Accurately predicting the concentration of fine particulate matter (PM_2.5) is crucial for evaluating air pollution levels and public exposure. Recent advancements have seen a significant rise in using deep learning (DL) models for forecasting PM_2.5 concentrations. Nonetheless, there is a lack of unified and standardized frameworks for assessing the performance of DL-based PM_2.5 prediction models. Here we extensively reviewed those DL-based hybrid models for forecasting PM_2.5 levels according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We examined the similarities and differences among various DL models in predicting PM_2.5 by comparing their complexity and effectiveness. We categorized PM_2.5 DL methodologies into seven types based on performance and application conditions, including four types of DL-based models and three types of hybrid learning models. Our research indicates that established deep learning architectures are commonly used and respected for their efficiency. However, many of these models often fall short in terms of innovation and interpretability. Conversely, models hybrid with traditional approaches, like deterministic and statistical models, exhibit high interpretability but compromise on accuracy and speed. Besides, hybrid DL models, representing the pinnacle of innovation among the studied models, encounter issues with interpretability. We introduce a novel three-dimensional evaluation framework, i.e., Dataset-Method-Experiment Standard (DMES) to unify and standardize the evaluation for PM_2.5 predictions using DL models. This review provides a framework for future evaluations of DL-based models, which could inspire researchers to standardize DL model usage in PM_2.5 prediction and improve the quality of related studies.

Preclinical studies have shown that the gut microbiota can play a role in schizophrenia (SCH) pathogenesis via the gut-brain axis. However, its role in the antipsychotic treatment response is unclear. Here, we present a 24-week follow-up study to identify gut microbial biomarkers for SCH diagnosis and treatment response, using a sample of 107 first-episode, drug-naïve SCH patients, and 107 healthy controls (HCs). We collected biological samples at baseline (all participants) and follow-up time points after risperidone treatment (SCH patients). Treatment response was assessed using the Positive and Negative Symptoms Scale total (PANSS-T) score. False discovery rate was used to correct for multiple testing. We found that SCH patients showed lower α-diversity (the Shannon and Simpson’s indices) compared to HCs at baseline (p = 1.21 × 10⁻⁹, 1.23 × 10⁻⁸, respectively). We also found a significant difference in β-diversity between SCH patients and HCs (p = 0.001). At baseline, using microbes that showed different abundance between patients and controls as predictors, a prediction model can distinguish patients from HCs with an area under the curve (AUC) of 0.867. In SCH patients, after 24 weeks of risperidone treatment, we observed an increase of α-diversity toward the basal level of HCs. At the genus level, we observed decreased abundance of Lachnoclostridium (p = 0.019) and increased abundance Romboutsia (p = 0.067). Moreover, the treatment response in SCH patients was significantly associated with the basal levels of Lachnoclostridium and Romboutsia (p = 0.005 and 0.006, respectively). Our results suggest that SCH patients may present characteristic microbiota, and certain microbiota biomarkers may predict treatment response in this patient population.

Schizophrenia (SCZ) is associated with complex crosstalk between the gut microbiota and host metabolism, but the underlying mechanism remains elusive. Investigating the aberrant neurotransmitter processes reflected by alterations identified using multiomics analysis is valuable to fully explain the pathogenesis of SCZ. We conducted an integrative analysis of multiomics data, including the serum metabolome, fecal metagenome, single nucleotide polymorphism data, and neuroimaging data obtained from a cohort of 127 drug-naïve, first-episode SCZ patients and 92 healthy controls to characterize the microbiome–gut–brain axis in SCZ patients. We used pathway-based polygenic risk score (PRS) analyses to determine the biological pathways contributing to genetic risk and mediation effect analyses to determine the important neuroimaging features. Additionally, a random forest model was generated for effective SCZ diagnosis. We found that the altered metabolome and dysregulated microbiome were associated with neuroactive metabolites, including gamma-aminobutyric acid (GABA), tryptophan, and short-chain fatty acids. Further structural and functional magnetic resonance imaging analyses highlighted that gray matter volume and functional connectivity disturbances mediate the relationships between Ruminococcus_torgues and Collinsella_aerofaciens and symptom severity and the relationships between species Lactobacillus_ruminis and differential metabolites l-2,4-diaminobutyric acid and N-acetylserotonin and cognitive function. Moreover, analyses of the Polygenic Risk Score (PRS) support that alterations in GABA and tryptophan neurotransmitter pathways are associated with SCZ risk, and GABA might be a more dominant contributor. This study provides new insights into systematic relationships among genes, metabolism, and the gut microbiota that affect brain functional connectivity, thereby affecting SCZ pathogenesis.