bias-detection

What problem does it solve?

This Skill systematically identifies and assesses methodological biases in scientific hypotheses, acting as a crucial "code smell" detector for study designs to ensure robust and reliable conclusions.

Core Features & Use Cases

• Comprehensive Bias Taxonomy: Covers a wide range of biases including time-zero, censoring, selection, confounding, and information biases.
• Type-Specific Emphasis: Tailors bias assessment based on the specific type of study (e.g., RCTs, observational studies).
• Baseline Characteristics Analysis: Mandates detailed examination of baseline data to check for imbalances and assess confounding.
• Use Case: When reviewing a new clinical trial hypothesis, this Skill will automatically scan for potential biases like immortal time bias or confounding by indication, providing a structured assessment of their applicability, severity, and potential mitigation strategies.

Quick Start

Use the bias-detection skill to assess the hypothesis in ./parsed_hypothesis.json and write the findings to ./bias_assessment.json.