Background: Traumatic brain injury (TBI) and depression are major global health burdens, yet their bidirectional causal relationship remains unclear. Objective: To explore the causal relationship between depression and TBI, and to clarify whether depression is one of the potential risk factors for TBI and whether TBI is one of the pathogenic factors for depression. Methods: This bidirectional two‐sample Mendelian randomization (MR) analysis investigated causal relationships between depression ( n = 170,756) and TBI ( n = 3193) using genome‐wide association study (GWAS) summary statistics. Genetic instruments were selected as single nucleotide polymorphisms (SNPs) significantly associated with exposures (depression/TBI) and outcomes (TBI/depression) at genome‐wide significance ( P < 5 × 10⁻⁶). The inverse variance weighted (IVW) method under fixed‐effects and multiplicative random‐effects models served as the primary analytical approach, with Cochran's  Q  test evaluating SNP heterogeneity. To address horizontal pleiotropy, MR‐Egger regression and MR‐PRESSO(MR Pleiotropy RESidual Sum and Outlier)outlier correction were applied. Sensitivity analyses included weighted median, penalized weighted median, maximum likelihood estimation, and leave‐one‐out validation to ensure robustness. All analyses were conducted using the TwoSampleMR package in R (v4.3.2), with effect estimates reported as odds ratios (OR) and 95% confidence intervals (CI). Results: MR analyses revealed bidirectional causal relationships between depression and TBI. In forward analyses, depression increased TBI risk across multiple IVW frameworks (fixed‐effects IVW: OR = 1.137, 95% CI = 1.019–1.271, P = 0.022; multiplicative random‐effects IVW: OR = 1.137, 95% CI = 1.014–1.277, P = 0.028), corroborated by maximum likelihood estimation (OR = 1.137, 95% CI = 1.017–1.274, P = 0.024). Reverse analyses demonstrated TBI's causal effect on depression through IVW models (fixed‐effects: OR = 1.083, 95% CI = 1.036–1.131, P < 0.001; multiplicative random‐effects: OR = 1.083, 95% CI = 1.043–1.124, P < 0.001) and penalized weighted median methods (OR = 1.079, 95% CI = 1.018–1.145, P = 0.011). Robustness was confirmed by null heterogeneity (Cochran's Q: forward P = 0.209, reverse P = 0.596) and absence of horizontal pleiotropy (MR‐PRESSO: forward P = 0.218, reverse P = 0.672; MR‐Egger intercepts: forward P = 0.661, reverse P = 0.874). All effect estimates remained stable in sensitivity analyses, supporting unconfounded causal inference. Conclusion: Our MR analyses robustly demonstrate bidirectional causality: depression is a risk factor for TBI (OR = 1.137, 95% CI = 1.019–1.271), and TBI subsequently increases depression risk (OR = 1.083, 95% CI = 1.036–1.131), advocating integrated clinical monitoring.

Pressure injury management requires coordinated decisions on wound staging, debridement, and dressing selection, yet most computational studies focus on isolated tasks using a single data modality. This study developed a multimodal, multi-task framework integrating wound photographs and clinical text to support pressure injury stage classification, debridement necessity prediction, primary dressing recommendation, and secondary dressing recommendation. Models were trained on an expert-annotated public dataset and externally validated using an independent private cohort. We systematically compared image-only models, text-only models, large language model-based approaches, conventional image–text fusion strategies, and image–language model combinations. Among the evaluated models, ResNet-50 combined with DeepSeek-R1-Distill-Qwen-1.5B showed the best overall balance across classification performance, text-generation quality, and validation robustness. On the validation set, the final model achieved 82.92% accuracy, 81.26% macro-F1, and 99.01% area under the curve for stage classification, and 89.91% accuracy, 89.83% macro-F1, and 96.24% area under the curve for debridement prediction. For dressing recommendation, the model achieved ROUGE-1 scores of 58.30% and 61.06% for primary and secondary dressings, respectively. Interpretability analyses indicated clinically relevant image attention and text attribution patterns. These findings suggest that multimodal learning may provide a clinically aligned decision-support approach for pressure injury assessment and treatment recommendations.

Background/Objectives: Genetic overlap between any ischaemic stroke (AIS) and coronary artery disease (CAD) can obscure associations less closely tied to coronary risk. We aimed to separate coronary-shared from residual AIS genetic effects. Methods: We applied GWAS-by-subtraction to GIGASTROKE AIS and UK Biobank/CARDIoGRAMplusC4D CAD summary statistics. A Cholesky model separated a CAD-related shared component (F1) from a model-defined residual AIS component (F2). We tested model order, factor covariance, population prevalence, CAD input, LD reference and sample-size parameter then characterised loci using FUMA, SuSiE-RSS, genetic correlation, MAGMA, SMR/HEIDI and targeted colocalisation. Results: AIS and CAD had a genetic correlation of 0.490. F1 and F2 had SNP heritabilities of 0.0251 and 0.00414, and F2 contained 14 FUMA loci. F2 Z scores remained correlated across covariance (minimum r = 0.994), prevalence (minimum r = 0.998) and alternative-CAD analyses (r = 0.984). Reverse ordering produced reallocation of source-trait signal. At rs974819, the T allele was associated positively with CAD and negatively with marginal AIS, producing a model-derived CAD-AIS-discordant F2 association. Source-trait colocalisation showed that NBEAL1, PHACTR1 and PDGFD signals were present in one or both input-trait analyses, whereas the F11 protein signal aligned with AIS and F2. In a potentially overlapping FinnGen cross-implementation analysis, all 13 F2 lead variants showed concordant directions. Within overlapping stroke-subtype analyses, F2 showed stronger alignment with cardioembolic stroke. Conclusions: GWAS-by-subtraction resolved a model-defined residual AIS component that was stable across covariance, prevalence and CAD-input specifications, supporting a component-based view of AIS genetic effects after modelling shared coronary architecture.