Error, or true exclusion?
Every mismatch is pre classified with a confidence and the reason, which marker, the pattern, the call quality. Analysts confirm instead of hunting marker by marker in Excel.
The science is unchanged, the same ISAG panel, the same Mendelian exclusion, the same LOD. The pilot reproduces the parentage results you get from Cervus today, then automates the manual middle: the marker by marker mismatch hunt, the macros, the certificate and the Stamboek upload.
The statistics stay deterministic and auditable. AI takes the first pass on the steps that eat analyst time, then a named reviewer confirms every call.
Every mismatch is pre classified with a confidence and the reason, which marker, the pattern, the call quality. Analysts confirm instead of hunting marker by marker in Excel.
Only the doubtful calls are surfaced for review. The analyst looks at what the model was unsure about, not every call.
In production, trained on your resolved case history. The reviewer confirms or overrides every call. The AI never decides parentage.
Same ISAG panel, same Mendelian exclusion, same LOD. We automate the tools that run it, not the genetics.
The pilot ships only once it reproduces your Cervus verdicts on your own solved cases, with a concordance report you keep.
ISO 17025, ring tests and exclusion probabilities are yours. We supply the validation evidence and preserve the audit trail.
Replacing GeneMarker's calling is next: we prove it reproduces your calls on your own traces, in parallel, before it ever ships.