Platform gap analysis · 2026-07-25

What everyone else ships that we don’t

Thirteen safety-monitoring and clinical-review platforms, two reference catalogues, and sixty-three capabilities scored against this portfolio. The chart set is in better shape than expected. The layer around it barely exists.

The short version

On chart types this portfolio is competitive and in two places ahead — the hepatic tooling for abnormal-baseline populations exists on none of the thirteen platforms, and no one else publishes requirement-keyed evidence for each display. The real hole is not a chart. It is the review layer: nine of the ten workflow capabilities surveyed are missing here, and the pattern repeats across commercial and open-source tools alike. A reviewer cannot mark a participant reviewed, cannot see what changed since last week, cannot annotate a finding, cannot track an issue to resolution, and cannot carry a filter or a selection from one display to the next.

The single most-repeated capability we lack, present on five platforms including the one open-source medical-monitoring package that competes with us directly, is “show me what changed since my last review”. It is filed nowhere — not on the hub, not in safety.viz, and not in last night’s goal-atlas requirement spike.

15
sources surveyed — 13 platforms, 2 reference catalogues
63
capabilities scored
17
we have today
7
already filed on the roadmap
37
missing or only partial, nothing filed

The headline list

Capabilities that at least four of the thirteen platforms ship, that we do not have and have not filed. Ordered by how many platforms have them. This is the answer to “common views and functionality we don’t support”.

The cut at four is for readability, not a finding: eight further capabilities sit at exactly three platforms — among them the AE volcano plot, the graded lab shift, mean-change-with-CI and the data-integrity family — and they are all in the matrix, three of them carried into the proposals. Some rows here were also reached independently from the inside: Kaplan-Meier (A17) is the goal atlas’s candidate C3, and the data-mapping surface (D1) is its A3. Where that happens the proposals page defers rather than proposing the same work twice.

Five findings

The chart migration is nearly complete; the review workflow was never started

Of the 30 chart and view types surveyed, we have 11 and have filed 5 more. Of the 10 review-workflow capabilities, we have zero and have filed zero — nine are missing and the tenth (raising EDC queries) is a deliberate no. That asymmetry is the report. Every commercial platform in the survey and two of the five open-source ones treat review state, annotation and issue tracking as core; this portfolio treats the chart as the whole product.

“What changed since I last looked” is the capability with the broadest support and the least awareness here

JReview highlights patients with new data since the last review; elluminate identifies data changes since the last review; Veeva CDB builds its automation on change detection; Empirica tracks alerts across successive data updates; and clinDataReview — open source, R, and aimed at exactly our users — ships batch-to-batch comparison. Safety review is a repeated activity over a growing dataset, and none of our displays know that.

Charts that stand alone are the exception everywhere else

Roche’s Spotfire template puts a linked listing under every page and a filter panel beside all of them; JMP broadcasts one subject selection to every open report; teal and Spotfire both let a reviewer name and restore a filter state. Our renderers filter and select individually and share nothing. hub#29 asks for charts side by side over one dataset and deliberately leaves the mechanism to its Design stage — this survey says the mechanism is the point, not a detail.

The missing chart types cluster into two families, not a long tail

Time-and-comparison (Kaplan-Meier, risk difference with intervals, exposure-adjusted rates, mean change with CI) and cross-domain context (multi-domain timeline, concomitant medications and medical history, exposure, demographics and disposition). Each family is coherent enough to be one requirement rather than six. The long tail that remains — tendril plots, ROC, scatter-plot matrices, geographic maps — is genuinely low priority and is marked as such in the matrix.

Two things here are ahead of the field, and both are easy to undersell

The abnormal-baseline DILI tools appear on no surveyed platform. And while several vendors claim validation, none publishes per-requirement evidence for each display on a public site. Those are the September slides: not “we rebuilt what exists” but “we built what the literature asked for last year and showed our work”.

Where we stand, by family

Have is shipped and on the site. On roadmap is a filed issue, in any stage. Partial is present on some renderers or for some purposes but not as a general capability. Out of scope is a judgment recorded on purpose — see proposal P12, which asks for that judgment to be written down rather than assumed.

What we have that they don’t

A gap analysis that only counts deficits misreads its own subject. Four things in this portfolio have no equivalent in the survey.

Method, and what it can’t tell you

Read the limits before the conclusions.