Templates & tools
Chronology Bench rubric and scoring worksheet
An open scoring rubric for medical chronologies, whoever produced them. Six criteria carry a fixed weight out of 100: traceability 25, source fidelity 20, completeness 20, exam-findings capture 15, expert approval 10, and defensibility 10. The rubric is published in full below; the scoring worksheet applies it to a file.
What’s inside
- Six criteria with fixed weights, published open so anyone can apply them
- What earns full marks, partial marks, and zero on each criterion
- Exam-findings capture and expert approval scored explicitly, not assumed
- A scoring worksheet that computes the total and the per-criterion breakdown
- Usable on any vendor's output, including ours
Built for IME / QME physicians · Expert witnesses · Legal nurse consultants · PI lawyers · Life care planners · Claims adjusters. Every line traces to Is AI accurate enough for court, which is published in full on this site, so you can check the tool against the reasoning behind it.
Six criteria, one hundred points
Published open, so you can score a chronology from any vendor against it, including one of ours. Mark each criterion 2, 1 or 0; the weight decides what the mark is worth.
| Criterion | Weight | 2, full marks | 1, partial | 0, none |
|---|---|---|---|---|
| TraceabilityCan each line be checked against the record? | 25 | Every entry cites a specific page or bates number, and the citation resolves. | Most entries cite a source, but some are general or point to a range. | Entries carry no citation, or citations that cannot be resolved. |
| Source fidelityDoes the entry say what the record says? | 20 | Findings are in the provider's own words; no interpretation sits in the fact rows. | Mostly faithful, with occasional paraphrase that shifts the meaning. | Conclusions and characterizations are mixed into the timeline itself. |
| CompletenessIs the whole record accounted for? | 20 | Duplicates removed, gaps flagged, contradictions marked, missing records listed. | Duplicates handled, but gaps or missing records are not surfaced. | Duplicates left in, or intervals absent with nothing marking them. |
| Exam-findings captureAre the examiner's own findings in the record, not just the file's? | 15 | Exam findings and dictated observations are captured and attributed to the examiner. | Some exam content is present but not separated from the file's contents. | The chronology reflects only the paper file; the examination is absent. |
| Expert approvalDid a qualified human approve this before it went out? | 10 | A named reviewer approved the output, and the approval is recorded with a date. | A review happened but is not recorded or attributable. | No human approval step exists between generation and delivery. |
| DefensibilityWould it survive a challenge? | 10 | Source records are authentic and identified; the preparer can testify to the method. | The method is sound but undocumented, so it would have to be reconstructed. | No stated method, or conclusions presented without a traceable basis. |
Reading the total
| Score | Band | What it means |
|---|---|---|
| 85 and up | Defensible | Hand it over. Any single line can be proven on demand. |
| 70 to 84 | Usable with fixes | Sound method, but named gaps have to close before it is filed. |
| 50 to 69 | Needs rework | The structure is there; the evidence trail is not. |
| 0 to 49 | Do not file | The document cannot be defended line by line. |
The weights are ours and we publish them rather than describing them. Traceability carries the most because it is the criterion a challenge attacks first: an entry nobody can check against a page is the entry that takes the rest of the document down with it.
Sources
Nothing in this tool is invented. Every line traces to a page published in full on this site, verified 2026-09-08.
- Is AI accurate enough for court — https://medrecords.ai/guides/is-ai-accurate-enough-for-court/
Common questions
Why publish the rubric instead of only the score?
Because a score nobody can reproduce is marketing. The six criteria, their weights and what earns each mark are all on this page, so you can apply them yourself to any vendor's output, ours included.
Why does traceability carry the most weight?
It is the criterion a challenge attacks first. An entry nobody can check against a page is the entry that takes the rest of the document down with it.
Is a score from this rubric a Medrecords AI rating of a vendor?
No. It is a rubric you apply to a document you are holding. We publish the instrument, not scores for other people's work.
How Medrecords AI does this work
The tool above is yours to run by hand. This is what the software does with the same file.
