Direct Data Access for RIS/PACS

Potential Change

Template:Potential Change We could stop having collectors enter imaging data available from IMPAX onto their laptops and use the combined data from dump and from imaging.mdb instead.

To do that we would need to:

• import all data received from PACS and not yet imported
• confirm that the data is of comparable quality
• decide how to store this:
  • if storing in Ed's DB, confirm that we have all required data to output the same way Delphic does, and that Ed can import our data
  • if imaging.mdb is sufficient, ensure that we can generate output to support whatever this data is used for now (what is it used for? might want to put that into Labs).

Direct access to data (work in progress)

We are going to test accessing this data directly from RIS. Once we know it is feasible we will ask the Database Steering Committee if we can stop manual collection.

Data Acquisition - RIS/PACS

Breanne Batters from the RIS/PACS team will provide us with a monthly report with columns (hospital|unit|phin|chart|image_DateTime|imaging_type). This data will get imported into Regional Server\output\imaging.mdb, table name images. I have reports for July and August, but none after that... emailed Breanne Ttenbergen 16:28, 18 November 2011 (CST)

RIS/PACS ambiguous types

Things like "CT ANGIO ABDOMEN COMBINED" or "CT ANGIO ABDOMEN COMBINED" may be in the table s_aggregate_types twice to be counted as both Angios and CTs. Template:Discussion Is that how collectors would count it? Ttenbergen 15:37, 2015 August 19 (CDT)

• No, we only count arterial angios done by arterial access in our lab counts as anglos. CT angios are counted as CT's in the lab counts.

Data Acquisition - Collectors

Additionally, during sending, complete patients' only (hosp|loc|serial|phin|chart|admit_datetime|disch_datetime) will be sent to the table patients; this information is required to link images to our database.

• hopefully fixed with update CCMDB.mdb_Change_Log_2011#ver_2011-11-18 Ttenbergen 16:28, 18 November 2011 (CST)

combining

combine patients and images: see Imaging.mdb query combiner

Aggregating

There will be one line per image, so we will need to aggregate the lines into counts image_type. This will be done by a query combiner_pivoter which will provide a count for lines of the same combination of (Serial/PHIN /MRN/location/admit_dt_tm/discharge_dt_tm/imaging_type). The query will start to pick up records from the image file only once a line is present in the patients table, i.e. after a patient has been sent. Further, the query will of course only pick up images that have been imported.

Where to store

Data can be stored in the imaging.mdb program. I think it would be a long time before we would hit any size restrictions with this, and by then we will hopefully have moved on to a different data repository that can handle the volume.

There is another option for long-term storage. We are currently importing some values for HSC into TMSX. Do we have any documentation for the importer? If I stream the reports from RIS into the right format, would we be able to import into Ed's program directly? Ttenbergen 11:30, 4 August 2011 (CDT)

• In the Labdownloading program, the study and chart#, admit datetime and disch datetime are entered manually by batch to download the lab dump file from New Delphic. Then Ed's interface program picks up the corresponding match, sums it up and upload to the master database. I don't think we have the documentation of the importer. JMojica 12:13, 4 August 2011 (CDT)

Backups

eHealth backs up the server, so we should not have to back this up. Considering that we could get new dump files generated when needed we don't need to keep or back those up either.