HmmBiasResolver

Introduction

HmmBiasResolver is an Hmm-based model for resolving the bias in Electronic Medical Records (EMR), therefore, improving the performance of EMR data analytics.

Functionality

HmmBiasResolver takes raw array-like data as input, fills the missing data with an Hmm-based model and then outputs the transformed data in the same shape.

Input:

• raw array-like data with shape (n_patients, n_timewindows, n_features)
• value range {-1, 0, +1}, with -1 "abnormal", "0" missing, "+1" normal

Output:

• transformed array-like data with shape (n_patients, n_timewindows, n_features)
• value range [-1, +1]

Requirements

hmmlearn==0.2.0
numpy==1.13.3
progressbar==2.3

Reference

K. Zheng, J. Gao, K. Y. Ngiam, B. C. Ooi, and W.L.J. Yip.
Resolving the Bias in Electronic Medical Records.
Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (SIGKDD), pages 2171-2180, 2017.