Merge pull request #15 from mmcdermott/build_task_cmd

A quick helper to help build ACES task labels.

README.md

@@ -25,3 +25,27 @@ TODO
 ### To Add Results
 
 TODO
+
+## Helpers
+
+### To extract a task
+
+First, clone the repo and install it locally with `pip install .` Then, make sure you have the desired task
+criteria and dataset predicates yaml files in their respective locations in the repo.
+
+Finally, run the following:
+
+```bash
+./src/MEDS_DEV/helpers/extract_task.sh $MEDS_ROOT_DIR $DATASET_NAME $TASK_NAME
+```
+
+E.g.,
+
+```bash
+./src/MEDS_DEV/helpers/extract_task.sh ../MEDS_TAB_COMPL_TEST/MIMIC-IV/ MIMIC-IV mortality/in_icu/first_24h
+```
+
+which will use the `datasets/MIMIC-IV/predicates.yaml` predicates file, the
+`tasks/criteria/mortality/in_icu/first_24h.yaml` task criteria, and will run over the dataset in the root
+directory at `../MEDS_TAB_COMPL_TEST/MIMIC-IV`, reading data from the `data` subdir of that root dir and
+writing labels to the `task_labels` subdir of that root dir, in a name dependent manner.

pyproject.toml

@@ -3,7 +3,7 @@ requires = ["setuptools>=64", "setuptools-scm>=8.0", "wheel"]
 build-backend = "setuptools.build_meta"
 
 [project]
-name = "MEDS-DEV"
+name = "MEDS_DEV"
 dynamic = ["version"]
 authors = [
     {name="Matthew B. A. McDermott", email="[email protected]"},
@@ -28,7 +28,7 @@ classifiers = [
     "License :: OSI Approved :: MIT License",
     "Operating System :: OS Independent",
 ]
-dependencies = ["meds==0.3", "es-aces==0.3.0"]
+dependencies = ["meds==0.3.3", "es-aces==0.5.0"]
 
 [tool.setuptools_scm]
 

src/MEDS-DEV/__init__.py → src/MEDS_DEV/__init__.py

@@ -1,6 +1,6 @@
 from importlib.metadata import PackageNotFoundError, version
 
-__package_name__ = "MEDS-DEV"
+__package_name__ = "MEDS_DEV"
 try:
     __version__ = version(__package_name__)
 except PackageNotFoundError:

src/MEDS_DEV/configs/_ACES_MD.yaml

@@ -0,0 +1,23 @@
+defaults:
+  - _aces
+  - override data: sharded
+  - _self_
+
+dataset_name: ${oc.env:MEDS_DATASET_NAME}
+task_name: ${oc.env:MEDS_TASK_NAME}
+output_dir: "${oc.env:MEDS_ROOT_DIR}/task_labels"
+
+# TODO: find a nice way to have this be inferred automatically
+MEDS_DEV_dir: "${oc.env:MEDS_DEV_REPO_DIR}"
+
+data:
+  standard: meds
+  root: "${oc.env:MEDS_ROOT_DIR}/data"
+
+# Cohort directory and name: used for automatically loading configs, saving results, and logging.
+cohort_dir: ${output_dir}
+cohort_name: ${task_name}
+
+# Path to the task configuration file
+config_path: ${MEDS_DEV_dir}/tasks/criteria/${task_name}.yaml
+predicates_path: ${MEDS_DEV_dir}/datasets/${dataset_name}/predicates.yaml

src/MEDS_DEV/datasets/MIMIC-IV/predicates.yaml

@@ -0,0 +1,18 @@
+predicates:
+  hospital_admission:
+    code: { regex: "^HOSPITAL_ADMISSION//.*" }
+  hospital_discharge:
+    code: { regex: "^HOSPITAL_DISCHARGE//.*" }
+
+  ED_registration:
+    code: { regex: "^ED_REGISTRATION//.*" }
+  ED_discharge:
+    code: { regex: "^ED_OUT//.*" }
+
+  icu_admission:
+    code: { regex: "^ICU_ADMISSION//.*" }
+  icu_discharge:
+    code: { regex: "^ICU_DISCHARGE//.*" }
+
+  death:
+    code: MEDS_DEATH

src/MEDS_DEV/helpers/extract_task.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+
+export MEDS_ROOT_DIR=$1
+export MEDS_DATASET_NAME=$2
+export MEDS_TASK_NAME=$3
+
+shift 3
+
+MEDS_DEV_REPO_DIR=$(python -c "from importlib.resources import files; print(files(\"MEDS_DEV\"))")
+export MEDS_DEV_REPO_DIR
+
+SHARDS=$(expand_shards "$MEDS_ROOT_DIR"/data)
+
+aces-cli --config-path="$MEDS_DEV_REPO_DIR"/configs --config-name="_ACES_MD" \
+  "hydra.searchpath=[pkg://aces.configs]" "data.shard=$SHARDS" -m "$@"