Using Artificial Intelligence to Predict Surgical Shunts in Men with Ischemic Priapism.

Abstract

PURPOSE:

Ischemic priapism is a urological emergency that requires prompt intervention to preserve erectile function. Characteristics that influence escalation to surgical intervention remain unclear. We identified factors and developed machine learning models to predict which men presenting with ischemic priapism will require shunting.

MATERIALS AND METHODS:

We identified men with ischemic priapism admitted to the emergency department of our large county hospital between January 2010 and June 2019. We collected patient demographics, etiology, duration of priapism prior to intervention, interventions attempted and escalation to shunting. Machine learning models were trained and tested using R to predict which patients require surgical shunting.

RESULTS:

A total of 334 encounters of ischemic priapism were identified. The majority resolved with intracavernosal phenylephrine injection and/or cavernous aspiration (78%). Shunting was required in 10% of men. Median duration of priapism before intervention was longer for men requiring shunting than for men who did not (48 vs 7 hours, p=0.030). Patients with sickle cell disease as the etiology were less likely to require shunting compared to all other etiologies (2.2% vs 15.2%, p=0.035).

CONCLUSIONS:

Men with longer duration of priapism before treatment more often underwent shunting. However, phenylephrine injection and aspiration remained effective for priapism lasting more than 36 hours. Having sickle cell disease as the etiology of priapism was protective against requiring shunting. We developed artificial intelligence models that performed with 87.2% accuracy and created an online probability calculator to determine which patients with ischemic priapism may require shunting.

• Authors: Thomas A Masterson, Madhumita Parmar, Michael B Tradewell, Sirpi Nackeeran, Quinn Rainer, Ruben Blachman-Braun, Nicholas Heller, Aubrey Greer, Nicholas Hauser, Bruce R Kava, Ranjith Ramasamy
• Journal: The Journal of urology (J Urol) Vol. 204 Issue 5 Pg. 1033-1038 (Nov 2020) ISSN: 1527-3792 [Electronic] United States
• PMID: 32516073 (Publication Type: Journal Article)

Chemical References

• Phenylephrine

Topics

• Adolescent
• Adult
• Aged
• Aged, 80 and over
• Anastomosis, Surgical (statistics & numerical data)
• Emergency Service, Hospital (statistics & numerical data)
• Emergency Treatment (statistics & numerical data)
• Erectile Dysfunction (etiology, prevention & control)
• Humans
• Likelihood Functions
• Machine Learning
• Male
• Middle Aged
• Models, Statistical
• Paracentesis (statistics & numerical data)
• Penile Erection (drug effects, physiology)
• Penis (blood supply, drug effects, physiopathology, surgery)
• Phenylephrine (administration & dosage)
• Priapism (etiology, therapy)
• Retrospective Studies
• Risk Assessment (methods)
• Time Factors
• Treatment Outcome
• Urologic Surgical Procedures, Male (statistics & numerical data)
• Young Adult