Patient Thread Insight · transplant software

A virtual crossmatch that knows when not to answer.

Software for histocompatibility and immunogenetics laboratories. It brings the recipient's antibody results, sensitisation history and HLA type together with the donor's typing, gives one of three calls with every reason shown, and returns a result signed by two scientists to your laboratory system.

Research use only · not for use in diagnostic procedures

Watch the demonstration Become a research partner

Captions

Demonstration, 3 minutes 19 seconds, narrated. A synthetic deceased-donor case from raw inputs to a signed HL7 result. All patient data, bead values and thresholds are synthetic.

What it does

Every input, every rule, and the reason for every call.

A bead panel reports antibody in one serum on one day. A safe virtual crossmatch has to know what the beads cannot see, so the rules that override them are built in, and the software says plainly when it cannot answer.

Inputs

Any source, one record

HL7 from the laboratory information system, single-antigen bead exports, HLA typing files, manual entry, and scanned paper reports read on the device itself, so no patient data leaves the building. Every value keeps its source.

Checks

Anything that looks wrong is flagged

Identity mismatches between sources, alleles that do not exist, a typing that differs from the record, implausible or copied values, sudden changes between sera. Nothing is released until a named person has looked.

Human review

Four eyes on what people type

Scanned values are confirmed against the original. Manual entries are verified by a second person, and the software will not let anyone verify their own entry.

History

Rules that override the beads

A sensitising event within three months, or after the serum was drawn, voids the virtual crossmatch. So do pregnancy-related donor relationships and repeat mismatches, even with a negative antibody screen.

Every serum

An earlier positive is not a negative

All sera on file are assessed against the donor. If an earlier serum would have been positive and today's is clean, a physical crossmatch is required. HLA and non-HLA antibodies alike.

Uncertainty

Honest at the edges

Values near a cut-off are treated as uncertain using the laboratory's own measurement variability. Untyped loci, ambiguous typing and alleles that may be null are named, never assumed away.

Three calls, worst first

One answer, and everything behind it.

Predicted positive

Donor-specific antibody at or above the laboratory's threshold, or a listed unacceptable antigen.

Physical crossmatch required

Anything that makes the virtual result unreliable, with each reason listed.

Predicted negative

None of the above. Whether to proceed is the transplant unit's decision, under its own protocol.

Thresholds are set by each laboratory from its own validation. Until they are set, the software will not call any crossmatch negative. Results are released only when two different authorised scientists have signed the same content, then returned as an HL7 v2.5.1 message.

How it is checked

Verified as software. Not yet validated as a test.

94automated tests covering every rule, the safety-critical cases and the interfaces

2,000synthetic donor and recipient pairs, drawn from real population HLA frequencies, run through the engine

0violations of the safety properties tested, such as a stronger antibody or a new sensitising event never making the call less cautious

2 wayscalculated reaction frequency computed by two independent methods, which agree

What this does and does not show These checks show that the software does what its design says. They do not show clinical accuracy. Performance against physical crossmatch has not yet been established, and that is the purpose of the research evaluation below. Interfaces are standards-based and are configured and validated with each laboratory's own systems.

Research partners

We are looking for H&I laboratories to evaluate it with us.

A research evaluation compares the software's calls with physical crossmatch results on real sera, with each laboratory's own thresholds. If your laboratory would like to take part, we would like to hear from you.

Research use only. Patient Thread Insight is not for use in diagnostic procedures, is not UKCA or CE marked, and has not been validated against physical crossmatch. The demonstration uses synthetic data and thresholds.

Patient Thread Insight is created and owned by Patient Thread Ltd and licensed to Scira Diagnostics Ltd, the laboratory and assay house it is developed with.