Fixed checks
The same supplied records produce the same review findings. Each finding points a reviewer to the evidence that needs attention.
Methodology and validation
WardTrace is evaluated on labeled synthetic ground truth today. Those results are not field performance. A calibration pilot on the customer's own data is required before field claims are made.
Labeled test cases
Synthetic cases include by-construction labels so false positives and missed findings can be measured during development.
Customer pilot
Production use begins with a governed pilot that measures performance, review burden, and data quality in the customer's environment.
What is shared publicly
We publish the validation design, the synthetic-data limitation, the role of human review, and the boundaries of the product's claims. We do not publish detector names, thresholds, signatures, tuning, per-check performance, source identifiers, or machine-readable evidence bundles.
What a qualified evaluation includes
- A documented data scope and review protocol.
- Performance measured against an agreed reference standard.
- False-positive and missed-finding analysis.
- Human-review workload and escalation outcomes.
- Known limitations and unresolved data-quality gaps.
What WardTrace does not establish
- Medical necessity. WardTrace reviews whether supplied records agree. It does not decide whether care was clinically warranted.
- Clinical accuracy. WardTrace does not determine whether a measurement or chart statement is medically correct.
- Fraud. A finding routes evidence to a human. It is not an automatic accusation or denial.
Detailed technical materials are shared with qualified evaluators under appropriate confidentiality and legal review.