We read the work by Ovalle-Chao et al.,1 “Performance of the predictive criteria of the American Society for Gastrointestinal Endoscopy in the diagnosis of choledocholithiasis at a secondary care public hospital in the state of Nuevo León, Mexico”, with interest. Their aim was to validate the performance of said criteria for predicting choledocholithiasis.

It is possible to develop mathematical algorithms through artificial intelligence (AI) models that analyze information and create prediction models that recognize data patterns with great accuracy. In the context of diagnosis and prognosis, these algorithmic processes play an essential role in the application of new technologies to medical practice. There is evidence that AI models have the capacity to classify patients diagnosed with choledocholithiasis, thus accurately aiding in selecting an efficacious treatment.2

In their study, Ovalle-Chao et al.1 corroborated the diagnosis of choledocholithiasis with 64.2% accuracy, 68.7% sensitivity, and 52% specificity. Dalai et al.3 developed a predictive machine-learning model that analyzed 270 patients with choledocholithiasis diagnoses confirmed through magnetic resonance cholangiography or ultrasound, resulting in 77% accuracy, 77% sensitivity, 75% specificity, a negative predictive value (NPV) of 37% and a positive predictive value (PPV) of 94%, for predicting the presence of choledocholithiasis.

In a study that included 94 patients, Herrera et al.4 applied a mathematical model to predict the presence of choledocholithiasis, comparing it with the model established by the American Society for Gastrointestinal Endoscopy (ASGE). They reported that the high risk for choledocholithiasis had 70.3% accuracy, 61% sensitivity, 85.7% specificity, a PPV of 87.5%, and a NPV of 57.1%.

In conclusion, given that AI-based diagnostic accuracy and prediction were higher, compared with other approaches, AI should be recognized as a useful instrument for predicting diagnoses, with a high potential for aiding in clinical decision-making. The implementation of machine-learning is imperative because it will improve clinical prediction models. Adding it to daily practice will always result in more accurate approaches, fewer cost overruns, and be in line with the dynamics of global change. Characterization and validation studies of these choledocholithiasis predictors should be encouraged, with future research embracing AI in the study methodologies.