Criteria Weighting

The core mechanism that assigns a numerical coefficient to each criterion based on its clinical criticality. Hard constraints (e.g., life-threatening exclusion) receive infinite or maximum weight, while soft preferences (e.g., age range) receive lower scores.

• Weight scale: Typically 0.0 to 1.0 or 1 to 10
• Hard gates: Boolean pass/fail for non-negotiable criteria
• Soft scoring: Continuous values for preferred but flexible requirements
• Example: A trial requiring a specific EGFR mutation assigns weight 1.0 to that biomarker, but only 0.3 to a preferred prior therapy history

The process of determining weight values through collaboration between clinical scientists and data architects. Protocol-defined primary endpoints receive the highest weights, while exploratory or logistical criteria are deprioritized.

• Safety exclusions: Always maximum weight (e.g., severe organ impairment)
• Efficacy indicators: High weight for proven predictive biomarkers
• Demographic preferences: Low weight for age, sex distribution targets
• Logistical factors: Minimal weight for geographic proximity or visit frequency

The mathematical function that combines individual weighted criterion scores into a single patient-trial match score. Common approaches include weighted sum, weighted product, and hierarchical cascade models.

• Weighted sum: Σ(weight_i × score_i) — simple and interpretable
• Weighted product: Π(score_i^weight_i) — penalizes zero scores more severely
• Cascade model: Sequential evaluation where failing a high-weight criterion halts further scoring
• Normalization: Final scores are typically normalized to a 0–100 scale for cross-trial comparison

The adaptive mechanism that adjusts criterion weights during active recruitment based on real-world screening data. If a criterion proves to be a bottleneck — eliminating too many otherwise qualified patients — its weight may be reduced after protocol amendment review.

• Screen failure analysis: Identifies criteria causing disproportionate exclusions
• Weight drift detection: Monitors for unintended scoring biases over time
• Amendment integration: Automatically updates weights when protocols are modified
• Cohort balancing: Adjusts weights to ensure diverse demographic representation

The transparency layer that records and justifies every weight assignment for regulatory compliance and investigator review. Each weight decision must be traceable to a specific protocol section or clinical rationale.

• Weight provenance: Links each coefficient to its source justification
• Audit logs: Immutable record of all weight changes and approvals
• Investigator dashboards: Visual breakdown of why a patient scored highly
• Regulatory alignment: Supports FDA and EMA expectations for algorithmic transparency in trial recruitment

The application of weighted scores to classify patients into actionable tiers for recruitment coordinators. Tier 1 represents high-confidence matches requiring immediate contact, while Tier 3 indicates marginal candidates for later review.

• Tier 1 (≥85): Automatic notification to site coordinators
• Tier 2 (70–84): Flagged for manual review by clinical staff
• Tier 3 (50–69): Deferred for potential protocol amendment consideration
• Below 50: Archived; not presented to investigators to reduce alert fatigue