Multiple Imputation by Chained Equations: What is it and how does it work?

By M. Azur, Elizabeth A. Stuart, C. Frangakis, P. Leaf. International Journal of Methods in Psychiatric Research, 2011

3,291 citations recorded by the source corpus.

Multivariate imputation by chained equations (MICE) has emerged as a principled method of dealing with missing data. Despite properties that make MICE particularly useful for large imputation procedures and advances in software development that now make it accessible to many researchers, many psychiatric researchers have not been trained in these methods and few practical resources exist to guide researchers in the implementation of this technique. This paper provides an introduction to the MICE method with a focus on practical aspects and challenges in using this method. A brief review of software programs available to implement MICE and then analyze multiply imputed data is also provided. Copyright © 2011 John Wiley & Sons, Ltd.

Subfields: Artificial Intelligence, Experimental and Cognitive Psychology, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/8ebeade389cb9508c60f71a3945b906b4db77a59

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yuan Luo
2. Lei Li
3. Yang Feng
4. Peter Szolovits
5. Isabel Valera
6. Hao Zhou
7. Wei Cao
8. Jian Li

In Experimental and Cognitive Psychology

1. Melanie M. Wall
2. Yuanjia Wang

In Statistics and Probability

1. Isabel Valera
2. J. Baron
3. Y. Liu
4. Peter Szolovits
5. Yuan Luo
6. Zoubin Ghahramani
7. Wenjie Du
8. Hao Zhou

Citations in the corpus

• Handling Incomplete Heterogeneous Data using VAEs (2018, methodology): “• MICE: Multiple Imputation by Chained Equations (Azur et al., 2011), which is an iterative method that performs a series of supervised regression models, in which missing data is modeled conditional upon the other variables in the data, including those imputed in previous rounds of the algorithm.”
• Use of Machine Learning to Develop and Evaluate Models Using Preoperative and Intraoperative Data to Identify Risks of Postoperative Complications (2021, neutral): “8) Height, median (IQR), cm 170 (163-178) Weight, median (IQR), kg 83 (69-100) BMI, median (IQR) 28 (24-34) Functional capacity, METS”
• SAITS: Self-Attention-based Imputation for Time Series (2022, background): “Amounts of prior work are proposed to solve this problem with statistics and machine learning methods [2, 15, 16, 17, 18, 19, 20].”
• BRITS: Bidirectional Recurrent Imputation for Time Series (2018, methodology): “Multivariate Imputation by Chained Equations (MICE) [2] first initializes the missing values arbitrarily and then estimates each missing variable based on the chain equations.”
• 3D-MICE: integration of cross-sectional and longitudinal imputation for multi-analyte longitudinal clinical data (2017, positive): “of each missing data element by first performing linear regression to predict the missing values for one variable using the nonmissing and the imputed (or, in the first iteration, initialized) values of the other variables, and then repeating this process to predict missing results for the other var”
• VARIABLE SELECTION AND PREDICTION WITH INCOMPLETE HIGH-DIMENSIONAL DATA (2016, methodology): “As initial values, MICE imputes every missing value of a variable by the mean of observed values or a simple random draw from the data.”
• Predicting Peptide-MHC Binding Affinities with Imputed Training Data (2016, methodology): “For each allele, we train a MHCflurry model using the measured peptide affinities for the allele and the values imputed by MICE based on other alleles in the training set.”