I recently returned from Snowflake Summit 22, and I am certainly impressed at the depth and breadth of the new features announced during the conference.
Snowflake has made a huge leap from a data warehouse to a full-scale data application platform.
I am most excited about Snowpark for Python and the Native Application Framework.
Photo by Author - Never Summer Wilderness
The latest Snowflake release has a number of excellent features for Data Science in Healthcare.
User Defined Functions and Streamlit
User defined functions in Snowflake are a fundamental change that transform it into a true application platform on the same lines as AWS or Azure. It is no longer just a data warehouse.
Snowpark Developer Guide for Python
On the Data Science Front, a great announcement at the Summit was the Snowpark API for Python.
How to use UDF to for the HL7 Healthcare Interchange Format
HL7 is a standard interchange for exchanging health information. It stands for Health Level 7, and is a form of EDI.
As an EDI format, HL7 doesn't lend itself easily to parsing, and HL7apy (pronounced 'happy') library makes it easy. See Chaos in Standardization, Sample HL7 Messages.
HL7 is used as an interchange to update patient records. For example, a VXU message is a common way to update a patient's vaccine record in a registry.
Snowflake support JSON, and hl7apy can be used (see Patrick Schmid's Blog).
Converting the HL7 to JSON supports the concept of Operational Data Store, and we can use Snowflake's native JSON capabilities to build Healthcare applications
-- Query the data structure
select t."$1":msh:"none"[0]:varies_1:st::string last_name
, t."$1":msh:"none"[0]:varies_2:st::string first_name
, t."$1":msh:"none"
, $1
from @stage.HL7_STG (
file_format => stage.JSON_FILE_FORMAT,
pattern => '.*.json'
) t;More to come. The value here is that we can utilize Snowflake as an application distribution environment.
Comments