Photo Credit: Jomkwan
The following is a summary of “Risk Classification for Interstitial Cystitis/Bladder Pain Syndrome (IC/BPS) Using Machine Learning Based Predictions,” published in the April 2024 issue of Urology by Lamb et al.
To enhance the diagnosis of Interstitial Cystitis/Bladder Pain Syndrome (IC/BPS), the researchers present a novel approach utilizing machine learning algorithms to develop an improved risk classification system for IC. Leveraging a national crowdsourcing effort, the investigators amassed a dataset comprising 1,264 urine samples, including 536 from individuals with IC (513 female, 21 male, 2 unspecified) and 728 age-matched controls (318 female, 402 male, 8 unspecified), each accompanied by corresponding Patient-Reported Outcome (PRO) pain and symptom scores.
Additionally, 296 urine samples were collected from three academic centers, comprising 78 IC cases (71 female, 7 male) and 218 controls (148 female, 68 male, 2 unspecified). Urinary cytokine biomarker levels were quantified using Luminex assay. The study group developed a machine learning predictive classification model, the Interstitial Cystitis Personalized Inflammation Symptom (IC-PIS) Score, which integrates PROs and cytokine levels, and compared its performance to a challenger model. The results indicate that the top-performing model utilizing biomarker measurements and PROs achieved an AUC of 0.87, outperforming PROs alone (AUC=0.83).
Although biomarkers alone exhibited modest predictive performance (AUC=0.58), their combination with PROs significantly enhanced predictive accuracy. The IC-PIS model thus represents a promising advancement in IC/BPS diagnosis, offering a comprehensive approach that integrates both patient-reported symptoms and objective biomarkers. Moreover, the study underscores the robustness and scalability of the findings through innovative sample collection logistics, including one of the largest crowdsourced biomarker development studies utilizing ambient shipping methods across the United States.
Source: sciencedirect.com/science/article/abs/pii/S0090429524002851