Presentation (CTI)

OBJECTIVES: To develop an automated decision-support tool for early feasibility assessment and method selection for indirect treatment comparisons (ITCs), and demonstrate its application in a realistic scenario.
METHODS: We developed a rule‑based feasibility assessment tool implemented in R Shiny to evaluate suitability for pairwise meta-analysis, network meta-analysis (NMA), meta-regression, multilevel network meta-regression (ML-NMR), population-adjusted indirect comparison (PAIC), and external control arm (ECA) analyses. The tool applies automated logic to structured systematic literature review outputs to assess evidence connectivity, alignment of treatments, comparators, outcomes, and populations across PICOS elements, and method-specific data requirements such as availability of individual-level data, covariate overlap and network structures. Users can flexibly refine analytical scope and validate assumptions related to treatment alignment and exchangeability. Automated outputs are explicitly designed to support expert decisions to confirm clinical plausibility and appropriate interpretation.
RESULTS: The platform is implemented as an interactive dashboard and an automatically generated, decision support-focused slide deck summarizing feasibility assessments across ITC methods and recommended analytical options. In an illustrative application evaluating progression-free survival for everolimus in combination with hormonal therapy versus chemotherapy in advanced breast cancer, as reported in the Cope framework, the tool was used to identify feasible analytical paths while highlighting challenges with treatment alignment and exchangeability that required expert validation.
CONCLUSIONS: This automated feasibility tool enables efficient, transparent, and reproducible early assessment of the suitability of ITC methods. It improves consistency and planning efficiency, producing high-quality, standardized outputs quickly with reduced manual intervention, particularly useful in evolving data landscapes and under tight turnaround timelines such as those required for Joint Clinical Assessments. However, structured expert validation remains essential, particularly for assessing treatment alignment and exchangeability, which cannot be fully resolved algorithmically.