Skip to content

The dashboard for management

In this chapter

  • Open the study dashboard and find the result of each endpoint of the protocol.
  • Tell a result of the SAP from an exploratory finding on the dashboard.
  • Read the gate that checks every number before the dashboard is shown.
Time
20 minutes
At the end you have
the study dashboard of the pilot, checked: 12 pages, 18,392 numbers, each one equal to the ARD.

19.1 The idea: one place for the people who decide

Section titled “19.1 The idea: one place for the people who decide”

A clinical study report takes months. Management, the medical team and the project lead need the result earlier, and in a form they can explore: by sex, by age group, by site. The risk of a dashboard is that its numbers drift from the tables: a filter computed differently, a rounding, an old data cut. Then two numbers for the same thing circulate.

The study dashboard of Sapling reads its numbers from the ARD, the same numbers as the tables. Before it is shown, its gate reads every number on every page and compares it with the ARD. One number different, and the gate fails.

Terminal
$ sapling dashboard --check --no-browser
the configuration: the omakase default (out/dashboard/addendum.md)
contract: CONTRACT 12 pages (RESULT: CLEAN)
10 dashboard: 12 pages, 18392 numbers equal to the ARD: DASHBOARD PASS
dashboard DASHBOARD PASS (55.0 s)

The gate opened every page of the dashboard in a browser and compared what is on screen with the ARD: 12 pages, 18,392 numbers, all equal. Open the dashboard yourself with sapling dashboard: it starts on a free port and opens in your browser.

The first page answers the first question of a manager: what did the protocol ask, and where is the answer?

The Protocol page of the study dashboard: a table with the endpoints of the protocol, the decision of the SAP for each with the cited protocol sentence, and a link to its result.
The Protocol page: each endpoint of the protocol (USDM), the SAP's decision with its protocol sentence, and the link to its result. At the top: the data cut, the subgroup filters, and the source of the numbers (the chain's ARD).Source: sapling dashboard, the pilot after chapter 20
The Efficacy page of Table 14.3.1 on the dashboard: the protocol sentence of section 4.3.3, then the table of the ADAS-Cog (11) by treatment group with the p-value 0.245.
The primary analysis on the dashboard: the same Table 14.3.1 as chapter 15, with the protocol sentence that plans it. A click on a number shows its analysis, its SAP block and the protocol's sentence.Source: sapling dashboard, page Efficacy · T14_3_1
The AE explorer of the dashboard: adverse events by body system and preferred term with counts and percentages per treatment group; application site pruritus and erythema are frequent in the xanomeline groups.
The AE explorer: the treatment-emergent adverse events by body system and term. A count opens the list of its patients. It is marked exploratory: a query on the data, not a table of the SAP.Source: sapling dashboard, page AE explorer
The Insights page of the dashboard: findings from the data, each with its subjects, such as 3 subjects with a serious adverse event, 3 deaths, more discontinuations on xanomeline low dose, and pruritus more often on xanomeline high dose.
Insights: findings that the dashboard computes from the data, each with its subjects and its test. They are exploratory, not results of the SAP, and the page says so.Source: sapling dashboard, page Insights

The badge NOT SEALED at the top says that this run has no entry in the ledger of the study. A study that uses the ledger seals each refresh of the results, each input and product with its SHA-256, and the audit trail page shows it. The book’s run does not use the ledger.

Your turn 19.1 Read the gate's report

The gate writes its report to out/dashboard/output/dashboard_check.txt. How does it compare the numbers, and on which pages?

Show the answer
out/dashboard/output/dashboard_check.txt
app: apps/dashboard on port 3102 (a free port of 3100-3199), the study's output
app log: 14 lines, 0 error(s)
timing: the app answered 2.3 s after its R process started (the cache: used); the first page in a new browser tab 0.6 s; a page of the ARD view: slowest 0.5 s (T14_1_1), median 0.3 s
groups: 12 pages in 3 menus of the addendum (Efficacy 2, Safety 6, Subjects 4), 0 page(s) of their own
1 ARD T14_1_1: 36 of 36 numbers on screen equal to the ARD
1 ARD T14_1_2: 92 of 92 numbers on screen equal to the ARD
1 ARD T14_2_1: 113 of 113 numbers on screen equal to the ARD
1 ARD T14_3_1: 58 of 58 numbers on screen equal to the ARD
1 ARD T14_3_2: 22 of 22 numbers on screen equal to the ARD
1 ARD T14_4_1: 57 of 57 numbers on screen equal to the ARD
1 ARD T14_5_1: 33 of 33 numbers on screen equal to the ARD
1 ARD T14_5_2: 2303 of 2303 numbers on screen equal to the ARD
...

It starts the dashboard app on a free port, reads its log, and then compares, page by page, the numbers on screen with the ARD: for example 58 of 58 for Table 14.3.1 and 2303 of 2303 for Table 14.5.2. The full report ends with all 12 pages and the verdict DASHBOARD PASS.

What you learned

  • The dashboard reads its numbers from the ARD, the same numbers as the tables.
  • Its gate compares every number on every page with the ARD: 18,392 numbers on 12 pages.
  • Planned results and exploration are on one screen, and marked apart.

Next: 20 · Four final products →

Found an error or have a question? Write to datastep.studio@gmail.com· The plan is written in Sapling; every R program runs in DataStep Studio for R.

Data: the CDISC pilot study (cdiscpilot01). The protocol as CDISC USDM from the CDISC 360i project (MIT licence); the SDTM data from the PhUSE scripts repository (MIT licence). Public example data: no real patient.

R is a project of the R Foundation. SAS is a registered trademark of SAS Institute Inc. CDISC, SDTM, ADaM and USDM are trademarks of the Clinical Data Interchange Standards Consortium. Claude is a trademark of Anthropic. This book is not affiliated with or endorsed by any of them.

© 2026 DataStep Labs. All rights reserved.