Background/Objectives: In routine practice, colonoscopy findings are saved as unstructured free text, limiting secondary use. Accurate named-entity recognition (NER) is essential to unlock these descriptions for quality monitoring, personalized medicine and research. We compared named-entity recognition (NER) models trained on real, synthetic, and mixed data to determine whether privacy preserving synthetic reports can boost clinical information extraction. Methods: Three Spark NLP biLSTM CRF models were trained on (i) 100 manually annotated Romanian colonoscopy reports (ModelR), (ii) 100 prompt-generated synthetic reports (ModelS), and (iii) a 1:1 mix (ModelM). Performance was tested on 40 unseen reports (20 real, 20 synthetic) for seven entities. Micro-averaged precision, recall, and F1-score values were computed; McNemar tests with Bonferroni correction assessed pairwise differences. Results: ModelM outperformed single-source models (precision 0.95, recall 0.93, F1 0.94) and was significantly superior to ModelR (F1 0.70) and ModelS (F1 0.64; p < 0.001 for both). ModelR maintained high accuracy on real text (F1 = 0.90), but its accuracy fell when tested on synthetic data (0.47); the reverse was observed for ModelS (F1 = 0.99 synthetic, 0.33 real). McNemar χ2 statistics (64.6 for ModelM vs. ModelR; 147.0 for ModelM vs. ModelS) greatly exceeded the Bonferroni-adjusted significance threshold (α = 0.0167), confirming that the observed performance gains were unlikely to be due to chance. Conclusions: Synthetic colonoscopy descriptions are a valuable complement, but not a substitute for real annotations, while AI is helping human experts, not replacing them. Training on a balanced mix of real and synthetic data can help to obtain robust, generalizable NER models able to structure free-text colonoscopy reports, supporting large-scale, privacy-preserving colorectal cancer surveillance and personalized follow-up.

Background/Objectives: Colorectal cancer (CRC) is a widespread malignancy that can be mitigated through early detection and removal of precancerous polyps. Artificial intelligence (AI), particularly deep learning, enhances polyp detection during colonoscopy but faces challenges due to limited medical imaging datasets. This study evaluates whether synthetic and pseudosynthetic data—augmented data generated from original datasets—can improve AI accuracy for polyp detection. Methods: We used multiple real and synthetic polyp datasets, applying data augmentation techniques to create pseudosynthetic data, and employing a modified U-Net architecture for polyp segmentation. The model was trained and evaluated across ten experiments using different combinations of real, pseudosynthetic, and synthetic data. Performance metrics included accuracy, precision, recall, Dice coefficient, Intersection over Union (IoU), and F1 score. External validation used the CVC-Colon-DB dataset with 612 image-mask pairs. Results: Combining real and pseudosynthetic data achieved the best performance, with a Dice score of 0.7638, precision of 0.7638, recall of 0.6774, and F1 score of 0.8979. Models trained solely on CycleGAN-generated data performed poorly, while diffusion-based synthetic data offered better generalization (precision: 0.7488, recall: 0.6695, F1: 0.8987). Conclusions: U-Net models trained with synthetic and pseudosynthetic data outperform those trained solely on real data, effectively addressing data scarcity, diversity, and ethical concerns. Models benefit more from pseudosynthetic data alone compared to mixed sources. Diffusion-generated synthetic data leads to better model performance than GAN-generated data. These findings confirm that synthetic and pseudosynthetic data are effective tools to improve model generalization and address ethical concerns in AI-assisted diagnostic environments.

Background: Synthetic and pseudosynthetic images can be used to extend colonoscopy datasets, which, in turn, are used to train AI-detection models, yet their clinical acceptability depends on whether medical professionals can still recognize non-real content. Aim: To quantify the ability of practicing gastroenterologists to discriminate real, pseudosynthetic, and synthetic polyp images and to determine how training level and synthesis method impact detection. Materials and Methods: A total of 32 Romanian gastroenterologists (18 residents and 14 seniors) reviewed 24 images (8 real, 8 augmented, 4 CycleGAN, and 4 diffusion) via an online form. Classification accuracy, 95% confidence intervals (CI), class sensitivity and precision, 3 × 3 confusion matrices, and Fleiss’ κ were calculated. Resident vs. senior differences were tested with Pearson χ²; CycleGAN versus diffusion detectability was analyzed with the Wilcoxon signed-rank test (α = 0.05). Results: Overall accuracy was 61.2% (95% CI 57.7–64.6). Residents and seniors performed similarly (62.3% vs. 59.8%; χ²₁ = 0.38, p = 0.54). Sensitivity/precision were 70.7%/62.2% for real, 51.6%/58.9% for augmented, and 61.3%/62.1% for synthetic images. Collapsing to “real vs. non-real” yielded 70.7% sensitivity and 78.5% specificity for real images. CycleGAN images were always recognized as synthetic (128/128; 97.1–100% CI), whereas diffusion images were correctly classified only 22.7% of the time (16.3–30.6%; Wilcoxon p < 0.001). The training level did not impact detection performance (χ²₂ < 1.2, p > 0.5). Inter-rater agreement was fair (κ = 0.30, 95% CI 0.15–0.43). Conclusions: Clinicians detect non-real colonoscopy images only slightly above chance, irrespective of experience. The diffusion synthesis method creates images that escape human scrutiny, suggesting the need for automated authenticity safeguards before synthetic datasets are applied in clinical or AI-validation contexts.

Background/Objectives: Previous studies have demonstrated the prognostic value of the C-reactive protein-to-albumin ratio (CAR) in acute pancreatitis, but the incremental value of serial measurements remains unclear. We evaluated whether CAR measured at 24 and 48 h provides additional prognostic information beyond admission CAR for predicting severe acute pancreatitis (SAP). Methods: In this retrospective cohort study, 150 consecutive patients with acute pancreatitis were included. CAR was calculated at admission (CAR-T0), 24 h (CAR-T24), and 48 h (CAR-T48). Univariable and multivariable logistic regression, receiver operating characteristic (ROC) analysis, and DeLong’s test were used to evaluate prognostic performance. Secondary analyses explored associations with pancreatic pseudocyst, pancreatic ascites, and splanchnic venous thrombosis. Results: Forty-four patients (29.3%) developed SAP. CAR was significantly associated with SAP at all three time points, with CAR-T24 demonstrating the highest predictive performance (AUC 0.874), followed by CAR-T48 (AUC 0.857) and CAR-T0 (AUC 0.829). CAR-T24 showed the highest individual discriminative performance, although its AUC was not significantly higher than that of CAR-T0 (DeLong p = 0.067). In contrast, adding CAR-T24 to the prespecified multivariable model significantly improved discrimination compared with the admission model (AUC 0.890 vs. 0.841; p = 0.016). Dynamic changes in CAR (ΔCAR) were not independently associated with SAP. Admission NLR showed poor predictive performance (AUC 0.578) and was significantly inferior to all CAR time points. Exploratory analyses suggested a potential association between CAR-T24 and splanchnic venous thrombosis; however, given the small number of thrombotic events (n = 10), this finding should be considered preliminary and hypothesis-generating. Conclusions: Serial CAR assessment, particularly at 24 h after admission, provides incremental prognostic value beyond admission CAR alone and may provide additional prognostic information during early clinical reassessment in acute pancreatitis.

Background and Objectives: The most common method for treating malocclusions today is fixed orthodontic therapy, during which brackets and tubes are bonded to the surface of the teeth, which makes oral hygiene difficult to maintain. Increased plaque retention, gingival bleeding, and gingivitis can be diagnosed in the early phases of treatment. The periodontal response to plaque accumulation can be explained by quantitative and qualitative changes in the subgingival microbiota. The aim of our research was to investigate the changes in the subgingival microbiota that occurred within 6–8 weeks after bonding when two different orthodontic adhesives were used. Materials and Methods: Thirty patients were followed; molar tubes were bonded with a composite (C) in fifteen cases, and in the other fifteen cases, they were bonded with glass ionomer cementum (GIC). A microbiological sample was taken from the gingival sulcus of the maxillary first molars at the time of appliance placement (T1) and 6–8 weeks (T2) after bonding. Bacterial DNA detection was performed using the micro-IDent^®plus11 (Hain Lifescience GmbH, Germany) PCR (polymerase chain reaction) method. The statistical analysis included McNemar’s test to analyze the paired binary data and Fisher’s Exact Test to compare the prevalence of each of the 11 bacteria at T1 and, ultimately, T2 between the two studied groups. The Bonferroni correction was also applied. Results: When analyzing GIC vs. C at T1 and T2, none of the studied pathogens showed significant differences. Conclusions: Given the lack of statistical significance, these trends do not confirm a definitive impact of the procedure on bacterial presence. The increased presence of periodontal pathogens might suggest that bonding does not reduce the bacterial loading of subgingival microbiota. Less protective effects of the GIC intervention against Tannerella forsythia and Eubacterium nodatum bacteria were detected.