How it works
The stack behind every Sutrix study
01Review & governance
Statisticians stay in control at five checkpoints: cleaning proposal, dataset approval, plan review, validation and final report. Nothing moves forward without a recorded decision, and 41 statistical-practice rules with citations mark what blocks, what must be declared and what is advisory.
02Workflow orchestration
Each study runs the same ordered workflow: intake and audit, codebook, cleaning, analysis plan, execution, independent validation, report. Work is broken into controlled, repeatable steps with a fixed set of operations and a bounded method repertoire, never left to a general-purpose chatbot.
03Study record
The complete chain behind every output: upload hashes, cleaning code, the signed plan, the analysis code, the validation reconciliation and the report, sealed into a reproducibility bundle so any study can be re-run or audited later.
04Permissions and access control
Workspaces and studies are scoped per team, connector credentials are stored encrypted with AES-256-GCM, identifier fields are excluded by default, and every upload is screened for 18 identifier classes before anything else runs.
05Knowledge & data
Study files in their own slots: dataset, codebook, protocol, scoring key and prior paper. REDCap dictionaries and Qualtrics question rows become the codebook, the plan is extracted from the protocol, and PubMed literature is searched and attached to the study.
06Connectors
File upload for CSV, Excel, SAS transport, REDCap and Qualtrics exports, and a live REDCap connector scoped to one project. Public files such as NHANES load directly. Every source follows the same rule: identifiers excluded by default, access recorded.