Added value of chest CT in a machine learning-based prediction model to rule out COVID-19 before inpatient admission: A retrospective university network study.

PURPOSE: During the coronavirus disease 2019 (COVID-19) pandemic, hospitals still face the challenge of timely identification of infected individuals before inpatient admission. An artificial intelligence approach based on an established clinical network may improve prospective pandemic preparedness. METHOD: Supervised machine learning was used to construct diagnostic models to predict COVID-19. A pooled database was retrospectively generated from 4437 participant data that were collected between January 2017 and October 2020 at 12 German centers that belong to the radiological cooperative network of the COVID-19 (RACOON) consortium. A total of 692 (15.6 %) participants were COVID-19 positive according to the reference of the reverse transcription-polymerase chain reaction test. The diagnostic models included chest CT features (model R), clinical examination and laboratory test features (model CL), or all three feature categories (model RCL). Performance outcomes included accuracy, sensitivity, specificity, negative and positive predictive value, and area under the receiver operating curve (AUC). RESULTS: Performance of predictive models improved significantly by adding chest CT features to clinical evaluation and laboratory test features. Without (model CL) and with inclusion of chest CT (model RCL), sensitivity was 0.82 and 0.89 (p < 0.0001), specificity was 0.84 and 0.89 (p < 0.0001), negative predictive value was 0.96 and 0.97 (p < 0.0001), AUC was 0.92 and 0.95 (p < 0.0001), and proportion of false negative classifications was 2.6 % and 1.7 % (p < 0.0001), respectively. CONCLUSIONS: Addition of chest CT features to machine learning-based predictive models improves the effectiveness in ruling out COVID-19 before inpatient admission to regular wards.

Artificial intelligence for assistance of radiology residents in chest CT evaluation for COVID-19 pneumonia: a comparative diagnostic accuracy study.

BACKGROUND: In hospitals, it is crucial to rule out coronavirus disease 2019 (COVID-19) timely and reliably. Artificial intelligence (AI) provides sufficient accuracy to identify chest computed tomography (CT) scans with signs of COVID-19. PURPOSE: To compare the diagnostic accuracy of radiologists with different levels of experience with and without assistance of AI in CT evaluation for COVID-19 pneumonia and to develop an optimized diagnostic pathway. MATERIAL AND METHODS: The retrospective, single-center, comparative case-control study included 160 consecutive participants who had undergone chest CT scan between March 2020 and May 2021 without or with confirmed diagnosis of COVID-19 pneumonia in a ratio of 1:3. Index tests were chest CT evaluation by five radiological senior residents, five junior residents, and an AI software. Based on the diagnostic accuracy in every group and on comparison of groups, a sequential CT assessment pathway was developed. RESULTS: Areas under receiver operating curves were 0.95 (95% confidence interval [CI]=0.88-0.99), 0.96 (95% CI=0.92-1.0), 0.77 (95% CI=0.68-0.86), and 0.95 (95% CI=0.9-1.0) for junior residents, senior residents, AI, and sequential CT assessment, respectively. Proportions of false negatives were 9%, 3%, 17%, and 2%, respectively. With the developed diagnostic pathway, junior residents evaluated all CT scans with the support of AI. Senior residents were only required as second readers in 26% (41/160) of the CT scans. CONCLUSION: AI can support junior residents with chest CT evaluation for COVID-19 and reduce the workload of senior residents. A review of selected CT scans by senior residents is mandatory.

Virtual inverted classroom to replace in-person radiology lectures at the time of the COVID-19 pandemic - a prospective evaluation and historic comparison.

BACKGROUND: In the time of the coronavirus disease 2019 (COVID-19) pandemic, in-person lectures had to be shifted to online learning. This study aimed to evaluate students' and lecturers' perception and effectiveness of a virtual inverted classroom (VIC) concept on clinical radiology in comparison to a historic control. METHODS: In the winter semester 2020/21, 136 fourth year medical students who completed the clinical radiology VIC during the pandemic, were included in the single centre, prospective study. Results were compared with a historic control that had finished the physical inverted classroom (PIC) in the immediately preceding year. The VIC consisted of an initial phase of self-determined preparation with learning videos and a second interactive phase of clinical case studies alternating between the virtual lecture hall and virtual buzz groups. At the end of the lecture series, students rated the lecture on a scale of 1 (most positive assessment) to 6 (most negative assessment) through an online survey platform. Additionally, they reported their impressions in free-form text. Lecturers were invited to comment on the VIC in a group interview. Main outcomes were final grades and student perception of the VIC. RESULTS: Students' general impression of VIC was lower than that of PIC (median value of 3 [IQR 4, 2] and 1 [IQR 0, 0], p < 0.001), respectively, p < 0.001). The highest rating was achieved concerning use of the audience response system (median 1 [IQR 1, 0]), and the lowest concerning the buzz groups (median 4 [IQR 5, 3]). Students stated that they would have appreciated more details on reading images, greater focus on plenary case studies, and provision of exam related scripts. Lecturers would have liked better preparation by students, more activity of students, and stronger assistance for group support. Exam grades after VIC were better than after PIC (median 1 [IQR 2, 1] and 2 [IQR 2,1], respectively, p < 0.001). CONCLUSIONS: Students' overall perception of VIC was satisfactory, although worse than PIC. Final grades improved compared to PIC. Provided an adapted buzz group size and support, VIC may serve as complement in medical education once the pandemic is over.

Ruling out COVID-19 by chest CT at emergency admission when prevalence is low: the prospective, observational SCOUT study.

BACKGROUND: It is essential to avoid admission of patients with undetected corona virus disease 2019 (COVID-19) to hospitals' general wards. Even repeated negative reverse transcription polymerase chain reaction (RT-PCR) results do not rule-out COVID-19 with certainty. The study aimed to evaluate a rule-out strategy for COVID-19 using chest computed tomography (CT) in adults being admitted to the emergency department and suspected of COVID-19. METHODS: In this prospective, single centre, diagnostic accuracy cohort study, consecutive adults (≥ 18 years) presenting with symptoms consistent with COVID-19 or previous contact to infected individuals, admitted to the emergency department and supposed to be referred to general ward were included in March and April 2020. All participants underwent low-dose chest CT. RT-PCR- and specific antibody tests were used as reference standard. Main outcome measures were sensitivity and specificity of chest CT. Predictive values were calculated based on the theorem of Bayes using Fagan's nomogram. RESULTS: Of 165 participants (56.4% male, 71 ± 16 years) included in the study, the diagnosis of COVID-19 was confirmed with RT-PCR and AB tests in 13 participants (prevalence 7.9%). Sensitivity and specificity of chest CT were 84.6% (95% confidence interval [CI], 54.6-98.1) and 94.7% (95% CI, 89.9-97.7), respectively. Positive and negative likelihood ratio of chest CT were 16.1 (95% CI, 7.9-32.8) and 0.16 (95% CI, 0.05-0.58) and positive and negative predictive value were 57.9% (95% CI, 40.3-73.7) and 98.6% (95% CI, 95.3-99.6), respectively. CONCLUSION: At a low prevalence of COVID-19, chest CT could be used as a complement to repeated RT-PCR testing for early COVID-19 exclusion in adults with suspected infection before referral to hospital's general wards. Trial registration ClinicalTrials.gov: NCT04357938 April 22, 2020.