Product launch · InovaMine
InovaMine is a biostatistics and programming function delivered as software. It ingests a dataset, interventional or real world, executes the statistical analysis plan against it, and returns the TLFs: Table 1 baseline characteristics, Table 2 endpoints and effect estimates, and every other table, listing and figure the SAP specifies. Every number comes with its lineage back to the source record. A statistician reviews and signs off before anything is released. Today we are making it available as a standalone product.
Every programme reaches the same point. The SAP is finalised, the data is locked, and then everything waits. Cohort derivation, variable mapping, programming, double programming, QC. On a claims dataset that is about six weeks. On a trial it is a quarter. Ask one follow up question, a new subgroup or a sensitivity analysis, and the clock restarts.
This is not a talent problem. Biostatisticians and statistical programmers do some of the most exacting work in the industry. It is a structural one: every study is bespoke. New data, new code, new validation, and the knowledge of how it all fits together held by a handful of people. The result is that evidence is rationed. Teams ask the three questions they can afford instead of the twenty they have.
Data enters through tested ingestion pipelines and lands in one of two common data models: a High Fidelity model for interventional studies, holding every SAP variable at subject and visit level, and a Market Wide model for large real world sources, hundreds of terabytes of hospital, claims and EHR data harmonised to a single clinical vocabulary. Thousands of automated tests run on every load, every transformation and every output. The engine then takes the SAP, or a cohort request, builds the population, executes the derivations and produces the TLFs. The customer works in a secure web platform where Table 1 and Table 2 are on demand, any cell opens to the rows behind it, and a patient journey view shows exactly why each subject met the inclusion criteria.
Tap to open full size. On a phone, swipe to pan or pinch to zoom.
Once a data source is mapped, the analysis code is already written and already tested, so the second study is configuration rather than engineering. A cohort such as “patients with comorbidity X, exposed to treatment Y, outpatient setting, index year Z” is no longer a new programme. It is a validated cohort, live within five working days, that anyone with access can interrogate in minutes.
Biostatisticians and epidemiologists move from building to reviewing. The first TLF set is already generated when they open the study, with lineage attached, so their time goes to estimand choices, design and clinical interpretation. Clinical development can test the feasibility of an external control arm or a natural history comparator in days. Medical affairs can turn around a journal reviewer’s sensitivity analysis the same week. Regulatory can demonstrate reproducibility rather than assert it. HEOR and market access draw from the same model as the submission, so the payer dossier and the regulatory package agree.
The quieter change is cultural. When a new cut costs hours instead of a change order, people ask the question they were going to leave unasked. Evidence stops being rationed.
A tool makes one person faster on one study. An engine changes the cost curve. Each additional study, data source and customer runs on the same tested models and the same code, so the marginal cost of evidence generation falls with scale while the quality bar stays fixed. InovaMine is already in production across interventional and real world data, licensed per dataset per year with analyses delivered on top.
It is also the foundation the rest of our platform stands on. Evidence strategy, protocol and SAP design, analysis, and regulatory and scientific writing now run end to end on validated workflows with expert sign off at each stage. InovaMine is the layer that makes the numbers in all of it defensible.
Strategy
Design
Evidence
Writing
Throughout