J. Qiu

Abnormal functional lateralization and cooperation in primary angle-closure glaucoma correlate with cell type-specific transcriptional signatures.

BACKGROUND Primary angle-closure glaucoma (PACG) damages retinal ganglion cells (RGCs) and is associated with neurodegeneration. This study used resting-state functional magnetic resonance imaging (fMRI) to analyze hemispheric lateralization and cooperative functional alterations in PACG. Machine learning assessed the classification efficacy of neuroimaging indicators, while integrated transcriptomics described spatial relationships between neuroimaging changes and genes, neurotransmitters, and cell types. METHODS Resting-state fMRI data were collected from 101 subjects (44 patients with PACG and 57 controls). Whole-brain maps of the Autonomy Index and connectivity of functional homotopic voxels (CFH) were constructed. Their classification efficacy was assessed using five machine learning classifiers. Partial Least Squares (PLS) spatial correlation analysis examined relationships between neuroimaging maps and gene-transcript profiles, cell type density, and neurotransmitter-receptor distribution maps. RESULTS PACG showed increased Autonomy Index in cerebellar Crus II and the right paracentral lobule, with widespread reductions in interhemispheric cooperation. Five machine learning models yielded classification results. Image-transcriptome analysis showed that positive Autonomy Index and CFH-positive gene features were enriched in synaptic signaling and neurodevelopment, whereas negative Autonomy Index and CFH-negative gene profiles correlated with neurodegeneration, metabolic dysfunction, and vascular-immune responses. Oligodendrocytes and endothelial cells showed spatial associations with the Autonomy Index and CFH. Brain lateralization and interhemispheric cooperation showed spatial correlations with VAChT topology and several neurotransmitter receptor/transporter maps. CONCLUSION PACG is characterized by enhanced cerebral lateralization and diminished interhemispheric cooperation. Explainable machine learning evaluated the classification efficacy of these imaging features. These findings provide a multimodal spatial-association framework relating PACG-related imaging alterations to normative molecular and cellular brain maps.

Genetic architecture of treatment-resistant schizophrenia across East Asian and European cohorts: insights from GWAS, TWAS, and synaptic pathway analyses

A software package for simple and rigorous survival machine learning analysis in biomedical research

Survival analysis is a fundamental technique in biomedical research for modeling time-to-event data. It enables the identification of prognostic factors in disease, compares survival outcomes across treatment groups, and performs targeted treatment selection. A variety of machine learning (ML) approaches to survival analysis have emerged to complement classical statistical methods, especially for high-dimensional datasets with complex, nonlinear interactions between features. However, using survival ML methods requires addressing challenges such as censoring-unaware evaluation, overfitting, selecting performance metrics, and data leakage. To address these and other difficulties in using survival ML models, we developed the mlsurv software package. mlsurv is an open-source Python package built around three major design principles: 1) methodological rigor, including evidence-based model selection, leakage-free pipelines, and multi-metric evaluation, 2) multi-scale evaluation and interpretation, including population and subpopulation evaluation, patient-level explanations, and feature analysis, and 3) automated trust and transparency, including limitation flagging and TRIPOD+AI-aligned reporting. mlsurv bundles ten models spanning linear, ensemble, kernel, and deep learning families within a unified software package. We demonstrate mlsurv on the Chowell immunotherapy cohort (n=1,479). The survival-trained models achieve a test concordance index of 0.73 for overall survival prediction. Further, risk scores strongly correlate with the response-trained LORIS clinical score (|{rho}| up to 0.84), reflecting the overlap between prognostic and predictive signal. mlsurv enables biomedical researchers to conduct rigorous, multi-model survival analysis and benchmarking using minimal code with default best practices rather than implementing custom scripts and methodological safeguards from scratch.