Estimating Average Causal Effects from Patient Trajectories
Authors: Dennis Frauen, Tobias Hatt, Valentyn Melnychuk, Stefan Feuerriegel
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We compare Deep ACE in an extensive number of experiments, confirming that it achieves state-of-the-art performance. We further provide a case study for patients suffering from low back pain to demonstrate that Deep ACE generates important and meaningful findings for clinical practice. |
| Researcher Affiliation | Academia | 1 LMU Munich 2 Munich Center for Machine Learning 3 ETH Zurich |
| Pseudocode | Yes | Algorithm 1: Iterative G-computation (Robins 1999; van der Laan and Rose 2018) |
| Open Source Code | Yes | 1Code available at https://github.com/Dennis Frauen/Deep ACE. |
| Open Datasets | Yes | For this purpose, we use the MIMIC-III dataset (Johnson et al. 2016), which includes electronic health records from patients admitted to intensive care units. |
| Dataset Splits | No | The paper mentions generating synthetic and semi-synthetic data for experiments and refers to the Appendix for details regarding data generation and method evaluation, but it does not specify concrete train/validation/test splits, percentages, or sample counts in the main text. Details on implementation, training, and hyperparameter tuning are also noted to be in the Appendix. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running experiments, such as GPU models, CPU types, or memory specifications. |
| Software Dependencies | No | The paper mentions using an LSTM layer and refers to implementation details in the Appendix, but it does not list specific software dependencies with version numbers (e.g., library names like PyTorch or TensorFlow with their versions). |
| Experiment Setup | No | The paper explicitly states, 'Details on our implementation, training, and hyperparameter tuning are in the Appendix.' Therefore, specific experimental setup details are not provided in the main text. |