Model complexity does not guarantee better health misinformation detection
Does Neural Complexity Improve Health Misinformation Detection? A Leakage-Controlled Cross-Corpus Benchmark
Computation and LanguageMachine Learning
Summary
Detecting false health information online is important but tricky. This study tested several types of neural network models and simpler methods on two health misinformation datasets, carefully controlling for duplicate content. The authors found that more complex models did not consistently perform better, and sometimes simpler models worked just as well or better. Their results suggest that the data and evaluation setup matter more than model complexity for this task.
What this means in practice
- •For health data teams: Select simpler machine learning models for health misinformation detection to ensure stable and interpretable results across varied datasets.
- •For social media platform developers: Design misinformation detection tools using reliable benchmarks that control data leakage rather than focusing solely on complex model architectures.
Authors
Mkululi SIKOSANA
Abstract
Increasing architectural complexity is often assumed to improve health misinformation detection, yet reported gains are difficult to interpret when studies use different corpora, preprocessing pipelines, data splits, and leakage controls. This study provides a controlled cross-corpus benchmark of five compact neural architectures (1D-CNN, LSTM, BiLSTM, CNN-LSTM, and CNN-BiLSTM), a soft-voting neural ensemble, and three classical machine-learning baselines using COVID19-FNIR and CONSTRAINT. Exact-text duplicate controls were applied before modelling; all neural systems used a common preprocessing and optimisation protocol, and neural results were repeated across three random seeds. On COVID19-FNIR, the deep ensemble achieved a mean macro-F1 of 0.9963 and ROC-AUC of 0.9994, while individual neural models ranged from 0.9945 to 0.9957 macro-F1. On CONSTRAINT, the ensemble achieved macro-F1 of 0.9272 and ROC-AUC of 0.9811, whereas a linear SVM achieved macro-F1 of 0.9574 and ROC-AUC of 0.9931. Architecture rankings changed across corpora, and simple sparse linear models remained highly competitive. The findings show that model complexity does not provide a stable performance advantage and that benchmark construction can dominate architecture choice. The study contributes a reproducible, leakage-controlled basis for evidence-driven model selection in health misinformation classification