Artificial intelligence-assisted system for the assessment of Forrest classification of peptic ulcer bleeding: a multicenter diagnostic study.

BACKGROUND: Inaccurate Forrest classification may significantly affect clinical outcomes, especially in high risk patients. Therefore, this study aimed to develop a real-time deep convolutional neural network (DCNN) system to assess the Forrest classification of peptic ulcer bleeding (PUB). METHODS: A training dataset (3868 endoscopic images) and an internal validation dataset (834 images) were retrospectively collected from the 900th Hospital, Fuzhou, China. In addition, 521 images collected from four other hospitals were used for external validation. Finally, 46 endoscopic videos were prospectively collected to assess the real-time diagnostic performance of the DCNN system, whose diagnostic performance was also prospectively compared with that of three senior and three junior endoscopists. RESULTS: The DCNN system had a satisfactory diagnostic performance in the assessment of Forrest classification, with an accuracy of 91.2% (95%CI 89.5%-92.6%) and a macro-average area under the receiver operating characteristic curve of 0.80 in the validation dataset. Moreover, the DCNN system could judge suspicious regions automatically using Forrest classification in real-time videos, with an accuracy of 92.0% (95%CI 80.8%-97.8%). The DCNN system showed more accurate and stable diagnostic performance than endoscopists in the prospective clinical comparison test. This system helped to slightly improve the diagnostic performance of senior endoscopists and considerably enhance that of junior endoscopists. CONCLUSION: The DCNN system for the assessment of the Forrest classification of PUB showed satisfactory diagnostic performance, which was slightly superior to that of senior endoscopists. It could therefore effectively assist junior endoscopists in making such diagnoses during gastroscopy.

AcornHRD: an HRD algorithm highly associated with anthracycline-based neoadjuvant chemotherapy in breast cancer in China.

PURPOSE: Our study aimed to develop and validate a homologous recombination deficiency (HRD) scoring algorithm in the Chinese breast cancer population. METHODS AND MATERIALS: Ninety-six in-house breast cancer (BC) samples and 6 HRD-positive standard cells were analyzed by whole-genome sequencing (WGS). Besides, 122 BCs from the TCGA database were down-sampled to ~ 1X WGS. We constructed an algorithm named AcornHRD for HRD score calculated based on WGS at low coverage as input data to estimate large-scale copy number alteration (LCNA) events on the genome. A clinical cohort of 50 BCs (15 cases carrying BRCA mutation) was used to assess the association between HRD status and anthracyclines-based neoadjuvant treatment outcomes. RESULTS: A 100-kb window was defined as the optimal size using 41 in-house cases and the TCGA dataset. HRD score high threshold was determined as HRD score ≥ 10 using 55 in-house BCs with BRCA mutation to achieve a 95% BRCA-positive agreement rate. Furthermore, the HRD status agreement rate of AcornHRD is 100%, while the ShallowHRD is 60% in standard cells. BRCA mutation was significantly associated with a high HRD score evaluated by AcornHRD and ShallowHRD (p = 0.008 and p = 0.003, respectively) in the TCGA dataset. However, AcornHRD showed a higher positive agreement rate than did the ShallowHRD algorithm (70% vs 60%). In addition, the BRCA-positive agreement rate of AcornHRD was superior to that of ShallowHRD (87% vs 13%) in the clinical cohort. Importantly, the high HRD score assessed by AcornHRD was significantly correlated with a residual cancer burden score of 0 or 1 (RCB0/1). Besides, the HRD-positive group was more likely to respond to anthracycline-based chemotherapy than the HRD-negative group (pCR [OR = 9.5, 95% CI 1.11-81.5, p = 0.040] and RCB0/1 [OR = 10.29, 95% CI 2.02-52.36, p = 0.005]). CONCLUSION: Using the AcornHRD algorithm evaluation, our analysis demonstrated the high performance of the LCNA genomic signature for HRD detection in breast cancers.