⬆️ bump medcat==1.7.0

.github/workflows/ci.yml

@@ -8,9 +8,6 @@ on:
     branches:
       - master
 
-env:
-  SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL: True
-
 jobs:
   Test:
     runs-on: ubuntu-20.04
@@ -29,6 +26,6 @@ jobs:
       - name: download model
         run: |
           python -m spacy download en_core_web_md
-          pip install https://huggingface.co/kormilitzin/en_core_med7_lg/resolve/9633b53e7f9488408eae5dfff51b831b3b787657/en_core_med7_lg-any-py3-none-any.whl
+          pip install https://huggingface.co/kormilitzin/en_core_med7_lg/resolve/main/en_core_med7_lg-any-py3-none-any.whl
       - name: run pytest
         run: pytest ./tests/*

Dockerfile

@@ -9,7 +9,7 @@ ADD ./notebooks/ /home/notebooks
 WORKDIR /home/app
 
 RUN python -m pip install --upgrade pip
-RUN pip install https://huggingface.co/kormilitzin/en_core_med7_lg/resolve/9633b53e7f9488408eae5dfff51b831b3b787657/en_core_med7_lg-any-py3-none-any.whl
 RUN pip install /home/
 
+RUN pip install https://huggingface.co/kormilitzin/en_core_med7_lg/resolve/main/en_core_med7_lg-any-py3-none-any.whl
 RUN python -m spacy download en_core_web_md

pyproject.toml

@@ -1,17 +1,41 @@
+[build-system]
+requires = [
+    "setuptools>=45",
+    "setuptools_scm[toml]>=6.2",
+]
+build-backend = "setuptools.build_meta"
+
 [project]
 name = "miade"
+authors = [
+    {name = "James Brandreth", email = "[email protected]"},
+    {name = "Jennifer Jiang", email= "[email protected]"}
+]
+description = "A set of tools for extracting formattable data from clinical notes stored in electronic health record systems."
+requires-python = ">=3.8"
 dynamic = ["version"]
 dependencies = [
-    "medcat==1.2.8",
-    "spacy==3.1.0",
-    "typing==3.7.4.3",
-    "pandas",
-    "typer==0.3.2",
-    "click==7.1.2",
-    "pathlib",
-    "pydantic",
-    "negspacy",
-    "scikit-learn",
+    "medcat==1.7.0",
+    "spacy>=3.4.2, <3.5.0",  # need to be compatible with latest med7 model
+    "pandas<=1.5.3, <2",  # pandas updated to 2.0.0 recently, should we update too?
+    "typer>=0.7.0",
+    'click>=8.0.4',
+    "pathlib>=1.0.1",
+    "pydantic>=1.10.0",  # compatibility with spacy
+    "negspacy>=1.0.3",
+]
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=7.1.3",
+    "jupyterlab>=3.4.7",
+    "black>=22.12.0",
+]
+dashboard = [
+    "streamlit>=1.20.0",
+    "streamlit-aggrid>=0.3.4.post3",
+    "spacy-streamlit>=1.0.5",
+    "seaborn>=0.12.2",
 ]
 
 [project.scripts]
@@ -28,11 +52,3 @@ where = ["src"]
 
 [tool.setuptools.package-data]
 miade = ["data/*.csv"]
-
-[build-system]
-requires = [
-    "setuptools>=45",
-    "setuptools_scm[toml]>=6.2",
-]
-build-backend = "setuptools.build_meta"
-

src/miade/model_builders/cdbbuilder.py

@@ -99,13 +99,13 @@ def preprocess_custom(self, output_dir: Path = Path.cwd()) -> Path:
     def preprocess(self):
         self.vocab_files = []
         if self.snomed:
-            self.vocab_files.append(self.preprocess_snomed(self.temp_dir))
+            self.vocab_files.append(str(self.preprocess_snomed(self.temp_dir)))
         if self.fdb_data_path:
-            self.vocab_files.append(self.preprocess_fdb(self.temp_dir))
+            self.vocab_files.append(str(self.preprocess_fdb(self.temp_dir)))
         if self.elg_data_path:
-            self.vocab_files.append(self.preprocess_elg(self.temp_dir))
+            self.vocab_files.append(str(self.preprocess_elg(self.temp_dir)))
         if self.custom_data_paths:
-            self.vocab_files += self.custom_data_paths
+            self.vocab_files += str(self.custom_data_paths)
 
     def create_cdb(self) -> CDB:
         cdb = self.maker.prepare_csvs(self.vocab_files, full_build=True)

tests/examples/common_doses_for_miade.csv

@@ -50,7 +50,7 @@ Clobetasone 0.05% cream,Apply Thinly Once Or Twice A Day,,,,,0.5,d,TRUE,FALSE,
 Adapalene 0.1% gel,"Apply Thinly At Night,45 gram",,,,g,1,d,FALSE,FALSE,
 Amoxicillin 500mg capsules,"One To Be Taken Three Times A Day,21 capsule",1,,,{cap},0.333333333333333,d,TRUE,FALSE,7
 ViATIM vaccine suspension for injection 1ml pre-filled syringes (Sanofi Pasteur),"Immediately,1 pre-filled disposable injection",1,,,{injection},,,,,1
-Amitriptyline 10mg tablets,"take one at night, building gradually to five at night,56 tablet",5,,,{tbl},1,d,FALSE,FALSE,
+Amitriptyline 10mg tablets,"take one at night, building gradually to five at night,56 tablet",1,,,{tbl},1,d,FALSE,FALSE,
 Propranolol 40mg tablets,"One To Be Taken Twice A Day,2 x 28 tablet",1,,,{tbl},0.5,d,TRUE,FALSE,28
 Sumatriptan 10mg/0.1ml nasal spray unit dose,"one spray as directed,2 x 2 unit dose",1,,,{unit},,,,,
 Phenoxymethylpenicillin 250mg tablets,"TAKE TWO TABLETS EVERY SIX HOURS FOR SEVEN DAYS,64 tablet",2,,,{tbl},6,h,FALSE,FALSE,7