Making antibody-identification reasoning visible.
Serologic is a clinical decision-support project exploring how explicit, source-grounded rules can help interpret red-cell antibody investigations. The goal is to show which conclusions the evidence supports, what remains unresolved, and what additional testing could clarify the result.
Research and educational prototype. Not validated for clinical use.
More than a pattern that fits.
An antibody investigation involves more than finding a pattern that fits. Interpretation also depends on the evidence available, the methods used, the alternatives that have been adequately assessed, and the findings that remain unexplained.
A useful decision-support tool should make those distinctions visible and open to review. Serologic is being developed to preserve that reasoning: a reproducible account of how observations lead to conclusions, with uncertainty and supporting sources intact.
The questions Serologic is designed to address
- 01
What does the evidence support?
What antibody or combination of antibodies is supported by the available evidence?
- 02
What remains possible?
Which possibilities are adequately excluded, and which remain unresolved?
- 03
What is still unexplained?
Does the proposed interpretation explain the reactions, and is the investigation complete?
- 04
What would help next?
What additional testing or expert review could help resolve the uncertainty?
The conclusion.
Then the reasoning.
The intended output connects conclusions to individual observations and the rules applied to them.
This illustrates the intended presentation, not a patient case or an engine result.
A concise account of what the evidence establishes—and what it does not.
Inspect the evidence
- Supporting observations
- The findings and rules behind a conclusion.
- Contradictory findings
- Evidence that does not fit the proposed explanation.
- Unresolved alternatives
- Possibilities the available evidence has not settled.
Keep two questions visible
Pattern explanation: does the interpretation account for the observed reactions?
Investigation completeness: have the required alternatives and remaining findings been adequately assessed?
Explaining the observed reactions and completing the investigation are separate questions. Serologic is designed to report both.
Explicit rules.
Human review.
AI may help turn a worksheet image into structured observations or explain an engine result in accessible language. The underlying antibody interpretation comes from explicit, versioned rules.
Extracted observations require human verification. Computed inference remains distinct from a reviewer’s accepted or reasoned-overridden interpretation.
A working research engine.
A developing public project.
A working research reference engine exists, and development is continuing. The initial focus is routine red-cell alloantibody interpretation; recognizing insufficient evidence and the need for expert escalation are part of the intended behavior.
This site currently explains the approach. A verified worked example is the next step toward a public demonstration. Reviewed photo input requires additional implementation and testing.
A project led by Justin Halls, MD, combining transfusion medicine and clinical informatics.
Illustration credits
Red-cell illustrations by Sarbasst Braian, via Wikimedia Commons: BloodCellState 001 and BloodCellState 159. Released under CC0 1.0. Illustrations, not microscopy or case evidence.
