How Deep is your Guess? A Fresh Perspective on Deep Learning for Medical Time-Series Imputation

TL;DR

This paper analyzes how deep learning models for EHR time-series imputation are affected by architectural biases, emphasizing the importance of design choices and clinical relevance over mere model complexity.

Contribution

It provides a comprehensive analysis of deep learning architectures for medical time-series imputation, highlighting biases, performance dependencies, and the need for clinically meaningful approaches.

Findings

Model effectiveness depends on bias alignment with data characteristics.

Larger models do not always outperform smaller, well-designed architectures.

Imputation performance varies up to 20\% based on preprocessing and implementation.

Abstract

We present a comprehensive analysis of deep learning approaches for Electronic Health Record (EHR) time-series imputation, examining how architectural and framework biases combine to influence model performance. Our investigation reveals varying capabilities of deep imputers in capturing complex spatiotemporal dependencies within EHRs, and that model effectiveness depends on how its combined biases align with medical time-series characteristics. Our experimental evaluation challenges common assumptions about model complexity, demonstrating that larger models do not necessarily improve performance. Rather, carefully designed architectures can better capture the complex patterns inherent in clinical data. The study highlights the need for imputation approaches that prioritise clinically meaningful data reconstruction over statistical accuracy. Our experiments show imputation performance variations of up to 20\% based on preprocessing and implementation choices, emphasising the need for standardised benchmarking methodologies. Finally, we identify critical gaps between current deep imputation methods and medical requirements, highlighting the importance of integrating clinical insights to achieve more reliable imputation approaches for healthcare applications.