Ask a question. Get evidence that shows its work.
Orbic takes real-world data from raw source to submission-ready output in one governed workspace. Connect, validate, harmonise, build cohorts, analyse, review and report, with every number traceable to its source.
For illustration only. Choose an example question, then press Ask.
Table 1. Baseline characteristics
| Characteristic | SMD |
|---|
SMD: standardised mean difference. Values under 0.10 are generally read as well balanced. Percentages use the cohort total as the denominator.
Click any number to see its proof
Define the question, not the query
Describe a cohort in plain language, review the logic it produces, and see attrition at every step.
Methods that are already there
Baseline balance, propensity weighting and standard summaries, with the choices recorded next to the result.
Code you can read and lock
Work in R, Python, SAS Viya or SQL. Programs are reviewed and approved, and what ran is what was approved.
From raw data to submission-ready output
Real-world studies slow down in the gaps between tools and teams. Orbic covers the whole path in one governed workspace. Select a stage to see the problem it removes.
From raw data to a table a reviewer can trust
Most teams stitch this together from notebooks, spreadsheets and email. Orbic keeps the whole chain in one workspace, so reviewers see how a result was made, not only the result.
Describe the population. See who is left at every step.
Type the cohort you want in plain language. Orbic drafts the definition, shows each inclusion and exclusion, and counts the patients that remain.
AI drafts. People decide.
A named reviewer signs off each program before it runs. The sign-off is recorded and locks that exact version.
Balance you can see
Weighting and diagnostics follow accepted practice, with the choices recorded beside the result.
A paper trail on every page
Tables export in the formats reviewers expect. The program, run time and inputs are printed on the page and kept in a run log.
Tables, figures and live dashboards
Orbic produces the full set a study needs, from submission-style tables to survival curves, forest plots and interactive dashboards. Every one carries the same evidence trail.
Figure 2. Overall survival by cohort
Figure 3. Subgroup hazard ratios
Controls like these are built in R or Python and published to your team, with the same approval and run log as any output.
Table 2. Time-to-event summary
| Statistic |
|---|
Programs behind the outputs
Assign it. Build it. Check it. Release it.
Statistical programming is a team effort with a paper trail. Every deliverable gets an owner, an independent QC step and a clear status, in the language your team already writes.
Know who owns what, and where it stands
Assign each table, figure and listing to a programmer and an independent QC reviewer. Click a status to move it along.
Illustrative example. Names are roles, not people.
Write in the language you know
Programmers keep their tools. Datasets move between languages in open formats, so a SQL step can feed an R model and a Python figure.
2
3
4
5
6
7
Create, edit and update programs in the built-in editor. Every save is a new version.
Give each deliverable an owner, a QC reviewer and a due date.
Build in R, Python, SAS Viya or SQL, with AI drafting if you want it.
An independent reviewer checks the output and records findings.
A named lead signs off the exact version, and that version is locked. Electronic signatures are part of the first release.
Outputs ship with the program, inputs and run log attached.
More than a place to run code
A statistical computing environment gives your team somewhere to run programs. Orbic adds the epidemiology: how cohorts are defined, how follow-up is judged, and how different data sources behave.
Time zero, aligned
Cohorts are built around a target trial. Eligibility, treatment and the start of follow-up meet at one index date, which guards against immortal-time bias.
Follow-up that knows its source
Claims have enrollment records. EHR data does not. Orbic detects which you have, tells you when follow-up is inferred, and keeps the assumption on record.
One engine, many sources
The same cohort logic runs across claims, EHR-derived, genomic and registry data, with no rewrite for each vendor.
Private by design
AI drafting runs on a model inside your own environment. Each organisation can run an isolated deployment with its own users, data and logs.
Know what is out of date
When a definition or input changes, every dependent table and figure is flagged, so nothing stale gets sent out.
Data cuts, with a freeze
When a new data cut arrives, dependent outputs are flagged. Re-run on the new cut, or freeze the study to the cut you submitted on.
Licence-aware data lifecycle
Track when each data licence ends. Expired data is quarantined and a signed deletion certificate is produced for the vendor.
Checks before anything runs
New data is checked for structure, duplicate keys and implausible dates before any analysis is allowed to use it.
The foundation reviewers expect
Everything a statistical computing environment should do, in one place.
- R, Python, SAS and SQL
- Versioned programs and datasets
- Lineage from data to table
- Assignments, QC and approval
- Role-based access and run log
- Private deployment
Change one input. See everything it touches.
Studies change. A definition is tightened, a dataset is refreshed. Orbic maps what depends on what and flags every output that is now out of date. Nothing re-runs behind your back.
Built for the way evidence is judged
Regulators, HTA bodies and internal reviewers ask the same questions: where did this come from, who approved it, and can you reproduce it?
Works with the data you already license
Bring licensed claims, EHR-derived and registry datasets into one governed environment. Orbic does not sell data. It connects through APIs, and we work with your team to integrate any data source you license.
Names are shown to describe the data you may license. They are trademarks of their owners, and Orbic is not affiliated with or endorsed by them.
Guardrails that stay on
- Patient-level values stay put. AI drafting works from your specification and variable names, not from patient records.
- No unapproved code runs. Programs execute only after a person approves them.
- Every result is reproducible. Inputs, program and run details are kept with the output.
Sign-offs record who approved which exact version. Electronic signatures, with identity re-confirmation at signing, and validation documentation are part of the first release and covered in demos.
See Orbic in action, end to end.
Request a demo and tell us what you would like to see. We will walk you through it live.
- A 30-minute walkthrough tailored to your role
- From a plain-language question to a table, with the evidence trail open
- Deployment options for your environment