Hypothesis-Based Phenotype Algorithms & Trigger-Provoked Latent-Disease EHR Case-Finding

In progress METHODOLOGYEHRPHENOTYPE_ALGORITHMNAMCHANNELOPATHY
Home / Projects / Hypothesis-Based Phenotype Algorithms & Trigger-Provoked Latent-Disease EHR Case-Finding

Hypothesis-Based Phenotype Algorithms & Trigger-Provoked Latent-Disease EHR Case-Finding

Motivation

A zebrafish Timothy-syndrome study (#6245, PMID:42426269) showed that fever elicits arrhythmia and seizures even in overtly normal CACNA1C heterozygotes. That implies a computable idea: scan EHRs for a new arrhythmia or seizure shortly after a fever, to surface latent/mild carriers a classical case definition would never catch. But such a query is valid only if the mechanism holds — so running it is simultaneously case-finding and a test of the mechanism. A standard PheKB/OHDSI phenotype library cannot say which of its algorithms is speculative; DisMech can, because it links the algorithm to the mechanism.

The archetype: trigger-provoked latent disease

A physiological / pharmacological / procedural trigger transiently unmasks a latent channel/enzyme defect. Most carriers are asymptomatic at baseline. An EHR query for trigger-associated events is enriched for latent carriers, and that enrichment tests the mechanism.

What was built

  1. Schema (enacted, @cmungall-approved) — Definition gained derivation_basis (ESTABLISHED_CRITERIA / MECHANISTIC_HYPOTHESIS / MODEL_SYSTEM_EXTRAPOLATION), a structured validation_status object (status + rationale + evidence), and reuses attaches_to to link the pathograph node(s) a hypothesis-based algorithm is predicated on. Gated by check_hypothesis_based_definition_attaches_to_foreign_keys. See the design write-up.
  2. Worked examples — each carries a PHENOTYPE_ALGORITHM case-finding definition and a NAM confirmation experiment (below).
  3. Candidate register + validation guidance — see the survey report.

Worked examples

Disease Trigger Trigger type System derivation_basis Genotype yield of a hit
Timothy Syndrome fever physiological cardiac + CNS MECHANISTIC_HYPOTHESIS (unknown — hypothesis)
Brugada Syndrome fever physiological cardiac ESTABLISHED_CRITERIA ~26% SCN5A
Long QT Syndrome QT-prolonging drug pharmacological cardiac ESTABLISHED_CRITERIA ~10–15%
Malignant Hyperthermia of Anesthesia volatile anesthetic / succinylcholine procedural skeletal muscle ESTABLISHED_CRITERIA 50–>70%

Timothy Syndrome is the one genuine MECHANISTIC_HYPOTHESIS (zebrafish-derived, human validity open); the other three are established mechanisms where the EHR operationalization is the new artifact. Together they span emerging↔established grounding × physiological/pharmacological/procedural triggers × cardiac/muscle.

Two-armed test strategy: population × mechanistic

Each hypothesis has two complementary tests, both first-class in the schema:

Disease Mechanistic NAM experiment
Timothy Syndrome iPSC-CM + iPSC-neuron + heart-on-chip + in-silico hyperthermia challenge (exp_ts_fever_ipsc_nam_challenge)
Brugada Syndrome iPSC-CM + heterologous Nav1.5 temperature ramp (exp_brs_nav15_temperature_nam)
Long QT Syndrome CiPA (in-vitro hERG/multi-channel + in-silico O'Hara-Rudy/ToR-ORd) + iPSC-CM drug challenge (exp_lqt_cipa_nam_confirmation)
Malignant Hyperthermia of Anesthesia RyR1 Ca²⁺-release + iPSC-myotube caffeine/4-CmC/halothane challenge, an in-vitro complement to the invasive CHCT (exp_mh_ryr1_ipsc_calcium_nam)

This connects to two adjacent projects: NAMO_RD_MODELS (the NAMO bridge) and MONDO_EHR_MAPPINGS (MONDO-driven OMOP concept sets the population arm would use).

Status & worklist