Your Clinical Data. Your Models. Your Measures.

Kelvin Cortex enables research and data science teams to build specialised analysis workflows in Python using Cortex Loom and publish them as custom Cortical Flows within Kelvin Cortex. In this example, the user selects Publish new Flow, installs the Cortex Loom Python package and generates an API token for publishing from their local development environment. A Python workflow called Average Tapping Velocity is then created to calculate the average velocity of left-hand finger tapping from pose-estimation data and published to Cortex. The new workflow appears in My Flows with a published status and can be selected alongside ready-to-use Cortical Flows. The user runs Pose Estimation and Average Tapping Velocity together on a clinical video, follows their processing progress to completion, selects the resulting pose-landmark and tapping-velocity outputs, and overlays the analysis outputs on the source video. The animation demonstrates how teams can develop and then retain ownership of their own analytical methods and intellectual property while using Kelvin Cortex to publish, run and scale those methods alongside existing analysis workflows.