Statistical Result Interpretation Visualization

The Evidence Agent completes a Cox proportional-hazards analysis comparing Treatment A versus placebo for all-cause mortality in a 12,000-patient EHR cohort. The model output object carries HR = 0.75 with 95% CI [0.63, 0.89]. The…

The table cell displays HR = 0.75. The prose states that Treatment A was associated with a 25% lower hazard of all-cause mortality compared to placebo — i.e., (1 − 0.75) × 100 = 25%. The numeric percent reduction implied by the prose is mathematically consistent with the HR shown in the table. No o…

An Evidence Agent runs a multi-endpoint Cox model returning a structured results object with three endpoint-specific HRs for SGLT2 inhibitors versus DPP-4 inhibitors in a heart failure cohort: all-cause mortality (HR = 0.75, fiel…

The agent selects the 'all_cause_mortality' endpoint (HR = 0.75) and reports it as the answer to the all-cause mortality question. The output explicitly names the endpoint as 'all-cause mortality' and cites HR = 0.75. HR = 0.62 and HR = 0.91 are not presented as the primary answer; if mentioned, th…

An Evidence Agent session previously computed and cached HR = 0.62 for Drug A versus Drug B (all-cause mortality, hypertension cohort). The user now submits a new query for Drug C versus Drug D in the same cohort and endpoint. Du…

The agent triggers a fresh model computation for Drug C versus Drug D. The returned HR is 1.15 (or the correct value for that comparison — not 0.62), labeled with Drug C as exposure and Drug D as reference. The response includes provenance indicating a new computation was performed (e.g., run times…