Glossary
Deterministic Scoring
Updated February 22, 2026
Deterministic scoring is the repeatable ranking layer built from stable inputs and fixed rules before optional adjustments are added.
Why it matters
Understanding which factors are fixed helps you know why the same profile usually generates the same ranked output.
Deterministic vs optional signals
- Deterministic factors: active goals, core contraindications, adherence class, and required exclusions.
- Optional/experimental factors: freshness of evidence, emerging interactions, and novelty bonuses.
Reproducibility rule
When profile, goals, and medication context are unchanged, ranking should be reproducible from cycle to cycle.
A new evidence flag, missing data event, or user override is what usually changes the exact order.
Example score direction shifts
Example with three signals:
- Goal focus changes from mood support to sleep support: sleep-focused items gain ranking priority.
- A missed-dose trend rises: higher-risk stimulants lose score weight.
- A new red-flag overlap appears: the candidate stack is pushed down or suppressed.
Cross-site references
Uncertainty
- Evidence is limited on how quickly experimental factors should be blended with deterministic scores.
- Evidence is limited on reproducibility after frequent profile edits in high-noise stacks.
How this appears in Unfair
Unfair first applies deterministic filters, then layers confidence and optional modifiers before final ranking.
Clinical safety note
If your output changes in ways that do not match your stable profile, pause major changes and verify your profile data is current.