reconciling-problem-lists

Condition reconciliation clusters synonymous diagnosis mentions—such as DM2 and diabetes mellitus—into unified concepts and resolves their clinical status. This process relies on negation detection, temporality axes, and synonym normalization to produce a clean, deduplicated problem list.

Negation detection identifies denied or hypothetical mentions within clinical notes and excludes them from the final problem list. By analyzing temporality, uncertainty, and section context, the system distinguishes active conditions from resolved or historical statuses.

Yes, clinical condition mentions are reconciled into USCDI Problem- and FHIR Condition-shaped records. The output includes deduplicated conditions with active, resolved, or historical statuses, mention counts, and source provenance for clinical review.

Yes, synonym normalization requires user-supplied SNOMED CT grounding or OpenMed Disease entities. This grounding maps varied clinical mentions like type 2 diabetes to unified concepts for accurate condition reconciliation.

The best way to exclude hypothetical mentions is applying negation and temporality axes alongside section context resolution. This ensures only active, resolved, or historical conditions are retained while denied symptoms are automatically filtered out.

Provenance tracking records the source mentions and counts for each reconciled condition within the clinical note. This provides clinicians with the exact evidence trail needed to review the deduplicated problem list and verify clinical status assignments.