2018-08-06 Meeting notes

• NHSBT feed to PV has not been restored. George Swinnerton has not chased this. He will follow up. Last file was sent 08 Jul 2018
• AKI issues and dump have not been specified yet. This will be followed up when the AKI Dump requirements are properly defined.
• NHSExtract has not been run by George Swinnerton. He will do this today 06 Aug 2018 
• Changes to the Nurture Dashboard have all been completed by George Swinnerton. There is still periodic issues with the graphs not displaying either because the sqlite file is corrupt or it needs syncing in metabase. The exact cause has not been identified.
• foghorn_leghorn (Unlicensed) has completed the programming for
  Error rendering macro 'jira' : Unable to locate Jira server for this macro. It may be due to Application Link configuration.
  and is awaiting Fran Benoy-Deeney's return to sign off for release.
  
• foghorn_leghorn (Unlicensed) NHSBT file hit an issue with a new bug introduced so is now being re-run.
  
• The meeting to review RaDaR tickets will be scheduled once the RaDaR feeds are re-enabled so as not to distract.
• rapolas (Unlicensed) is still looking at options for how to incorporate reports within the radar structure to allow specialist ones to be available to certain cohorts only etc. This will help shape how these could be added in the future to make them more user accessible.
  

The process of producing an extract for the data managers directly from the ukrdc was discussed. George Swinnerton pointed out that a relatively simple task could be to produce a data quality report highlighting missing data before the extract is run allowing the data managers to chase the correction of these values in advance. An example would be missing Date of Death. Exactly what could be included in this needs to be discussed and defined and whilst his may be useful the focus should be on creating an extract first. If the time to produce this increases then the Data Quality report could be a way for the managers to help anticipate and correct potential errors before validation. Once the process is established the DQ report would be a useful tool to monitor incoming data and flag potential issues in advance of the data extract.

Initially a specification for the extract needs to be written. This needs to define all the required elements where they will be pulled from and what rules need to be used to define the value when it is one of several that can be returned. George Swinnerton will define the initial specification for the extract once this is defined all fields that need clinical input on what rules should be used can be referred to the appropriate expert. Which elements are required as a minimum would need to be identified to help define the order the various parts of the dataset extract are written.

George Swinnerton

George Swinnerton has tested the Royal free files on staging and they all process without error. They are keen to start live feeds. George Swinnerton will contact them to switch on pdf feeds for one or two patients initially and then for all patients. live-mirth will need monitoring to ensure no untoward errors/issues arise.

George Swinnerton reported some changes that SSG have made to Patientview to improve its error handling.

Queuing has been updated to work more as expected. Messages are received into the queue and are being pulled out using 20 threads each appearing to have ten messages so the queue has up to 200 unacknowledged messages at anyone time until the messages is successfully saved to PV. Any unacknowledged message should get re-queued in the event that the importer restarts.

PV should handle issues with the postgres database stopping (running out of space) and reconnect on the database return.

In theory they have again fixed "Can't retrieve patient data" but this is difficult to verify.

Tom at SSG has also been looking at the programembership issue TNG-180 and verifying the cxml against the schema. George Swinnerton was not sure if he had managed to upload a new version before leaving on holiday.

Tim Whitlock (Deactivated) noted that the current test code for the importer changes is now logging complete PV files into the tomcat-importer log file. This should not be done. George Swinnerton to followup.