Merge branch 'release/4.3.0'

.coveragerc

@@ -1,7 +1,7 @@
 # .coveragerc to control coverage.py
 [run]
 branch = True
-source = pyomop
+source = src.pyomop
 # omit = bad_file.py
 
 [paths]

.github/workflows/bump.yml

@@ -13,7 +13,7 @@ jobs:
 
     steps:
     - name: Checkout Latest Commit
-      uses: actions/checkout@v3
+      uses: actions/checkout@v4
 
     - name: Generate changelog
       uses: charmixer/auto-changelog-action@v1

.github/workflows/docs.yml

@@ -10,9 +10,9 @@ jobs:
     runs-on: ubuntu-latest
     timeout-minutes: 15
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python
-      uses: actions/setup-python@v2.3.1
+      uses: actions/setup-python@v5.0.0
       with:
         python-version: '3.10.13'
     - name: Install dependencies

.github/workflows/publish.yml

@@ -10,9 +10,9 @@ jobs:
     runs-on: ubuntu-latest
     timeout-minutes: 20
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python
-      uses: actions/setup-python@v2.3.1
+      uses: actions/setup-python@v5.0.0
       with:
         python-version: '3.10.13'
     - name: Install dependencies

.github/workflows/pytest.yml

@@ -13,18 +13,21 @@ jobs:
   build:
 
     runs-on: ubuntu-latest
-    timeout-minutes: 15
+    env:
+      CI: true
+    timeout-minutes: 10
     strategy:
       max-parallel: 4
       matrix:
         python-version: [3.10.13]
 
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v2.3.1
+      uses: actions/setup-python@v5.0.0
       with:
         python-version: ${{ matrix.python-version }}
+        cache: 'pip' # caching pip dependencies
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip

.github/workflows/tox.yml

@@ -1,4 +1,4 @@
-name: Python Test
+name: Tox Test
 
 on:
   push:
@@ -9,18 +9,19 @@ jobs:
   build:
 
     runs-on: ubuntu-latest
-    timeout-minutes: 15
+    timeout-minutes: 12
     strategy:
       max-parallel: 4
       matrix:
         python-version: [3.10.13]
 
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v2.3.1
+      uses: actions/setup-python@v5.0.0
       with:
         python-version: ${{ matrix.python-version }}
+        cache: 'pip' # caching pip dependencies
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip

CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## [Unreleased](https://github.com/dermatologist/pyomop/tree/HEAD)
+
+[Full Changelog](https://github.com/dermatologist/pyomop/compare/4.2.0...HEAD)
+
+**Closed issues:**
+
+- Google Cloud Platform Compatibility [\#9](https://github.com/dermatologist/pyomop/issues/9)
+
+## [4.2.0](https://github.com/dermatologist/pyomop/tree/4.2.0) (2023-11-19)
+
+[Full Changelog](https://github.com/dermatologist/pyomop/compare/4.1.0...4.2.0)
+
 ## [4.1.0](https://github.com/dermatologist/pyomop/tree/4.1.0) (2023-11-19)
 
 [Full Changelog](https://github.com/dermatologist/pyomop/compare/4.0.0...4.1.0)

README.md

@@ -9,6 +9,16 @@
 
 ### [Documentation](https://dermatologist.github.io/pyomop/)
 
+## UPDATE
+Recently added support for **LLM based natural language queries** of OMOP CDM databases using [llama-index](examples/llm_example.py). Please install the llm extras as follows. Please be cognizant of the privacy issues with publically hosted LLMs. Any feedback will be highly appreciated. [See usage](examples/llm_example.py)!
+
+```
+git clone https://github.com/dermatologist/pyomop.git@develop
+cd pyomop
+pip install pyomop[llm]
+```
+[See usage](examples/llm_example.py).
+
 ## Description
 
 The [OHSDI](https://www.ohdsi.org/) OMOP Common Data Model allows for the systematic analysis of healthcare observational databases. This is a python library to use the CDM v6 compliant databases using SQLAlchemy as the ORM. **pyomop** also supports converting query results to a pandas dataframe (see below) for use in machine learning pipelines. See some useful [SQL Queries here.](https://github.com/OHDSI/QueryLibrary)

dev-requirements.txt

@@ -12,16 +12,14 @@ babel==2.9.1
     # via sphinx
 backports-entry-points-selectable==1.1.0
     # via virtualenv
-certifi==2022.12.7
+certifi==2023.7.22
     # via requests
 charset-normalizer==2.0.6
     # via requests
 commonmark==0.9.1
     # via recommonmark
 coverage[toml]==5.5
-    # via
-    #   coverage
-    #   pytest-cov
+    # via pytest-cov
 distlib==0.3.2
     # via virtualenv
 docutils==0.17.1
@@ -38,7 +36,7 @@ imagesize==1.2.0
     # via sphinx
 iniconfig==1.1.1
     # via pytest
-jinja2==3.0.1
+jinja2==3.1.4
     # via sphinx
 markupsafe==2.0.1
     # via jinja2
@@ -58,7 +56,7 @@ py==1.10.0
     # via
     #   pytest
     #   tox
-pygments==2.10.0
+pygments==2.15.0
     # via sphinx
 pyparsing==2.4.7
     # via packaging
@@ -68,13 +66,13 @@ pytest==6.2.5
     #   pytest-cov
 pytest-cov==3.0.0
     # via -r dev-requirements.in
-pytz==2023.3.post1
+pytz==2024.1
     # via
     #   -c requirements.txt
     #   babel
 recommonmark==0.7.1
     # via -r dev-requirements.in
-requests==2.26.0
+requests==2.32.2
     # via sphinx
 setuptools-scm==6.3.2
     # via -r dev-requirements.in
@@ -110,7 +108,7 @@ tomli==1.2.1
     # via setuptools-scm
 tox==3.24.5
     # via -r dev-requirements.in
-urllib3==1.26.6
+urllib3==1.26.19
     # via requests
 virtualenv==20.8.0
     # via tox

examples/llm_example.py

@@ -0,0 +1,60 @@
+"""Example of using LLMs with PyOMOP
+git clone https://github.com/dermatologist/pyomop.git@develop
+cd pyomop
+pip install -e .[llm]
+"""
+
+from pyomop import CdmEngineFactory, Cohort, metadata, CdmLLMQuery, CDMDatabase
+import datetime
+import asyncio
+# Import any LLMs that llama_index supports and you have access to
+# Require OpenAI API key to use OpenAI LLMs
+from llama_index.llms import Vertex
+
+async def main():
+    # Create a sqllite database by default
+    # You can also connect to an existing CDM database using the CdmEngineFactory
+    cdm = CdmEngineFactory()
+    # Postgres example below (db='mysql' also supported)
+    # cdm = CdmEngineFactory(db='pgsql', host='', port=5432,
+    #                       user='', pw='',
+    #                       name='', schema='cdm6')
+
+    engine = cdm.engine
+    # Create Tables if required
+    await cdm.init_models(metadata)
+
+    async with cdm.session() as session:
+        async with session.begin():
+
+            # Adding  a cohort just for the example (not required if you have a OMOP CDM database)
+            session.add(Cohort(cohort_definition_id=2, subject_id=100,
+                cohort_end_date=datetime.datetime.now(),
+                cohort_start_date=datetime.datetime.now()))
+            await session.commit()
+
+            # Use any LLM that llama_index supports
+            llm = Vertex(
+                model="chat-bison",
+            )
+            # Include tables that you want to query
+            sql_database = CDMDatabase(engine, include_tables=[
+                "cohort",
+            ])
+            query_engine = CdmLLMQuery(sql_database, llm=llm)
+            # Try any complex query.
+            response  = query_engine.query("Show each in table cohort with a subject id of 100?")
+    print(response)
+    """
+
+| cohort_id | subject_id | cohort_name |
+|---|---|---|
+| 1 | 100 | Math |
+
+    """
+    # Close session
+    await session.close()
+    await engine.dispose()
+
+# Run the main function
+asyncio.run(main())

requirements.txt

@@ -4,25 +4,27 @@
 #
 #    pip-compile
 #
-aiosqlite==0.19.0
+aiosqlite==0.20.0
     # via pyomop (setup.py)
 click==8.1.7
     # via pyomop (setup.py)
-greenlet==3.0.1
+greenlet==3.0.3
     # via sqlalchemy
-numpy==1.26.2
+numpy==2.0.0
     # via pandas
-pandas==2.1.3
+pandas==2.2.2
     # via pyomop (setup.py)
-python-dateutil==2.8.2
+python-dateutil==2.9.0.post0
     # via pandas
-pytz==2023.3.post1
+pytz==2024.1
     # via pandas
 six==1.16.0
     # via python-dateutil
-sqlalchemy==2.0.23
+sqlalchemy==2.0.31
     # via pyomop (setup.py)
-typing-extensions==4.8.0
-    # via sqlalchemy
-tzdata==2023.3
+typing-extensions==4.12.2
+    # via
+    #   aiosqlite
+    #   sqlalchemy
+tzdata==2024.1
     # via pandas

setup.cfg

@@ -54,6 +54,11 @@ exclude =
 # `pip install pyomop[PDF]` like:
 # PDF = ReportLab; RXP
 # Add here test requirements (semicolon/line-separated)
+llm =
+    llama-index
+    langchain==0.0.350
+    overrides
+
 testing =
     pytest
     pytest-cov

src/pyomop/__init__.py

@@ -1,9 +1,15 @@
 import sys
-
+import logging
+_logger = logging.getLogger(__name__)
 from .engine_factory import CdmEngineFactory
 from .cdm6_tables import metadata
 from .vocabulary import CdmVocabulary
 from .vector import CdmVector
+try:
+    from .llm_query import CdmLLMQuery
+    from .llm_engine import CDMDatabase
+except ImportError:
+    _logger.warn("LLM is not installed. Please install LLM with pip install pyomop[llm] to use this package.")
 
 from .cdm6_tables import AttributeDefinition
 from .cdm6_tables import CareSite