Mammographic Density Evaluator

Script to evaluate mammographic density as a predictor of breast cancer risk. Given density measurements and censored cancer data, the script computes the relative risk, odds ratio, and c-statistic of the density measurements.

Environment

conda env create -f environment.yml
conda activate density

Data Format

An example of the data format is provided in example_input.csv. The data is a CSV file with the following columns:

• PatientID: Unique identifier for each patient
• ExamID: Unique identifier for each mammogram
• ExamDate: Date of the mammogram. The date should be in the format MM/DD/YYYY
• Density: Density measurement
• Age at Exam: Age of the patient at the time of the mammogram
• Ethnicity: Ethnicity of the patient
• Cancer (YES | NO): Whether the patient was diagnosed with cancer within T years of the mammogram
• Date of Cancer Diagnosis: Date of cancer diagnosis. The date should be in the format MM/DD/YYYY
• Date of Last Negative Mammogram: Last known date of a negative mammogram (i.e., date of the most recent negative mammogram). The date should be in the format MM/DD/YYYY

Usage

To compute the relative risk and odds ratio:

python evaluate_density.py \
-i example_input.csv \
--max_followup 5 \
--density_cutoff 3

To evaluate the performance of the density measurements on different age groups:

python evaluate_density.py \
-i example_input.csv \
--max_followup 4 \ # Maximum follow-up time in years, indexed at 0
--density_cutoff 3 \
--age_groups 40-50 50-60 60-70

To format and save the data with deidentified IDs add the arguments --deidentify_and_save --seed 123. The seed is required to generate the deidentified IDs in a random manner.

Output

The script outputs the relative risk and odds ratio of the density measurements. The output is saved in a CSV file with the same name as the input file, but with _results appended to the name. For example, if the input file is example_input.csv, the output file will be example_input_results.csv.