Predictive modeling of ethical and professional attitudes and behaviors in Latin American physicians using neural networks

Introduction: Medical professionalism is key to ensuring quality of care and collaboration within the healthcare team. However, negative attitudes can undermine trust and affect teamwork. Artificial intelligence, through neural networks, offers an innovative approach to predicting and analyzing these behaviors, potentially strengthening professional ethics in Latin American physicians. The objective was to generate a predictive model of ethical and professional attitudes and behaviors among Latin American physicians through the use of neural networks. Materials and method: A cross-sectional study was conducted using a survey of 424 physicians from Paraguay, Peru, and Cuba. The dependent variables were ignoring the opinions of other colleagues or healthcare professionals, and criticizing fellow physicians or other healthcare professionals in front of patients. The independent variables included age, gender, specialty, and indicators adapted from the questionnaire developed by Kwon HJ et al. A multilayer perceptron neural network was implemented. Results: The model achieved an overall accuracy of 71.20% in training and 69.40% in testing. The area under the curve (AUC) values were close to 0.75 for the categories “Ignoring the opinions of colleagues” and “Criticizing colleagues in front of patients”, reflecting good model performance. The most influential variables included ethical and professional behaviors, such as discrimination toward colleagues, provision of incorrect information, and fulfillment of job responsibilities. Conclusions: The multilayer perceptron neural network was efficient for analyzing ethical and professional attitudes and behaviors in medical practice.