[New Skill]: Clinical Trial NLP Matcher

[rosspeili]

Skill Name

bioinformatics/clinical_trial_matcher

What should this skill do?

Ideal for CERTH INAB's Digital Health & eCAN Plus Initiatives

Extracting inclusion/exclusion criteria from clinical trial PDFs and matching them to anonymized patient profiles requires high-level reasoning. This skill accepts an anonymized health JSON payload and a specific ClinicalTrials.gov ID. It evaluates the boolean logic of the trial's requirements and returns a definitive match score.

Contributors: Privacy is paramount. The manifest.yaml constitution must strictly enforce that only anonymized hashes or stripped parameters are processed. The test_skill.py must include mock patient permutations ensuring zero leaked PII.

Ideal Inputs & Outputs

Input:
{
"trial_nct_id": "NCT01234567",
"anonymized_patient_biomarkers": {"HER2": "positive", "age": 45, "prior_therapies": 1}
}

Output:
{
"match_status": "eligible",
"eligibility_score": 1.0,
"flagged_criteria": []
}

Targeted Models (if applicable)

Model Agnostic (All)