sebp / causal-effects-in-alzheimers-continuum

Code for the paper "Estimation of Causal Effects in the Presence of Unobserved Confounding in the Alzheimer's Continuum"

Estimation of Causal Effects in the Presence of Unobserved Confounding in the Alzheimer's Continuum

This repository contains the code to the paper "Estimation of Causal Effects in the Presence of Unobserved Confounding in the Alzheimer's Continuum." If you are using this code, please cite:

@inproceedings(Poelsterl2021-causal-effects-in-ad,
  title     = {{Estimation of Causal Effects in the Presence of Unobserved Confounding in the Alzheimer's Continuum}},
  author    = {P{\"{o}}lsterl, Sebastian and Wachinger, Christian},
  booktitle = {Information Processing in Medical Imaging},
  year      = {2021},
  pages     = {45--57},
  doi       = {10.1007/978-3-030-78191-0_4},
  url       = {https://arxiv.org/abs/2006.13135},
}

Installation

1. Install Docker.
2. Build Docker image causalad:

docker build -t causalad .

If you want to use the code for development, you can use conda to create an environment with all dependencies from requirements.yaml.

Data

This section provides an overview on how to obtain the data to reproduce the results presented in the paper. As data cannot be shared publicly, you will have to perform the data processing yourself to fill in the missing values of data/adni-data-template.csv and data/ukb-data-template.csv. These files list the patient ID, and visit and image ID for ADNI, which uniquely identify the data you need to obtain. We expect that you have been approved to access the data and are familiar with the data portals of ADNI and UK Biobank.

Alzheimer’s Disease Neuroimaging Initiative (ADNI)

1. Log in to the ADNI Data Portal.
2. Download ADNIMERGE.CSV, APOERES.csv and UPENNBIOMK_MASTER.csv.
3. Copy APOERES.csv to the data/ directory.
4. Use ABETA, PTAU, TAU from UPENNBIOMK_MASTER.csv to determine which patients have an Alzheimer's pathologic by creating a column ATN_status that describes the A/T/N scheme, e.g. A+/T+/N- if ABETA ≤ 192, PTAU ≥ 23, and TAU < 93, following the thresholds from Ekman et al., 2018:

The individual CSF values were considered pathological (+) if ≤192 pg/ml for Aβ42, ≥93 pg/ml for t-tau, and ≥23 pg/ml for p-tau.

5. Download T1 structural brain MRI from the ADNI Data Portal and segment each with FreeSurfer 5.3 to obtain volume and thickness measurements.
6. Fill in the values of data/adni-data-template.csv by taking ABETA, PTAU, TAU from UPENNBIOMK_MASTER.csv, ATN_status from above, volume and thickness measurements computed by FreeSurfer, and the remaining variables from ADNIMERGE.CSV. Save the resulting file as data/adni-data.csv.

UK Biobank (UKB)

1. Log in to the UK Biobank Access Management System.
2. Download data on Sex, and Age at first imaging visit.
3. Download T1 structural brain MRI and segment each with FreeSurfer 5.3 to obtain volume measurements.
4. Fill in the values of the data/ukb-data-template.csv and save the result as data/ukb-data.csv.

Estimating Causal Effects

Effect of Neuroanatomy on ADAS13

1. Make sure your created data/adni-data.csv as outlined above.
2. The workflow is split into 3 notebooks that have to be executed sequentially. Begin by starting the Jupyter notebook server:

docker run --rm -p 8888:8888 -v $(pwd)/data:/notebooks/data causalad

3. Go to https://localhost:8888 or click on the link that is printed when running the above command.

4. Click on the adni-estimate-substitute.ipynb notebook, which will open a new tab. In the menu at the top, go to Cell and click Run All. Once completed, this will create data/adni-transformed.csv and 4 files in data/outputs/adni/dim6/:

   1. adni_aug_BPMF.csv: Transformed features with 6 substitute confounders estimated by BPMF.
   2. adni_aug_PPCA.csv: Transformed features with 6 substitute confounders estimated by PPCA.
   3. adni_aug_regressout.csv: Transformed features with observed confounders regressed out by linear regression.
   4. pvalue.csv: Bayesian p-value of posterior predictive check for BPMF and PPCA.

5. Go back to https://localhost:8888, and click on the adni-estimate-effects.ipynb notebook. In the new tab, go to Cell and click Run All. Once completed, 4 additional files are created in data/outputs/adni/dim6/:

   1. coef_bpmf.csv: Estimated coefficients of Beta-regression model when accounting for observed confounders and 6 substitute confounders estimated by BPMF.
   2. coef_noconf.csv: Estimated coefficients of Beta-regression model when ignoring confounding.
   3. coef_ppca.csv: Estimated coefficients of Beta-regression model when accounting for observed confounders and 6 substitute confounders estimated by PPCA.
   4. coef_regout.csv: Estimated coefficients of Beta-regression model when accounting for observed confounders via the regress-out approach.

6. Finally, go back to https://localhost:8888 and open the adni-compare-effects.ipynb notebook. In the new tab, go to Cell and click Run All. This will display a figure comparing the estimated credible intervals for each model from the previous step. In addition, it writes the figure to data/outputs/adni/dim6/coef-horizontal.pdf.

Semi-Synthetic Simulation Study

1. Make sure your created data/ukb-data.csv as outlined above.
2. To execute all steps of the simulation study, you will need at least 64GB of RAM. The entire process takes approximately 4 hours and can be started by executing:

docker run --rm -v $(pwd)/data:/notebooks/data causalad ./run-synthetic-ukb.sh

3. This will create 4 files in data/outputs/synthetic_ukb/.

   1. augmented_data_bpmf_dim5.pkl: Transformed features with 5 substitute confounders estimated by BPMF.
   2. augmented_data_ppca_dim5.pkl: Transformed features with 5 substitute confounders estimated by PPCA.
   3. coefs_dim5.pkl: Estimated coefficients of all models for 1000 different simulated outcomes.
   4. evaluation_coefs_dim5.csv: RMSE of coefficients with respect to true coefficients for each model, across 1000 simulations.