Urology AI Assistant

What it solves

Urology workflows often split across records, imaging reports, guideline lookups, medication decisions, and research statistics. Urology AI Assistant is positioned as a structured workflow aid rather than a replacement for clinical judgment, reducing the need to switch between templates, literature tools, and analysis scripts.

How it works

The skill organizes work into two main areas:

• Clinical support: generate admission notes, operative notes, discharge summaries, and progress notes; structure findings from CTU, mpMRI, and ultrasound reports using frameworks such as PI-RADS v2.1 and R.E.N.A.L.; support planning references with TNM staging, Partin tables, and procedure-specific risk templates; search EAU, AUA, NCCN, and CUA guidelines; assist with antibiotic selection, postoperative analgesia, oncologic treatment options, and renal dose adjustments.
• Research support: search PubMed via API, frame questions with PICO, appraise RCTs with Jadad scoring, and run descriptive statistics, t-tests, chi-square tests, survival analysis, Cox regression, linear/logistic regression, and ROC curves.

Technically it relies on Python 3.8+ with libraries such as numpy, pandas, scipy, statsmodels, lifelines, matplotlib, and seaborn, and prefers JSON for standardized input and output, which makes it easier to integrate into existing data pipelines.

Boundaries and caveats

The output is primarily template-based, retrieval-based, and computational assistance, not a diagnosis, final surgical plan, or individualized order set. Imaging scores, staging, medication recommendations, and statistical results should be reviewed by clinicians against primary records and clinical context. Guideline results need version and scope verification, and PubMed results depend heavily on search terms.