/ImputeEHR

Explore the most effective imputation strategies for EHR data

Primary LanguagePythonMIT LicenseMIT

ImputeEHR

Explore the most effective imputation strategies for EHR data

The example dataset is TJH dataset, preprocessed following pyehr scripts.

Imputation Methods

Rule-based

  • LOCFImpute (last-observation-carried-forward)
  • ZeroImpute
  • MeanImpute
  • MedianImpute
  • ModeImpute
  • SimpleImpute (GRU-D Simple, concatenating the measurement with masking and time interval)

ML-based

  • InterpolationImpute
  • SmoothingImpute
  • SplineImpute
  • KNNImpute
  • MICEImpute (multiple imputation by chained equations)
  • RFImpute (Random Forest, e.g. MissForest)
  • MFImpute (matrix factorization)
  • PCAImpute
  • SoftImpute

DL-based

  • MLPImpute
  • RNNImpute
  • AEImpute (AutoEncoder)
  • GANImpute