Comparison of Imputation Strategies for Incomplete Electronic Health Data
Health Data Science · 01.01.2026 · 6
DOI: 10.34133/hds.0493 ↗Kurz gefasst
Fachartikel in der Zeitschrift Health Data Science, online erschienen am 02.11.2026. Verfasst von Shuo Zhang und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Science, Tianjin University of Technology, Tianjin, China. Die Arbeit nennt 33 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Tianjin Municipality, Peking University People’s Hospital Research And Development Funds.
Abstract (Originalsprache)
Background
Missing data are a common challenge in clinical tabular datasets and can substantially affect both statistical analyses and predictive modeling in healthcare.
Methods
In this study, we conducted an empirical evaluation of 8 imputation strategies—median imputation, median imputation with missing indicator, multiple imputation by chained equations (MICE), MissForest, generative adversarial imputation network (GAIN), missing data imputation using autoencoders (MIWAE), self-attention-based imputation for time series, and gated recurrent unit with decay—across 3 heterogeneous real-world clinical datasets. The methods were assessed under controlled missingness simulations covering different mechanisms (missing completely at random, missing at random [MAR], and missing not at random [MNAR]) and missingness rates ranging from 10% to 90%. We evaluated imputation quality using multiple statistical metrics and examined its relationship with downstream classification performance across multiple classifiers.
Results
Across the considered settings, MICE and MissForest showed consistently strong and stable performance, while deep generative approaches such as GAIN and MIWAE exhibited higher variability, particularly under the MAR and MNAR mechanisms and at higher missingness rates. Importantly, we observed that improvements in imputation accuracy did not necessarily translate into better downstream predictive performance.
Conclusion
Rather than drawing general conclusions about method superiority, the results highlight the sensitivity of imputation outcomes to data characteristics, missingness mechanisms, and evaluation objectives. These findings underscore the importance of aligning imputation strategy selection with the intended downstream task and data regime when working with heterogeneous clinical tabular data.
Autor:innen und Institutionen
- Department of Computer Science, Tianjin University of Technology, Tianjin, China.
- National Institute of Health Data Science, Peking University, Beijing, China.
- DCST, BNRist, RIIT, Institute of Internet Industry, Tsinghua University, Beijing, China.
- Department of Clinical Epidemiology and Biostatistics, Peking University People's Hospital, Beijing, China.
ORCID-Profile: Shuo Zhang
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Tianjin Municipality
- Peking University People’s Hospital Research And Development Funds
- Tianchi Elite Youth Doctoral Program
- Peking University People's Hospital Research And Development Funds
Zitieren
APA
Zhang, S., Zhang, Z., Zhou, Y., & Liu, H. (2026). Comparison of Imputation Strategies for Incomplete Electronic Health Data. Health Data Science, 6. https://doi.org/10.34133/hds.0493
Vancouver
Zhang S, Zhang Z, Zhou Y, Liu H. Comparison of Imputation Strategies for Incomplete Electronic Health Data. Health Data Science. 2026;6. doi:10.34133/hds.0493
BibTeX
@article{Zhang2026,
author = {Zhang, Shuo and Zhang, Zhilong and Zhou, Yuxi and Liu, Huixin},
title = {Comparison of Imputation Strategies for Incomplete Electronic Health Data},
journal = {Health Data Science},
year = {2026},
volume = {6},
doi = {10.34133/hds.0493}
}
EUVIMED