AI Reporting launch can use JWT for user authentication. The expected JWT structure and content is described.

A re-structure from the webhook documentation including workflows and examples helps our HIT partners to integrate with our webhook API.

With the introduction of AI reporting, documentation of the integration for our HIT partners is created. It comprises the launch - which needs to be configured in a PACS or RIS, the ingestion of clinical information from a RIS and the return of the final report text to the RIS.

Our AI Reporting system is now ready to be tested in German. Speech recognition and final report creation is based on German trained algorithms now.

For beta-customers and prospects it is possible to test the AI Reporting module. Several AI algorithms are used for speech recognition, automatic template selection, final report structuring and impression creation. All available in English.

The representation of data is re-organized by displaying all information based on studies instead of job, which represent the customer view on the data. Several jobs can be created for one study depending on configured AI solutions or if AI solutions work on series. This distincation is important for IT administrators and maintenance, but not the medical consumer.

Each Gate can be enlarged to see the all configured AI solutions and it's traffic. Requested, completed, failed and rejected unique job counts are displayed in a table view and as a graph.

A unique job is created depending on the AI solution on patient, study or series level.

AI solution performance

This tab contains a table called AI findings summary, which lists all findings of the system over a selected period of time, it’s AI solution product name, the absolute number of positive predictions and the positive prediction rate. And a graph for positive prediction rate, which is the proportion of jobs that have predicted a present finding per completed job.

AI Porfolio

This page displayes the latest AI solutions on the deepcOS® platform. It is searchable for specific keywords or filter for solutions concerning modalities, applications, body parts or regulatory clearance. For each solution a product detail page gives more information about the solution and if applicable shows samples and references.

Analytics

Using an interactive dashboard to investigate on their own data, customers can navigate through their usage of deepcOS® Cloud infrastructure. Following charts and data points are available:

• Success rate trend: This displays the trend of proportion of jobs which returned a successful response.
• Gates: List of all gates of the organization or by a customers of the organization (in case of channel partners)
  • Total requests per gate
  • Success Rate per gate: Rate of studies, which were successfully processed and a result created. It does not show if the result was displayed in the PACS.
  • Rejection Rate per gate: Rate of studies, which were rejected by the AI solution itself due to mismatching specifications.
• AI solutions: either all available to the customer or filtered by gate
  • Total requests for the AI solution
  • Success Rate for the solution
  • Rejection Rate for the solution
  • Average Execution Time for the solution in seconds
• Total requests per day

System Manager

System Manager shows a list of Gates for this user and their uptime of the last 24 hours or 7 days in percentage and visually. The user can then click to navigate to the deepcOS Gate manager.

Deverloper Docs

AI solution developers can find here the documentation how to integrate their solution with the deepcOS® platform. The two APIs to consume AI processing and output data are described here in detail for HIT integrations.

Submit AI

The is a submission workflow to support AI partners to submit their new AI solution or an update to the deepcOS® platform.

Support

This page enumerates links for different user groups to reach out to deepc or read the release note for deepcOS® Hub.