Related Experiment Video
Updated: Oct 5, 2025
05:19
Author Spotlight: Therapeutic Benefit of Closed-Loop Deep Brain Stimulation in Depression Treatment
Published on: July 7, 2023
2.6K
The Fitting Optimization Path Analysis on Scale Missing Data: Based on the 507 Patients of Poststroke Depression
Xiaoying Lv1,2, Ruonan Zhao3, Tongsheng Su4
1School of Chinese Medicine, Beijing University of Chinese Medicine, Beijing, China.
Evidence-Based Complementary and Alternative Medicine : Ecam
|January 24, 2022
Summary
This study identifies optimal methods for handling missing patient data in scales. Random Forest Regression (RFR) is often superior, especially with higher missing data rates, improving clinical research efficiency.
Area of Science:
- Biostatistics
- Clinical Data Science
- Health Informatics
Background:
- Missing data in patient scales poses challenges for accurate analysis.
- Understanding missing data mechanisms (MCAR, MAR, MNAR) is crucial for appropriate imputation.
Purpose of the Study:
- To determine the optimal data imputation strategies for missing values in patient scales.
- To evaluate different methods under various missing data scenarios.
Main Methods:
- Simulated missing data (5-40%) using MCAR, MAR, and MNAR mechanisms in a stroke patient dataset (n=507).
- Applied Mean Substitution (MS), Random Forest Regression (RFR), and Predictive Mean Matching (PMM) for data imputation.
- Evaluated imputation performance using Root Mean Square Error (RMSE), 95% Confidence Interval width, and Spearman Correlation Coefficient (SCC).
Main Results:
- Under MCAR, MS is suitable for <20% missing data; RFR is best for >20%.
- Under MAR, MS is convenient for <35% missing data; RFR performs better for >35%.
- Under MNAR, RFR is the optimal method, particularly for >30% missing data.
Conclusions:
- The choice of imputation method depends on the missing data mechanism and proportion.
- RFR offers a robust solution, expanding sample utility and reducing clinical research costs.
- Selecting the best method requires considering statistical expertise, method efficacy, and reader comprehension.
