A univariate analysis statistically evaluated the radiomic association between XR features and COVID-19 diagnosis, using the receiver operating characteristic (ROC) curve with sensitivity and specificity metrics. The Mann–Whitney U test evaluated the statistical difference between feature distributions from the groups of patients with pneumonia [9]. Each feature had the area under the ROC curve (AUC) and p-value calculated individually.

The short-term prognostic analysis was performed by correlating the radiomic features with overall and deterioration-free survival using the Kaplan–Meier time-to-event method. Higher and lower-risk groups of patients were split according to the median value of the quantitative features [12]. As the number of patients with follow-up data (survival time and outcome result) was relatively low for this analysis, we combined all cases with COVID-19 in a single set of 28 patients (14 from discovery and 14 from the validation set).

The mean follow-up time was 20.4 days (±7.1 of standard deviation). Overall survival analysis used death by any nature as event, and deterioration-free survival analysis used worsening on clinical/radiological conditions or death by any cause. Patients who survived or remained clinically stable or had loss of follow-up were censored. The log-rank test assessed the statistical difference between the survival curves from both stratified groups to identify features with potential prognostic value [15].

The SciPy v1.2.3 and R v3.4.4 packages were used to perform statistical analysis. Tests with p < 0.05 were considered statistically significant.

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.