IMO Precision Normalize

IMO’s deep expertise in clinical data is grounded in terminology used by over 740,000 US physicians during point-of-care documentation. IMO Precision Normalize leverages this foundation to efficiently standardize inconsistent clinical data from diverse systems into consistent, structured, clinically validated terminology that is comprehensively mapped to standard code systems behind the scenes. IMO Precision Normalize helps organizations across the health IT ecosystem to enrich clinical data, accelerate data readiness for analytics, and power accurate insights.

Features

1. Industry-leading clinical terminology: Every IMO solution starts with a foundation of the definitive source of clinical terminology, managing the complexity of 5.0 million clinical terms with current and comprehensive mappings to all major global coding systems:<br/>-- Problem/Diagnosis: ICD-9-CM, ICD-10-CM, SNOMED CT®<br/>-- Procedure: CPT®, ICD-10-PCS, HCPCS, LOINC, SNOMED CT®<br/>-- Medication: RxNorm, NDC

2. Robust normalization engine: The engine that drives IMO Precision Normalize is capable of processing up to 300,000 terms per minute across diagnosis, procedure, medication, and laboratory data domains. By comparing input data to IMO’s clinically-vetted, highly specific, and propriety terminology, the engine accurately matches most terms with a high degree of confidence to minimize manual review.

3. Description and code-based matching: Client supplied clinical descriptions or codes pertaining to problems, diagnoses, procedures and labs, and medications are matched to IMO terms. The engine’s matching algorithms can account for both exact and fuzzy matches of descriptions.

4. Auto domain resolution: In cases where the domain (problem, procedure, or medication) of a description is unknown, the engine can attempt to automatically detect a term’s domain and match it appropriately.

5. Data review tools: Intuitive UI enables simple and flexible review of matches by domain, volume of use, and confidence score. Enables accelerated localization, remapping, and follow-up normalization operations.

Improve clinical data quality – at scale

1. Enrich clinical data: Built on a dictionary of 5 million terms and comprehensive mapping to all appropriate standard code systems behind the scenes

2. Accelerate data review: Increase the accuracy of normalized data with an intuitive interface to review, accept, and adjust matched data as needed

3. Reduce maintenance burden: IMO content is updated 6 times a year to be compliant with health system and regulatory requirements

4. Increase data specificity: Foundational terminology is curated by clinical SMEs, increasing the accuracy and specificity of terms and codes

This application currently only allows term normalization of diagnosis to ICD-10-CM codes. To retrieve full set of codes provided by the IMO Precision Normalize API please contact IMO.

<p><br/></p>

<p><br/></p>

Pre-requisites:

1. Contact IMO to get credentials to start using IMO Precision Normalize API.
2. Set up network rule and create integration. See Readme docs accompanied with the application package for the necessary script to run.

3. Set up a database table named 'patient_terms' with patient conditions having the following columns (RECORD_ID, CONDITION, NORMALIZED_ICD10CM, REQUEST_ID)

4. Populate the table created in step 2. with your dataset leaving Normalized_ICD10CM and Request_ID columns empty. Here is a link for an example dataset https://github.com/imohealth/snowflake-normalize-integration/blob/master/patient_conditions.csv

<p><br/></p>

Usage:

1. Go to DataProducts -> Apps.
2. Click on Normalize_Connector_Instance (This takes less than a minute to load the streamlit application).
3. The app will prompt you to give itself access to your sample dataset, the 'patient_terms' table created in pre-requisite step 2
4. Populate IMO Precision Normalize API production ClientID and Secret in the form presented.
5. Set Batch Size to 30
6. Set the Term Description Column Name to 'condition'
7. Click 'Start Normalization'

This will kick of the process of your dataset normalization. 30 records are being sent to the Normalize API at a time getting the icd10cm code. You can now query your 'patient_terms' table to see the results