This preprint reports a case–control style analysis of 107 adolescents with emotional disorders, subdivided into 54 individuals with current non-suicidal self-injury (NSSI) and 53 individuals without NSSI. The study was conducted at Shanghai Mental Health Center, Shanghai Jiao Tong University School of Medicine. Ethical approval was granted by the Ethics Committee of Shanghai Mental Health Center (approval no. 2022-74C3). The authors state this manuscript is a preprint and has not been certified by peer review; results should not be used to guide clinical practice until validated.
The investigators aimed to identify peripheral biological signatures associated with current NSSI by integrating multiple molecular and clinical data layers. The study combined metabolomic, inflammatory, and routine clinical blood measures with genome-derived variables. Genetic-context features included a polygenic risk score and measures of rare variant burden to provide genome-aware information for modeling.
The authors developed a genome-aware multi-omics modeling framework to classify current NSSI status. The framework fused heterogeneous feature types across molecular and clinical domains rather than relying on any single biomarker layer. The genome-derived metrics were incorporated as contextual variables to inform the model about inherited risk backgrounds.
Modeling included repeated runs to evaluate the stability of selected features and to identify a subset of markers consistently prioritized by the fusion approach. The authors emphasize that distributed multi-omics patterns, rather than single-layer effects, better represented NSSI status in their dataset.
The integrated fusion model achieved the strongest classification performance reported by the authors, with a mean area under the receiver operating characteristic curve (AUC = 0.811). The fusion model outperformed single-omics alternatives, indicating that combining metabolomic, inflammatory, clinical blood, and genome-derived features improved discrimination of current NSSI status.
Specific performance metrics beyond the mean AUC (for example sensitivity, specificity, confidence intervals, or cross-validation procedure details) were not reported in the abstract. Likewise, the precise performance of individual single-omics models and their numeric comparisons to the fusion model were not detailed in the abstract and would require consulting the full preprint for exact values and methodology.
Repeated modeling prioritized 42 stable features that contributed to classification. The authors note that many of these prioritized features were not significant in conventional univariate testing, underscoring the value of multivariate, integrative approaches in detecting distributed signals.
To improve clinical interpretability, the authors derived a compact sentinel marker panel intended to translate the comprehensive model into clinically readable profiles. The abstract does not list the specific markers included in the sentinel panel or their effect directions; those details were not reported in the abstract and would be found in the full manuscript or supplementary materials.
Beyond classification, the authors reconstructed group-specific peripheral networks and identified a pattern of peripheral reorganization associated with current NSSI. Specifically, they report that several modules observed in non-NSSI participants converged into an NSSI-associated module. This NSSI-associated module reportedly linked signals of inflammatory recruitment with indications of weaker immune communication and reduced repair and support-related signals.
These network-level observations suggest that NSSI status may be reflected in an altered peripheral state characterized by coordinated changes across inflammatory, immune-support, and repair-related pathways. The abstract does not provide specific node identities, pathway names, or quantitative network metrics; such details would require examination of the full preprint.
The study also included exploratory analyses connecting the peripheral multi-omics signals to other biological domains. Reporting indicates that exploratory MRI, gut-related, and stress-endocrine analyses were performed and provided additional biological anchors for the peripheral findings. The abstract presents these as supplementary or contextual analyses; the specific results, measures used, and statistical outcomes for these exploratory assessments were not described in the abstract.
Therefore, interpretation of how central nervous system imaging, gut-related measures, or endocrine stress markers relate to the peripheral multi-omics signature requires direct review of the full manuscript.
The authors declare no competing interests. Ethical approval was obtained from the Ethics Committee of Shanghai Mental Health Center (approval no. 2022-74C3). The preprint status indicates the findings are preliminary and have not undergone peer review.
Limitations implicit in the abstract include sample size constraints (107 participants) and the exploratory nature of ancillary analyses. The abstract does not report external validation or replication in independent cohorts; it instead frames the results as a framework for future validation of multi-omics state markers in adolescent emotional disorders.
Next steps suggested by the authors include validating the distributed, genome-aware peripheral signature in additional samples and further evaluating the compact sentinel panel for clinical translation. Detailed methodological descriptions, full feature lists, network graphs, and quantitative outcomes for exploratory analyses were not provided in the abstract and should be consulted in the full preprint for implementation, replication, or clinical application planning.