NBS Test Cases

.gitignore

@@ -73,3 +73,6 @@ __pycache__/
 
 # Databases
 *.sqlite3
+
+# Test result files
+output.csv

compose.yml

@@ -76,3 +76,25 @@ services:
     depends_on:
       api:
         condition: service_healthy
+
+  algo-test-runner:
+      build:
+        context: tests/algorithm
+        dockerfile: Dockerfile.algo
+      env_file:
+        - tests/algorithm/algo.env
+      environment:
+        DB_URI: "postgresql+psycopg2://postgres:pw@db:5432/postgres"
+        API_URL: "http://api:8080"
+      volumes:
+        - ./tests/algorithm/scripts:/app/scripts
+        - ./tests/algorithm/data:/app/data
+        - ./tests/algorithm/results:/app/results
+        - ./tests/algorithm/configurations:/app/configurations
+      depends_on:
+        db:
+          condition: service_healthy
+        api:
+          condition: service_healthy
+      profiles:
+        - algo-test

tests/algorithm/Dockerfile.algo

@@ -0,0 +1,12 @@
+# Use the official Python 3.11 slim image as the base
+FROM python:3.12-slim
+
+# Set the working directory
+WORKDIR /app
+
+# Copy the scripts and data directories into the image
+COPY scripts /app/scripts
+COPY data /app/data
+
+# Install Python dependencies
+RUN pip install --no-cache-dir requests

tests/algorithm/README.md

@@ -0,0 +1,104 @@
+# Record Linkage Algorithm Testing
+
+This repository contains a project to evaluate the match accuracy performance of the RecordLinker algorithm.
+
+## Prerequisites
+
+Before getting started, ensure you have the following installed:
+
+- [Docker](https://docs.docker.com/engine/install/)
+- [Docker Compose](https://docs.docker.com/compose/install/)
+
+## Directory Structure
+
+- `/`: Contains the `.env` file and `Dockerfile` to build
+- `configurations/`: Contains the configuration `.json` file that will be used for the test
+- `data/`: Contains the data `.csv` files used for the algorithm test (seed file and test file)
+- `results/`: Contains the results `.csv` file after running the test
+- `scripts/`: Contains the scripts to run the test
+
+## Setup
+
+1. Build the Docker images:
+
+    ```bash
+    docker compose --profile algo-test build
+    ```
+
+2. Add seed and test data files
+    You can use the sample data files provided in the `data` directory or add your own data files.
+    The format of the input files should be a CSV file with the same column headers as shown in the sample files.
+
+        `/data/sample_seed_data.csv`
+
+        `/data/sample_test_data.csv`
+
+
+3. Configure environment variables
+
+    `/algo.env`
+
+    Edit the environment variables in the file
+
+4. Edit the algorithm configuration file
+
+    `/configurations/algorithm_configuration.json`
+
+    Edit the configuration file to tune the algorithm parameters
+
+## Running Algorithm Tests
+
+1. Run the test
+
+    ```bash
+    docker compose run --rm algo-test-runner scripts/run_test.py
+    ```
+
+2. Analyze the results
+
+    The results of the algorithm tests will be available in the `results/output.csv` file.
+
+    The results will be in a CSV formatted file with the following columns: 
+    `Test Case #`, `Expected Result`, `Match Result`, `Details`
+
+## Rerunning Algorithm Tests
+
+After you've run the algorithm tests, you may want to rerun the tests with different seed data, test data, or configurations.
+
+Edit the csv files and/or the configuration file as needed and then run the following commands to rerun the tests.
+
+1. Reset the mpi database
+
+    ```bash
+    docker compose run --rm algo-test-runner python scripts/reset_db.py
+    ```
+2. Run the tests
+
+    ```bash
+    docker compose run --rm algo-test-runner scripts/run_test.py
+    ```
+
+## Environment Variables
+
+1. `env file`: The attributes that should be tuned for your particular algorithm test,
+    are located in the `algo_test.env` file.
+
+2. `environment`: The attributes that should likely remain static for all algorithm tests are located directly in the `compose.yml` file.
+
+### Algorithm Test Parameters
+
+The following environment variables can be tuned in the `algo-test.env` file:
+
+- `SEED_FILE`: The file containing person data to seed the mpi with
+- `TEST_FILE`: The file containing patient data to test the algorithm with
+- `ALGORITHM_CONFIGURATION`: The file containing the algorithm configuration json
+- `ALGORITHM_NAME`: The name of the algorithm to use (either the name of your `ALGORITHM_CONFIGURATION` or can be the built in `dibbs-basic` or `dibbs-enhanced` algorithms)
+
+
+## Cleanup
+
+After you've finished running algorithm tests and analyzing the results, you can stop and remove the Docker containers by running:
+
+```bash
+docker compose --profile algo-test down
+```

tests/algorithm/algo.env

@@ -0,0 +1,4 @@
+SEED_FILE="data/sample_seed_data.csv"
+TEST_FILE="data/sample_test_data.csv"
+ALGORITHM_CONFIGURATION="configurations/algorithm_configuration.json"
+ALGORITHM_NAME="test-config"

tests/algorithm/configurations/algorithm_configuration.json

@@ -0,0 +1,66 @@
+{
+    "label": "test-config",
+    "description": "test algorithm configuration",
+    "is_default": false,
+    "include_multiple_matches": true,
+    "belongingness_ratio": [0.75, 0.9],
+    "passes": [
+        {
+            "blocking_keys": [
+                "BIRTHDATE"
+            ],
+            "evaluators": [
+                    {
+                        "feature": "FIRST_NAME",
+                        "func": "func:recordlinker.linking.matchers.feature_match_fuzzy_string"
+                    },
+                    {
+                        "feature": "LAST_NAME",
+                        "func": "func:recordlinker.linking.matchers.feature_match_exact"
+                    }
+            ],
+            "rule": "func:recordlinker.linking.matchers.eval_perfect_match",
+            "cluster_ratio": 0.9,
+            "kwargs": {
+                "thresholds": {
+                    "FIRST_NAME": 0.9,
+                    "LAST_NAME": 0.9,
+                    "BIRTHDATE": 0.95,
+                    "ADDRESS": 0.9,
+                    "CITY": 0.92,
+                    "ZIP": 0.95
+                }
+            }
+        },
+        {
+            "blocking_keys": [
+                "ZIP",
+                "FIRST_NAME",
+                "LAST_NAME",
+                "SEX"
+            ],
+            "evaluators": [
+                    {
+                        "feature": "ADDRESS",
+                        "func": "func:recordlinker.linking.matchers.feature_match_fuzzy_string"
+                    },
+                    {
+                        "feature": "BIRTHDATE",
+                        "func": "func:recordlinker.linking.matchers.feature_match_exact"
+                    }
+            ],
+            "rule": "func:recordlinker.linking.matchers.eval_perfect_match",
+            "cluster_ratio": 0.9,
+            "kwargs": {
+                "thresholds": {
+                    "FIRST_NAME": 0.9,
+                    "LAST_NAME": 0.9,
+                    "BIRTHDATE": 0.95,
+                    "ADDRESS": 0.9,
+                    "CITY": 0.92,
+                    "ZIP": 0.95
+                }
+            }
+        }
+    ]
+}

tests/algorithm/data/sample_seed_data.csv

@@ -0,0 +1,6 @@
+Match Id,ID,BIRTHDATE,FIRST,LAST,SUFFIX,MAIDEN,RACE,ETHNICITY,GENDER,ADDRESS,CITY,STATE,COUNTY,ZIP,SSN,
+1,3020167,1951-06-02,Linda,Nash,Sr,Gutierrez,Asian,Hispanic,F,968 Gonzalez Mount,South Emilybury,GU,North Kennethburgh County,93236,675-79-1449,
+2,9488697,1942-08-03,Jose,Singleton,Sr,Ingram,Asian,Hispanic,M,631 Fowler Causeway,Port Williamfurt,IN,Wardburgh County,90637,587-60-3668,
+3,1805504,1963-01-29,Ryan,Lawrence,IV,Armstrong,Black,Non-Hispanic,M,5256 Lisa Light,Port Monica,GA,South Christine County,51813,371-33-0433,
+4,1792678,1950-08-10,Thomas,Brady,II,Cobb,White,Unknown,M,944 Hayes Port,Jonesville,FM,Jonesview County,6015,272-78-9905,
+5,1332302,1972-08-26,Angie,Murphy,Sr,Mcmahon,Black,Non-Hispanic,F,60015 Edward Vista Suite 518,Lake Andreaview,UT,North Rodney County,46540,740-16-5170,

tests/algorithm/data/sample_test_data.csv

@@ -0,0 +1,7 @@
+Test Case #,Match Id,ID,BIRTHDATE,FIRST,LAST,SUFFIX,MAIDEN,RACE,ETHNICITY,GENDER,ADDRESS,CITY,STATE,COUNTY,ZIP,SSN,Expected Result
+1,1,3020167,1951-06-02,Linda,Nash,Jr,Gutierrez,Asian,Hispanic,F,968 Gonzalez Mount,South Emilybury,GU,North Kennethburgh County,93236,675-79-1449,Should be a Match
+2,2,9488697,1942-08-03,Singleton,Jose,Sr,Ingram,Asian,Hispanic,M,631 Fowler Causeway,Port Williamfurt,IN,Wardburgh County,90637,587-60-3668,Should be a Match
+3,3,1805504,1963-01-29,Ryan,Law-rence,IV,Armstrong,Black,Non-Hispanic,M,5256 Lisa Light,Port Monica,GA,South Christine County,51813,371-33-0433,Should be a Match
+4,4,1792678,1950-08-10,Tho-mas,Brady,II,Cobb,White,Unknown,M,944 Hayes Port,Jonesville,FM,Jonesview County,6015,272-78-9905,Should be a Match
+5,4,1792678,1950-08-10,ThoMas,Brady,II,Cobb,White,Unknown,M,944 Hayes Port,Jonesville,FM,Jonesview County,6015,272-78-9905,Should be a Match
+6,0,1792679,1950-18-10,ThoMas,Brady,II,Cobb,White,Unknown,M,944 Hayes Port,Jonesville,FM,Jonesview County,6015,272-78-9905,Should fail

tests/algorithm/scripts/helpers.py

@@ -0,0 +1,43 @@
+import json
+
+
+def dict_to_pii(record_data) -> dict | None:
+    # convert row to a pii_record
+    pii_record = {
+        "external_id": record_data.get('ID', None),
+        "birth_date": record_data.get("BIRTHDATE", None),
+        "sex": record_data.get("GENDER", None),
+        "address": [
+            {
+                "line": [record_data.get("ADDRESS", None)],
+                "city": record_data.get("CITY", None),
+                "state": record_data.get("STATE", None),
+                "county": record_data.get("COUNTY", None),
+                "postal_code": str(record_data.get("ZIP", ""))
+            }
+        ],
+        "name": [
+            {
+                "given": [record_data.get("FIRST", None)],
+                "family": record_data.get("LAST", None),
+                "suffix": [record_data.get("SUFFIX", None)]
+            }
+        ],
+        "ssn": record_data.get("SSN", None),
+        "race": record_data.get("RACE", None)
+    }
+
+    return pii_record
+
+
+def load_json(file_path: str) -> dict | None:
+    """
+    Load JSON data from a file.
+    """
+    with open(file_path, "rb") as fobj:
+        try:
+            content = json.load(fobj)
+            return content
+        except json.JSONDecodeError as exc:
+            print(f"Error loading JSON file: {exc}")
+            return None

tests/algorithm/scripts/reset_db.py

@@ -0,0 +1,18 @@
+import os
+
+import requests
+
+
+def reset_db(api_url):
+    print("Resetting the database...")
+    try:
+        response = requests.delete(f"{api_url}/seed") 
+        response.raise_for_status()  # Raise an error for bad status codes
+        print("Database reset successfully")
+    except requests.exceptions.RequestException as e:
+        print(f"Failed to reset the database: {e}")
+
+
+if __name__ == "__main__":
+    api_url = os.getenv("API_URL")
+    reset_db(api_url)

tests/algorithm/scripts/run_test.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python3
+
+import os
+
+from helpers import load_json
+from seed_db import seed_database
+from send_test_records import send_test_records
+from set_configuration import add_configuration
+from set_configuration import check_if_config_already_exists
+from set_configuration import update_configuration
+
+
+def main():  
+    # Get the environment variables
+    api_url = os.getenv("API_URL")
+    algorithm_name = os.getenv("ALGORITHM_NAME")
+    algorithm_config_file = os.getenv("ALGORITHM_CONFIGURATION")
+    seed_csv = os.getenv("SEED_FILE")
+    test_csv = os.getenv("TEST_FILE")
+
+    # setup the algorithm configuration
+    algorithm_config = load_json(algorithm_config_file)
+    if check_if_config_already_exists(algorithm_config, api_url):
+        update_configuration(algorithm_config, api_url)
+    else:
+        add_configuration(algorithm_config, api_url)
+
+    seed_database(seed_csv, api_url)
+
+    send_test_records(test_csv, algorithm_name, api_url)
+
+if __name__ == "__main__":
+    main()

tests/algorithm/scripts/seed_db.py

@@ -0,0 +1,43 @@
+import csv
+
+import requests
+from helpers import dict_to_pii
+
+
+def seed_database(csv_file, api_url):
+    MAX_CLUSTERS = 100
+    cluster_group = []
+
+    print("Seeding the database...")
+
+     # Read the CSV file using the csv module
+    with open(csv_file, mode='r', newline='', encoding='utf-8') as file:
+        reader = csv.DictReader(file)
+
+        for row in reader:
+            record_data = {k: ("" if v in [None, "NaN"] else v) for k, v in row.items()}
+
+            # convert dict to a pii_record
+            pii_record = dict_to_pii(record_data)
+
+            # nesting for the seeding api request
+            cluster = {"records": [pii_record]}
+            cluster_group.append(cluster)
+
+            if len(cluster_group) == MAX_CLUSTERS:
+                send_clusters_to_api(cluster_group, api_url)
+                cluster_group = []
+
+    if cluster_group:
+        send_clusters_to_api(cluster_group, api_url)
+
+    print("Finished seeding the database.")
+
+
+def send_clusters_to_api(cluster_group, api_url):
+    """Helper function to send a batch of clusters to the API."""
+    try:
+        response = requests.post(f"{api_url}/seed", json={"clusters": cluster_group})
+        response.raise_for_status()  # Raise an error for bad status codes
+    except requests.exceptions.RequestException as e:
+        print(f"Failed to post batch: {e}")