The dataset was sourced from ClinicalTrials.gov and processed to filter only interventional studies with industry sponsorship, ensuring the inclusion of trials with a clearly defined therapeutic focus. In the preprocessing stage, text normalization and regex-based extraction were employed specifically on intervention names, rather than titles, to capture radioisotope mentions accurately using a robust synonym mapping. This mapping standardizes various forms of isotope nomenclature, and each isotope is then evaluated based on its established therapeutic or diagnostic classification.

Verification is achieved through a dual approach: a correctness check that randomly samples studies to display key details (e.g., NCT IDs, phases, interventions, and sponsors) for manual review, and a comparison process where aggregated statistics are matched against a pre-defined manual pipeline.

Therapeutic intent is further verified by cross-referencing recognized clinical trial phases, scanning for an extensive list of therapeutic-related keywords in both titles and intervention details, and subsequently flagging studies that fulfill any of these criteria. The pipeline also aggregates statistical summaries (such as the distribution of trial phases, isotope frequency, and sponsor counts), and visualization data is built to stratify the results by phase, company, and clinical application.

Although potential challenges remain, such as deviations in input formatting, ambiguous sponsor identifiers, or limitations in the static keyword and synonym lists the comprehensive verification steps embedded in the pipeline offer a robust mechanism for validating data extraction and categorization.