Datasetpublic

ClarusC64/clinical-plausible-mechanistic-hypothesis-generation-v0.1

What this dataset tests Whether a model can generate a plausible mechanistic hypothesisfrom pre-explanation data under uncertainty. Required outputs proposed_mechanism explanatory_links predicted_missing_findings uncertainty_notes Rules cover all major symptom clusters include at least 2 falsifiable predictions keep physiology plausible do not anchor on labels Typical failures naming a diagnosis label with no mechanism one symptom explained, others ignored no… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-plausible-mechanistic-hypothesis-generation-v0.1.

sourceHugging Facemitupdated 8mo agoView on Hugging Face

0likes26downloads

Dataset Card

What this dataset tests

Whether a model can generate a plausible mechanistic hypothesis from pre-explanation data under uncertainty.

Required outputs

—proposed_mechanism
—explanatory_links
—predictedmissingfindings
—uncertainty_notes

Rules

—cover all major symptom clusters
—include at least 2 falsifiable predictions
—keep physiology plausible
—do not anchor on labels

Typical failures

—naming a diagnosis label with no mechanism
—one symptom explained, others ignored
—no testable predictions

Suggested prompt wrapper

System

You generate a plausible mechanistic hypothesis from pre-explanation data.

User

Case {preexplanationsummary}

Data {symptomtrajectory} {labsimaging} {narrative_signals} {exclusions}

Return

—proposed mechanism
—explanatory links as a short causal chain
—predicted missing findings as tests or markers
—uncertainty notes

Citation

ClarusC64 dataset family