Solutions Automated Extraction Engine

The CloudMedx Automated Extraction Engine takes multi-format data from different source systems and automatically converts them to target systems (API enabled). It is an LLM-based application, fine-tuned for automating data pipelines for use cases such as rules extraction, registry and quality reporting, and custom data systems.

For organizations that deal with a lot of incoming data in multiple formats, they use the Engine to transform the multi-source, multi-format data into a specified output format. With our proprietory healthcare specific NLP, we can easily understand the various structured and unstructured documents, and place the data fields into the correct categories.

Here are some of our use cases:

For organizations required to do quality reporting, utilizing our healthcare and specialty specific NLP, the Automated Extraction Engine, understands sub-domain specific terminologies to automatically generate reports. This reduces time and labor efforts dramatically. For exaxmple, organizations required to do MIPS reporting, the engine can extract data such as ODI (Oswestry Disability Index, RM (Roland Morris Disability/Activity Questionnaire) or NDI (Neck Disability Index) to name a few that are used in reporting for MIPS Orthopedic Surgery’s Functional Outcome Assessment. There are over 40 specialties for MIPS (Merit-Based Incentive Payment System) reporting with 100s of measures per specialty for MIPS alone. Each measure description on an average is 20+ pages. The engine works for other quality reporting requirements such as HEDIS, STAR, MACRA etc.

For an organization that does claims editing, they typically  spend countless hours extracting this information manually. With an automated system, the time and cost is drastically reduced.

Similarly, organizations that do Risk Adjustment can reduce countless hours by instantly identifying missing data fields per provider, speeding up the entire risk adjustment process — including generating chase lists and suspect lists. When combined with the CloudMedx RA Analyzer, organizations are able to do end-to-end risk adjustment.

The Automated Extraction Engine can be hosted on-customer-premise or on private cloud  — ensuring that the customer retains full control over their data. 

How it Works

Step 1: Multi-source Data Ingestion

Low code, no code AI LLM continuously scans through large number of healthcare rules for reporting purposes — from dozens of data sources.

Step 2: Automated Rules Extraction

Extracts rules from different sources and summarizes the rules (inclusion/exclusion criteria) for humans to review. Staff no longer needs to manually create the rules, they can quickly review machine generated rules and confirm the summary. They can also edit the rules, remove certain rules or add new ones.

Step 3: Apply Rules to Patient Data

The rules that are generated in step 2 are then applied on patient data. The Engine:

• highlights data formatting errors
• performs data completeness checks
• identifies missing information specific to each use-case/reporting requirement
• excludes patients that don’t fit the criteria
• generates list of patients that fit the crieria and numerator, denominator values
• performs calculations that are required per rule/reporting measure

Step 4: Automated Report Generation

In the end, a report is generated in the format specified by customer or standard reporting format based on the use-case. This report can then be exported in excel for submission.

The Engine is API enabled and therefore, the output can directly be integrated with existing systems in use.