Cleveland Clinic urologists show how machine learning could revolutionize overactive bladder care

Machine learning isn’t new to medicine or to urology, but its potential remains largely untapped, according to the authors of two new Cleveland Clinic–led studies.

Urologists at Cleveland Clinic have found a new application for machine learning within their field, and they’re using it to improve shared decision making in the treatment of a common urologic diagnosis: overactive bladder (OAB).

Machine learning isn’t new to medicine or to urology, but its potential remains largely untapped, according to the authors of two new Cleveland Clinic-led studies.

“Algorithms can “learn” data patterns and trends and make inferences about the relationship between input and output data and, with this knowledge, make new predictions,” explains Glenn Werneburg, MD, PhD, first author on both of the studies and a fellow in the Department of Urology.

“In the context of our work, these predictions can help inform patient decision making about the effectiveness of a particular therapy,” says Dr. Werneburg.

OAB as a diagnostic target

Many patients with OAB will respond to behavioral approaches or oral medications, but for those who don’t, third-line options, bladder injection of onabotulinumtoxinA (OBTX‐A) and sacral neuromodulation (SNM), are two similarly effective therapies. Since those treatments are invasive, it’s important to know in advance which individual patients are more likely to respond and, if so, to which one.

OAB is common and costly, and improved clinical management of OAB is needed. The condition currently affects about 16% adults of the United States population, and its global expenditure is expected to increase in coming years.

A novel approach with robust dataset

Dr. Werneburg, senior author Sandip Vasavada, MD, and coauthor Howard Goldman, MD, both female pelvic medicine and reconstructive surgeons in the Department of Urology, developed neural networks using a series of novel approaches. They then applied the networks to the prediction problem in OAB using datasets from the ROSETTA study.

The ROSETTA study, which was sponsored by the National Institute of Child Health and Development, is one of the most complete datasets in the field. The open-label, randomized trial included 381 women with refractory urgency incontinence across nine different U.S. centers to compare OBTX-A and SNM. These findings were published in JAMA in 2016.

Machine learning studies explained

The first study showed that the algorithms were extremely accurate in predicting treatment responses to both modalities; they correctly predicted who was a responder and a nonresponder about 90% of the time. In fact, the algorithms generally outperformed human experts in predictions. The study was published in Neurourology and Urodynamics.

In the design of the second study, Dr. Werneburg explains, blinded expert urologists were given the same training data, and this time tasked to predict patient-reported outcomes.

The top algorithms showed excellent predictive accuracy for patient-reported bladder function improvement for both OBTX-A and were noninferior to expert urologists. The algorithms were also highly accurate in predicting patient-reported bladder leakage improvement for both modalities and were noninferior to experts.

They presented the findings in May 2022 at the American Urological Association meeting, and the study was later published in the International Urogynecology Journal.

Taking a prospective approach

Plans for a future prospective analysis are underway, wherein the study design will begin with the clinician, and not the algorithm. The clinician will counsel the patient on the options, determine what is important to them, and direct their questions based on the clinical picture and the patient’s history.

They also plan to test if these data can be extrapolated to males with OAB, as their current dataset only includes women.

Machine learning complements, not replaces, clinical judgement

Despite their accuracy, the algorithms won’t replace clinician expertise, says Dr. Werneburg. “Some aspects of the physician‐patient interaction are subtle and uncomputable, and thus machine learning may complement, but not replace, a physician’s judgment.”

Related Videos
Related Content
© 2023 MJH Life Sciences

All rights reserved.