|
 News
Customer Stories
About Us
Contacts
|
 SDMS Processes
 Step 1: Capture – Save files from LIMS, instrument files and reports, remote computers, direct uploads, and Microsoft Office®
Files generated during the LIMS workflow–including reports, documents, multimedia, images–are stored transparently and seamlessly in the SDMS repository. In addition, STARLIMS SDMS automates content capture by polling preconfigured data sources, such as instrument data stations. This function is performed by the SDMS Grabber, which monitors operating system file creation events; upon file creation, it immediately moves the file to the secure, 21CFR Part 11–compliant SDMS repository for further processing. Files can also be saved directly to the STARLIMS SDMS repository by using STARLIMS SDMS’s application-specific buttons in Microsoft Office® applications.
Step 2: Parsing - Artificial Intelligence streamlines content extraction
STARLIMS uses unique Artificial Intelligence algorithms to process unstructured documents in an automated manner. STARLIMS SDMS extracts content from the document and stores it together with the relevant metadata tags. This process transforms unstructured data to an XML format that is both standardized and highly accessible. Based on the values of the metadata tags, the file is routed to a workflow. Subsequently, a link to the document is created within the appropriate STARLIMS module, and in the case of instrument outputs, the results table is updated with results parsed from the captured files.
Before being able to automate the parsing and routing of documents, STARLIMS SDMS is configured to:
 |
Identify the file source (Varian Galaxy, Chemstation, Finigan GCMS, SOP, COA from vendors, etc.) |
|
|
Extract the appropriate metadata and content from the file (sample ID, instrument name, operator) and associated tables (i.e peak name, retention time, amount units for instrument files) |
|
|
Map extracted data to a user-configured XML format, also referred to as the Unified XML |
|
|
Define routing directions, for example:
|
|
All files captured from the CDS need to be reviewed prior to updating the LIMS results table |
|
|
All method SOPs need to be reviewed by the lab manager, a QC representative and approved by QA prior to becoming an active method in STARLIMS |
|
What makes this parsing approach unique is the application of Artificial Intelligence (AI) in the creation of the parsing routines. STARLIMS SDMS will automatically attempt to identify the following key document components (see screen capture):
|
|
Identification tags - words that always appear in the target document and are used to determine the file source. |
|
|
Key words (green)- these are label value pairs used to populate the metadata tags. |
|
|
Table (yellow)- the parser automatically identifies table structures, such as peak tables. |
|
|
Charts (purple) - the parser identifies charts that need to be extracted, such as chromatograms, etc.
|
Any inaccuracies in the parsing of the unidentified document can be adjusted with simple drag-and-drop manipulations, and the SDMS will automatically follow the most recent update to the parsing routine when processing files. This approach renders the laborious task of producing parsing scripts trivial, greatly reducing time required and eliminating the need to learn complex parsing applications.
|
|
