Predicting Micro-TESE among Heterogeneous Nonobstructive Azoospermic Patients: The Impact on Surgical Decision and ICSI

Figure 3

Decision curve analysis for our prediction model. (a) Showing how many micro-TESEs would be done (red line) and how many would be successful (blue line) per 1000 patients using the model-based decision strategy at a given decision threshold. (b) The decision curve shows standardized net benefit which is defined as the expected benefit-cost. As an illustrative example, per 1000 patients with a decision threshold of 0.4 (at the purple vertical dash line), 458 would go through micro-TESE and 279 would have successful sperm retrieval. The standardized net benefit of a model-based decision for these patients is 0.388 compared with 0.289 for an etiology-based strategy (i.e., the surgical decision only considers etiology) and 0.045 for an all-micro-TESE strategy (all patients have micro-TESE). The model-based strategy has 34% more benefit than the etiology-based strategy and 8.6 times the benefit of the all-micro-TESE strategy.

(a)

(b)