal impairment. However, few clinical data are available on the toxicity and efficacy of docetaxel for patients with nondialysis renal impairment. Furthermore, some cases of tubular nephrotoxicity caused by docetaxel in NSCLC patients have been reported. Therefore, a retrospective cohort study was conducted to assess the influence of nondialysis CKD on the toxicity and efficacy of docetaxel in NSCLC patients. METHODS: NSCLC patients who received docetaxel were assessed for renal function, occurrence of adverse events and treatment efficacy. RESULTS: A total of 34 NSCLC patients who received docetaxel were studied. Eight (23.5%) patients had nondialysis CKD stage 3b or higher, with an estimated glomerular filtration rate (eGFR) <45 mL/min/1.73 m2 . Although the differences were not statistically significant, the starting dose of docetaxel (mg/m2 ) was lower (60 mg/m2 ; 37.5% vs. 69.2%) in patients with an eGFR <45 than that in patients with an eGFR ≥45. No significant association was observed between pretreatment eGFR and hematological and nonhematological toxicities. No significant difference was observed in the disease control rate (62.5% vs. 65.4%, P = 1.000) or in the median overall survival (10.7 vs. 11.7, P = 0.735) between patients with an eGFR <45 and those with an eGFR ≥45. CONCLUSION: Docetaxel is a reasonable option for NSCLC patients with nondialysis CKD stage 3b or higher. Dose reduction of docetaxel is also a possibility for NSCLC patients with CKD stage 3b or higher.