0
Article ? AI-assigned paper type based on the abstract. Classification may not be perfect — flag errors using the feedback button. Tier 2 ? Original research — experimental, observational, or case-control study. Direct primary evidence. Remediation Sign in to save

Limitations of SHAP-based interpretations in environmental and membrane filtration applications

Water Research 2025 3 citations ? Citation count from OpenAlex, updated daily. May differ slightly from the publisher's own count. Score: 48 ? 0–100 AI score estimating relevance to the microplastics field. Papers below 30 are filtered from public browse.
Yoshiyasu Takefuji

Summary

Researchers critically analyze the use of SHAP values — a machine learning interpretability method — in microplastic filtration studies, arguing that SHAP's dependence on model assumptions and inability to handle correlated variables can produce misleading conclusions about which process parameters matter most in complex environmental systems.

Maliwan et al. (2025) identified key parameters in microplastic ultrafiltration using interpretable machine learning (SHAP), attributing 57.6-70.6 % feature importance to factors like transmembrane pressure. This paper critically examines their methodological approach, highlighting significant concerns regarding SHAP's application. SHAP values are inherently model-dependent and lack ground truth for validating feature importance accuracy, leading to potentially biased and erroneous conclusions; high prediction accuracy does not ensure reliable insights. SHAP's underlying assumptions, particularly feature independence, rarely hold in complex environmental systems characterized by multicollinearity, potentially misattributing variable importance. We advocate for a more robust analytical framework incorporating unsupervised machine learning (e.g., feature agglomeration) and nonlinear nonparametric statistical methods (e.g., Spearman's correlation) to provide more reliable insights into variable relationships, moving beyond model-dependent interpretations.

Sign in to start a discussion.

More Papers Like This

Article Tier 2

Decoding the transport thresholds of emerging contaminants in watersheds using explainable machine learning

Researchers collected 517 water samples from the Huangshui River over four years and used an explainable machine learning framework with SHAP analysis to model how land use, landscape metrics, and climate variables drive the transport of microplastics, antibiotics, and heavy metals through the watershed.

Article Tier 2

Hybrid Ensemble Machine Learning Models with SHAP Explainability for Robust Prediction of Suspended Particle Attachment Efficiency in Complex Environmental Systems

Scientists developed a new computer model that can better predict how tiny particles—including microplastics—clump together and move through the environment. The model found that salt levels in water are the main factor controlling how single particles stick together, while electrical charge differences matter most when different types of particles interact. This research could help us better understand how microplastics and other harmful particles spread through water systems and potentially affect human health.

Article Tier 2

Elucidating microplastic adsorption mechanisms in biomass composite materials through interpretable machine learning

Researchers used interpretable machine learning to study how biomass composite materials adsorb microplastics from water. They found that initial microplastic concentration and surface electrical potential were the most important factors determining adsorption effectiveness. The study demonstrates that data-driven approaches can help design more efficient and sustainable materials for removing microplastics from contaminated water.

Article Tier 2

Membrane filter removal in FTIR spectra through dictionary learning for exploring explainable environmental microplastic analysis

Researchers developed a machine learning method to remove the interfering signal from filter membranes in infrared spectra used to identify microplastics, improving classification accuracy by 1.5-fold and maintaining explainability — making it easier to reliably identify plastic types in environmental water samples collected with filters.

Article Tier 2

Decoding the PlasticPatch: Exploring the Global MicroplasticDistribution in the Surface Layers of Marine Regions with InterpretableMachine Learning

Researchers applied four interpretable machine learning algorithms to a calibrated global marine microplastic dataset to construct a predictive model of surface-layer microplastic distribution, finding that biogeochemical and anthropogenic factors are the dominant drivers of global marine microplastic pollution patterns.

Share this paper