3D CT-Inclusive Deep-Learning Model to Predict Mortality, ICU Admittance, and Intubation in COVID-19 Patients.

Chest CT is a useful initial exam in patients with coronavirus disease 2019 (COVID-19) for assessing lung damage. AI-powered predictive models could be useful to better allocate resources in the midst of the pandemic. Our aim was to build a deep-learning (DL) model for COVID-19 outcome prediction inclusive of 3D chest CT images acquired at hospital admission. This retrospective multicentric study included 1051 patients (mean age 69, SD = 15) who presented to the emergency department of three different institutions between 20th March 2020 and 20th January 2021 with COVID-19 confirmed by real-time reverse transcriptase polymerase chain reaction (RT-PCR). Chest CT at hospital admission were evaluated by a 3D residual neural network algorithm. Training, internal validation, and external validation groups included 608, 153, and 290 patients, respectively. Images, clinical, and laboratory data were fed into different customizations of a dense neural network to choose the best performing architecture for the prediction of mortality, intubation, and intensive care unit (ICU) admission. The AI model tested on CT and clinical features displayed accuracy, sensitivity, specificity, and ROC-AUC, respectively, of 91.7%, 90.5%, 92.4%, and 95% for the prediction of patient's mortality; 91.3%, 91.5%, 89.8%, and 95% for intubation; and 89.6%, 90.2%, 86.5%, and 94% for ICU admission (internal validation) in the testing cohort. The performance was lower in the validation cohort for mortality (71.7%, 55.6%, 74.8%, 72%), intubation (72.6%, 74.7%, 45.7%, 64%), and ICU admission (74.7%, 77%, 46%, 70%) prediction. The addition of the available laboratory data led to an increase in sensitivity for patient's mortality (66%) and specificity for intubation and ICU admission (50%, 52%, respectively), while the other metrics maintained similar performance results. We present a deep-learning model to predict mortality, ICU admittance, and intubation in COVID-19 patients. KEY POINTS: • 3D CT-based deep learning model predicted the internal validation set with high accuracy, sensibility and specificity (> 90%) mortality, ICU admittance, and intubation in COVID-19 patients. • The model slightly increased prediction results when laboratory data were added to the analysis, despite data imbalance. However, the model accuracy dropped when CT images were not considered in the analysis, implying an important role of CT in predicting outcomes.

Liver biopsy: Analysis of results of two specialist teams.

AIM: To analyze the safety and the adequacy of a sample of liver biopsies (LB) obtained by gastroenterologist (G) and interventional radiologist (IR) teams. METHODS: Medical records of consecutive patients evaluated at our GI unit from 01/01/2004 to 31/12/2010 for whom LB was considered necessary to diagnose and/or stage liver disease, both in the setting of day hospital and regular admission (RA) care, were retrieved and the data entered in a database. Patients were divided into two groups: one undergoing an ultrasonography (US)-assisted procedure by the G team and one undergoing US-guided biopsy by the IR team. For the first group, an intercostal approach (US-assisted) and a Menghini modified type needle 16 G (length 90 mm) were used. The IR team used a subcostal approach (US-guided) and a semiautomatic modified Menghini type needle 18 G (length 150 mm). All the biopsies were evaluated for appropriateness according to the current guidelines. The number of portal tracts present in each biopsy was assessed by a revision performed by a single pathologist unaware of the previous pathology report. Clinical, laboratory and demographic patient characteristics, the adverse events rate and the diagnostic adequacy of LB were analyzed. RESULTS: During the study period, 226 patients, 126 males (56%) and 100 females (44%), underwent LB: 167 (74%) were carried out by the G team, whereas 59 (26%) by the IR team. LB was mostly performed in a day hospital setting by the G team, while IR completed more procedures on inpatients (P < 0.0001). The groups did not differ in median age, body mass index (BMI), presence of comorbidities and coagulation parameters. Complications occurred in 26 patients (16 G team vs 10 IR team, P = 0.15). Most gross samples obtained were considered suitable for basal histological evaluation, with no difference among the two teams (96.4% G team vs 91.5% IR, P = 0.16). However, the samples obtained by the G team had a higher mean number of portal tracts (G team 9.5 ± 4.8; range 1-29 vs IR team 7.8 ± 4.1; range 1-20) (P = 0.0192) and a longer mean length (G team 22 mm ± 8.8 vs IR team 15 ± 6.5 mm) (P = 0.0001). CONCLUSION: LB can be performed with similar outcomes both by G and IR. Use of larger dimension needles allows obtaining better samples, with a similar rate of adverse events.