Before data moves, the clinical team gets the study cleared to start. That means an IRB approval or exemption, data use and site agreements, a decision on ClinicalTrials.gov registration, and a small pilot to test the forms and the reading workflow. In parallel the clinical lead gives the risk team the clinical side of the hazard analysis: what a wrong output would do to a patient, how likely a clinician is to act on it, and how biased data could mislead the device.
On the map: IRB submission and site agreements. ClinicalTrials.gov registration. Pilot subjects. Input to the hazard analysis.
Rows are disciplines, columns are stages 1 to 8. Select a square to open that task.
Back to the full map
Bring site, consent and data agreements to quality before the IRB filing. Give engineering the hazards clinicians raise, so they become risk controls and requirements.
Decide from the design. A retrospective study on de-identified archived data often qualifies for an IRB exemption or waiver, but the IRB makes that call. A prospective study that assigns patients to an intervention is a clinical trial and may need an IDE and registration. Record the reasoning, since the eSTAR asks about it.
Large enough to find problems with the reading software, the instructions and the case report forms before the full set is read. A few practice cases per reader is often enough, and the pilot cases stay out of the test set.
Name the clinical harm of a false positive, a false negative and a believable wrong measurement, and judge severity from what a clinician would do next. Add study-level hazards such as biased training or test data, and trace each one to a control such as a study report review.
See the open positions and apply.