Machine Learning-Based Risk Classification for Vascular Damage in Erectile Dysfunction with Ultrasound Analysis

Penile dynamic color Doppler duplex ultrasound (CDDU) has a debated role in diagnosing erectile dysfunction (ED), particularly vasculogenic ED, which is closely linked to a heightened risk of future cardiovascular (CV) events. This study sought to establish a machine learning-based vascular risk classification for patients diagnosed with ED. CDDU data from 301 men with ED were gathered prospectively, including baseline demographics such as age, body mass index (BMI), cardiovascular risk factors, smoking, and alcohol habits. Participants also completed the International Index of Erectile Function (IIEF-EF) questionnaire. Vasculogenic ED (V-ED) was identified by a peak systolic velocity under 35 cm/s and a resistance index below 0.8. Risk classifications for V-ED were created using a Chi-square Automatic Interaction Detector (CHAID), a machine learning algorithm for risk stratification, while ROC curve analysis confirmed classification accuracy.

Five risk groups emerged from this model: 1) age <53 years and BMI <25 kg/m², 2) age <53 years, BMI >25 kg/m², and non-smoker, 3) age ≥53 years and non-smoker, 4) age <53 years, BMI >25 kg/m², and smoker, and 5) age >53 years and smoker. These groups were classified from “very low” to “very high” risk for V-ED. In a validation cohort of 127 men treated for ED, both “high-risk” and “very high-risk” groups had significantly poorer ED treatment response rates. The “very high-risk” group also demonstrated a notably elevated risk of major CV events, with a 30% likelihood of developing cardiovascular disease within ten years. This machine learning model enables targeted CDDU use in high-risk ED patients, identifying those who may benefit from early cardiovascular risk interventions.

Machine Learning-Based Risk Classification for Vascular Damage in Erectile Dysfu…

by Echo Writer time to read: 1 min
0

Contact Support

If you're interested in posting an article and need assistance, please don't hesitate to contact our support team. We're here to help you through the process, answer any questions you may have, and ensure that your article is published smoothly and effectively.

support@ehealthcommunity.org