Abstract Background Acute Respiratory Distress syndrome (ARDS) is a common complication of Acute Pancreatitis (AP) and is associated with high mortality. This study used Machine Learning (ML) to predict ARDS in patients with AP at admission. Methods The authors retrospectively analyzed the data from patients with AP from January 2017 to August 2022. Clinical and laboratory parameters with significant differences between patients with and without ARDS were screened by univariate analysis. Then, Support Vector Machine (SVM), Ensembles of Decision Trees (EDTs), Bayesian Classifier (BC), and nomogram models were constructed and optimized after feature screening based on these parameters. Five-fold cross-validation was used to train each model. A test set was used to evaluate the predictive performance of the four models. Results A total of 83 (18.04%) of 460 patients with AP developed ARDS. Thirty-one features with significant differences between the groups with and without ARDS in the training set were used for modeling. The Partial Pressure of Oxygen (PaO2), C-reactive protein, procalcitonin, lactic acid, Ca2+, the neutrophil:lymphocyte ratio, white blood cell count, and amylase were identified as the optimal subset of features. The BC algorithm had the best predictive performance with the highest AUC value (0.891) than SVM (0.870), EDTs (0.813), and the nomogram (0.874) in the test set. The EDT algorithm achieved the highest accuracy (0.891), precision (0.800), and F1 score (0.615), but the lowest FDR (0.200) and the second-highest NPV (0.902). Conclusions A predictive model of ARDS complicated by AP was successfully developed based on ML. Predictive performance was evaluated by a test set, for which BC showed superior predictive performance and EDTs could be a more promising prediction tool for larger samples. (ARDS (AP mortality ML (ML admission 201 2022 analysis Then SVM, , (SVM) EDTs, (EDTs) BC, (BC) Fivefold Five fold crossvalidation cross validation 8 18.04% 1804 18 04 (18.04% 46 Thirtyone Thirty one modeling PaO2, PaO2 PaO (PaO2) Creactive C reactive protein procalcitonin acid Ca2 Ca Ca2+ neutrophillymphocyte neutrophil lymphocyte ratio count 0.891 0891 0 891 (0.891 0.870, 0870 0.870 870 (0.870) 0.813, 0813 0.813 813 (0.813) 0.874 0874 874 (0.874 0.891, 0.800, 0800 0.800 800 (0.800) F 0.615, 0615 0.615 615 (0.615) 0.200 0200 200 (0.200 secondhighest second 0.902. 0902 0.902 . 902 (0.902) samples 20 202 (SVM (EDTs (BC 18.04 180 1 (18.04 4 (PaO2 0.89 089 89 (0.89 087 0.87 87 (0.870 081 0.81 81 (0.813 (0.87 080 0.80 80 (0.800 061 0.61 61 (0.615 0.20 020 (0.20 090 0.90 90 (0.902 2 18.0 (18.0 (PaO 0.8 08 (0.8 (0.81 (0.80 06 0.6 6 (0.61 0.2 02 (0.2 09 0.9 9 (0.90 18. (18. 0. (0. (0.6 (0.9 (18 (0 (1 (

Abstract Background and aims: Potassium-Competitive Acid Blockers (P-CABs) have been used in Helicobacter pylori (H. pylori) eradication therapies in recent years. However, the efficacy and safety of P-CABs compared to ProtonPump Inhibitors (PPIs) in this setting remain controversial. Methods: The efficacy and safety of P-CABs and PPIs for H. pylori eradication were compared in a meta-analysis based on a systematic literature search of major electronic databases for relevant Randomized Controlled Trials (RCTs). Results: Seven studies and 1,168 patients were included. The pooled eradication rate determined by Intention-ToTreat (ITT) analysis was 90.2% for P-CAB-based and 75.5% for PPI-based triple therapy (pooled RR [95% CI] = 1.17 [1.08-1.28], p < 0.001). The Per-Protocol (PP) analysis also demonstrated significant superiority of P-CABs (pooled eradication rate = 92.4% vs. 77.8%; pooled RR [95% CI] = 1.14 [1.03–1.26], p < 0.01). In a subgroup evaluation, P-CABs were significantly better than PPIs as a first-line eradication therapy, in both the ITT analysis (pooled eradication rate = 91.8% vs. 76.4%; pooled RR [95% CI] = 1.18 [1.10-1.28], p < 0.0001) and the PP analysis (pooled eradication rate = 93.0% vs. 78.6%; pooled RR [95% CI] = 1.13 [1.02 –1.26], p < 0.05). However, P-CABs were not superior to PPIs when administered as salvage therapy, as determined in the ITT (75.0% vs. 66.0%, pooled RR [95% CI] = 1.11 [0.69–1.78], p = 0.66) and PP (85.7% vs. 70.0%, pooled RR [95% CI] = 1.20 [0.82–1.75], p = 0.34) analyses. In a subgroup analysis limited to Japanese patients, both the ITT analysis (pooled eradication rate = 89.6% vs. 73.9%; RR [95% CI] = 1.21 [1.14 –1.29], p < 0.01) and the PP analysis (pooled eradication rate = 92.0% vs. 75.7%; RR [95% CI] = 1.18 [1.06 –1.32], p < 0.01) showed that P-CABs were significantly superior compared to PPIs as triple eradication therapy. However, in the subgroup analysis of patients from other countries, there was no significant difference in either the ITT analysis (pooled eradication rate = 93.8% vs. 85.2%; RR [95% CI] = 1.10 [0.99-1.22], p = 0.07) or PP analysis (pooled eradication rate = 95.0% vs. 90.8%; RR [95% CI] = 1.05 [0.98–1.14], p = 0.17). The incidence of adverse events associated with the two regimens did not significantly differ (P-CABs vs. PPIs: 33.6% vs. 40.0%; RR [95% CI] = 0.84 [0.71‒1.00], p = 0.05). The incidence of serious adverse events and dropout rate due to adverse events also did not differ (p = 0.44 and p = 0.67, respectively). Conclusions: The efficacy of P-CAB-based triple therapy is superior to that of PPI-based triple therapy as a first-line approach to H. pylori eradication, particularly in Japanese patients. As salvage therapy, the efficacy of the two treatments did not significantly differ. The tolerability of P-CAB-based and PPI-based triple therapy was comparable, as was the incidence of adverse events. HIGHLIGHTS The efficacy of P-CAB-based triple therapy is superior to that of PPI-based triple therapy as a first-line approach to H. pylori eradication, particularly in Japanese patients. P-CABs were not superior to PPIs as a salvage triple eradication therapy. The safety and tolerability of P-CAB are comparable to PPI in H. pylori triple eradication therapies. Further large RCTs conducted in multiple regions and countries are necessary.