Souichi Oka

This critique evaluates Monti et al.'s investigation into associations between air pollution, apparent temperature, and schizophrenia severity. While their findings indicate significant short‑ and medium‑term effects of PM10 and thermal stress on PANSS scores, several methodological limitations warrant caution. Their study relies on residential exposure assignments, which may not capture individual mobility or indoor environments, potentially introducing substantial exposure misclassification. Despite appropriately modeling delayed and non-linear effects, the DLNM's reliance on predefined spline structures may oversimplify the complex, synergistic interactions among atmospheric variables. Seasonal discrepancies-such as the absence of PM10 effects in autumn-winter-may reflect unmodeled dependencies or limited pollutant data, particularly for PM2.5 and black carbon. To address these constraints, future research should incorporate flexible, data‑driven approaches, particularly those capable of uncovering latent structures within environmental mixtures. Unsupervised feature‑clustering methods can identify correlated pollutant groupings and reduce dimensional noise, while rank‑based correlation metrics provide robust assessment of non‑linear dependencies that are often obscured by parametric spline specifications. These non‑parametric techniques can complement DLNM by capturing multivariate synergies and interaction patterns that rigid basis structures may overlook. Overall, integrating such approaches is essential for advancing analytical capacity and improving risk assessment for vulnerable psychiatric populations.

Li et al. (CircRM: Profiling circular RNA modifications from nanopore direct RNA sequencing. Brief Bioinform 2026;27:bbaf726.) introduced Circular RNA Modifications (CircRM), a computational framework employing eXtreme Gradient Boosting and SHapley Additive exPlanations (SHAP) to profile RNA modifications in circular RNAs, achieving high predictive accuracy. However, we argue that strong predictive performance does not validate the biological reliability of the resulting feature-importance rankings. In heterogeneous feature spaces, tree-based models exhibit inherent biases, favoring continuous, high-cardinality variables-such as genomic position-over sparse sequence patterns, potentially obscuring true biological determinants. Furthermore, reliance on SHAP introduces theoretical vulnerabilities; recent findings on attribution limitations indicate that baseline sensitivity can decouple explanations from local mechanistic behavior. To address these analytical pitfalls, we advocate for a robust framework incorporating Highly Variable Gene Selection and Feature Agglomeration to mitigate multicollinearity, complemented by model-agnostic non-parametric methods such as Spearman's rho and Kendall's tau. Adopting these strategies ensures that computational profiling yields biologically actionable insights rather than reflecting statistical artifacts.