Objectives: To develop a risk prediction model for the identification of features involved in the prediction of erectile dysfunction (ED) at 12 months following COVID-19 recovery. Methods: We performed an observational prospective multicentre study. Participants were classified according to their history of COVID-19: I) Patients with a past history of COVID-19; and II) Patients without a previous microbiological diagnosis of COVID-19. A total of 361 patients (Past history of COVID-19, n=166; No past history of COVID-19, n=195) were assessed from January 2022 to March 2023. Patients with a past history of COVID-19 were assessed at 12 months following COVID-19 recovery. The primary outcome measure was ED, assessed through the 5-item International Index of Erectile Function (IIEF-5). Data concerning epidemiologic variables, comorbidities and active treatment were also collected. We performed a binary logistic regression to develop a risk predictive model. Among the models developed, we selected the one with the higher Area Under the Curve (AUC). Results: The median age was 55 years in both groups. The ED prevalence was 55.9% in patients with past history of COVID-19 and 44.1% in those with no past history of COVID-19. The best predictive model developed for ED comprised 40 variables and had an AUC of 0.8. Conclusions: We developed a regression model for the prediction of ED 12 months after COVID-19 recovery. The application of our predictive tool in a community setting could eventually prevent the adverse effects of ED on cardiovascular health, and the associated unfavourable economic impact.

Objectives: To develop a machine learning-based model for the identification of features involved in the prediction of long-term erectile dysfunction (ED) after COVID-19 recovery. Methods: We performed an observational prospective multicentre study. Participants were divided into 2 groups: I) Patients with a past history of COVID-19; II) Patients without a previous microbiological diagnosis of COVID-19. A total of 361 patients (COVID-19 group, n=166, Control group; n=195) were assessed from January 2022 to March 2023. COVID-19 group patients were assessed 12 months after COVID-19 recovery; Control group patients were assessed within the same time window. The primary outcome measure was ED. A machine learning-based model including three algorithms (random forest, logistic regression and support vector machines [SVM]) was applied to perform the automatic selection of those features involved in the prediction of ED. The algorithm SVM-recursive feature elimination (RFE) was also executed. Results: The median age was 55 years in both groups. The final selection of variables in our model was: group, SARS-CoV-2 vaccination status, hypertension, diabetes, beta blocker treatment, antiplatelet therapy and coffee intake. The algorithm SVM-RFE showed that a past history of COVID-19 and SARS-CoV-2 vaccination were both long-term ED predictors in our model. Conclusions: We developed a machine learning-based model for the prediction of long-term ED following COVID-19 recovery. A past history of COVID-19 and SARS-CoV-2 vaccination were both predictors in our model. The application of our predictive tool in a community setting could avoid the adverse effects of ED and the unfavourable economic impact associated.