The software itself does not pose a danger to either doctors or patients, but it can contribute
to dangerous situations. There are two main reasons for this – errors in software
specifications and incorrect coding. The risk analysis shows that after the measures taken to
reduce the risk, the residual risk is minimal.
Key Risk Categories Identified | Risk Control Measures Implemented |
---|---|
Wrong input data (e.g., unsupported studies, missing files, etc.) | – The study does not load; – A warning appears and study is loaded without CoLumbo analysis. |
Wrong analysis (e.g., incorrect segmentation or measurement, incorrect findings or modifiers, etc.) | – CoLumbo asks users to review and modify (or correct) the results of the analysis in 3 ways: o In the list of findings, o Through the visualizations, and o In the draft report. – Users independently review the original study in addition to CoLumbo analysis, modify and approve the report, and therefore are fully responsible for the diagnosis. |
Missed analysis (e.g., missed segmentation or measurement, missed finding, etc.) | – Users independently review the original study in addition to CoLumbo analysis, modify and approve the report, and therefore are fully responsible for the diagnosis |
Issues with operating environment (e.g., loss of power, lack of network connection, etc.) | – Study is not processed and can be loaded again upon power or network connectivity resumption; – User settings are stored on non-volatile memory; – Automatic switching to local hardware resources to continue the analysis if server connection is lost. |
Data security breach (e.g., unauthorized users, unsecured access, data tempering, etc.) | – User access controls and authentication; – All transferred data is encrypted. |
Misuse of software (i.e., radiologist users overly rely on the software) | – Present UI reminders, warnings for users to review, correct and approve displayed information and report; – CoLumbo creates summary tables that contain information about what changes the user made for each study. These tables can be generated weekly, monthly and annually and reviewed by the lead radiologist of the medical center/hospital. |