A machine learning algorithm to predict the success of a second microsurgical testicular sperm extraction
Akef Obeidat1, Belal Nedal Sabbah1, Hammam Mandourah2
1Alfaisal University, College of Medicine, Riyadh, Saudi Arabia.
Annals of Medicine and Surgery (2012)
|September 3, 2025
Summary
Predicting success for a second testicular sperm extraction (TESE) is crucial for azoospermia patients. A machine learning model using factors like hormone levels and procedure history achieved 80% accuracy in identifying sperm retrieval success.
Area of Science:
- Reproductive Medicine
- Artificial Intelligence in Healthcare
- Urology
Background:
- Testicular sperm extraction (TESE) is vital for azoospermia patients.
- Success rates for repeat TESE procedures are often low.
- Predictive tools for second TESE (microTESE) success are needed.
Purpose of the Study:
- Develop and evaluate a machine learning algorithm.
- Predict the success of a second microsurgical TESE (microTESE).
Main Methods:
- Retrospective analysis of 47 patients undergoing second microTESE.
- Utilized supervised machine learning (Support Vector Machine - SVM).
- Included variables: procedure side, histopathology, FSH, testosterone, testicular volume, comorbidities.
Main Results:
- SVM model achieved 80% accuracy after hyperparameter tuning.
- Bilateral procedures and longer intervals between surgeries increased success.
- Cancer history negatively impacted outcomes; FSH, testosterone, histopathology, varicocele, and procedure interval were key predictors.
Conclusions:
- Machine learning accurately predicts sperm presence in second microTESE for non-obstructive azoospermia.
- SVM model shows promise incorporating clinical and hormonal factors.
- Further validation is required for broader applicability.
