Citations

Purpose: Depression is prevalent yet frequently underdiagnosed. Although speech-based detection methods show promise, most prior studies have emphasized binary classification in nontonal languages. This study examined whether acoustic features of spontaneous Mandarin speech reflect depression as a categorical condition or a continuous, dimensional construct. Method: A validated emotional Mandarin speech corpus paired with self-reported depression severity scores was analyzed. Acoustic features were extracted using the extended Geneva Minimalistic Acoustic Parameter Set. A multistage analytic framework was applied, including random forest classification to distinguish depressed from nondepressed clips, linear mixed-effects modeling to examine group and severity effects, and unsupervised clustering to examine latent structure in the acoustic feature space. Results: The random forest classifier achieved an accuracy of 72.3% and an area under the receiver operating characteristic curve of 0.826 on the held-out test set. Model performance favored sensitivity over precision in identifying depressed clips. SHapley Additive exPlanations analysis identified features across multiple acoustic domains, including spectral, pitch-related, voice quality, cepstral, and intensity measures, as important contributors to model predictions. Among the Top 10 features, no acoustic features showed statistically significant group differences. In contrast, median and mean fundamental frequencies were significantly positively associated with depression severity. Clustering analysis revealed overlapping group structures and a continuous distribution of severity scores across clusters. Conclusions: These findings provide preliminary support for a dimensional conceptualization of depression in spontaneous Mandarin speech. While classification models can distinguish depressed from nondepressed speech with moderate accuracy, acoustic features appear to vary more consistently with symptom severity than with diagnostic group. Speech-based measures may therefore have potential for continuous monitoring of depression, although this possibility requires further longitudinal verification.

Background: Neurological diseases share overlapping systemic and molecular features, but the specificity and interpretation of cross-disease metabolomic signatures remain uncertain. Methods: We analyzed summary-level Nightingale NMR metabolomic and genetic data from 274,241 UK Biobank participants across eight neurological endpoints. The disease rankings used here were derived from baseline plasma metabolites predicting future incident disease rather than from contemporaneous diagnostic case–control contrasts. Results: Among 57 priority metabolites with genome-wide significant instruments (2472 SNPs; mean F = 162.4), nine ranked in the top 30 for at least seven of the eight endpoints, defining a shared pre-diagnostic neurological signature confirmed as non-random cross-endpoint convergence by permutation testing (p < 0.0001). Because we did not perform a quantitative non-neurological disease control analysis, this signature should not be interpreted as neurologically specific and may partly reflect systemic morbidity, renal function, body composition, or frailty-related physiology. Forward Mendelian randomization across 45 metabolite–disease pairs found no Bonferroni-significant causal effects; three nominal protective associations are consistent with the number of false-positive findings expected under multiple testing and require replication. Conclusions: These results support a shared systemic, pre-diagnostic metabolic signature across the neurological disease spectrum, while the null MR findings and specificity limitations favor interpretation as a biomarker or prodromal downstream signal rather than a proven causal mechanism. External prospective validation is essential.

Abstract Objective Epilepsy duration is a modifiable risk factor in the outcome of definitive epilepsy surgery; however, an analogous effect in palliative procedures has not been shown. We reviewed the Pediatric Epilepsy Surgery Database data for an association between epilepsy duration and seizure reduction in palliative procedures. Methods Patients enrolled between January 2018 and April 2025 who underwent their first epilepsy surgery with palliative intent with 6 months of follow‐up were included. Procedures included neuromodulation, corpus callosotomy, hemispherotomy, lesionectomy, and lobectomy where surgical intent was not seizure freedom. Outcomes of seizure freedom, 90% seizure reduction, and 50% seizure reduction were considered at 6–12 months and >12 months from surgery. Duration from epilepsy onset to surgery was compared for patients above and below each outcome threshold at each time point. Logistic regression analysis for the association between epilepsy duration and seizure reduction adjusted for potential confounders including procedure type, etiology, and other clinical factors. Logistic regression analysis was performed on the overall cohort and subgroups of patients with each procedure. Results A total of 588 patients were included. Initial univariate analysis suggested that epilepsy duration at time of surgery was significantly associated with seizure freedom and 90% seizure reduction at both 6–12 months and >12 months. After adjusting for confounders, only seizure freedom at >12 months was significantly associated with duration of epilepsy. When individual procedures were considered, only lobectomy was sensitive to duration of epilepsy in multivariate analysis, with significant impacts on >50% and >90% seizure reduction at >12 months. Lesional epilepsy predicted seizure freedom at >12 months. Neuromodulation and corpus callosotomy were less likely to achieve seizure reduction than other procedures. Significance We did not find an association between early epilepsy surgery and seizure reduction. This reflects the heterogeneity of our population, including different types of surgical procedures and lesional and nonlesional epilepsies.

BACKGROUND: Anthracyclines are a widely used and effective class of chemotherapy. However, a major limitation for their use is cardiotoxicity, manifesting as systolic dysfunction, congestive heart failure, and arrhythmias. The objective of this study was to identify genes and genetic risk variants for doxorubicin-induced arrhythmia. METHODS: We performed genome-wide association studies and meta-analysis across 2 independent nonoverlapping, genetically homogenous data sets from the Molecular Epidemiology Resource (cases N=77, controls N=1184) and Tapestry (cases N=13, controls N=172). Doxorubicin-related arrhythmia was the primary cardiac outcome, followed by investigation of a broader phenotype of cardiac events. Effects of doxorubicin on top-associated genes were assessed by quantitative polymerase chain reaction on RNA extracted from human cardiomyocytes following treatment with doxorubicin and by echocardiography in a zebrafish model of doxorubicin-induced cardiomyopathy. RESULTS: In the meta-analysis, rs12151014 at the KCTD15 locus was associated with doxorubicin-related arrhythmia at the genome-wide significance (minor allele frequency in Molecular Epidemiology Resource cases 0.195 versus 0.069 in controls; Tapestry cases 0.192 versus 0.079 in controls) P =9.01×10 −9 ; odds ratio, 3.61. MMP 16 rs1094199 was associated with the broader phenotype of arrhythmia and heart failure (minor allele frequency in Molecular Epidemiology Resource cases 0.079 versus 0.019 in controls; Tapestry cases 0.038 versus 0.015) P =1.98×10 − 8 ; odds ratio=6.15. In human cardiomyocytes, MMP16 and KCTD15 gene expression was significantly downregulated by doxorubicin, and in a zebrafish MMP16 knockout was protective of doxorubicin-induced decline in left ventricular ejection fraction. CONCLUSIONS: Genetic variants at KCTD15 and MMP16 are associated with doxorubicin-related arrhythmia and cardiac events. MMP16 is a potential cardioprotective therapeutic target.

Abstract Frailty influences treatment outcomes in older patients with multiple myeloma (MM), yet existing frailty tools capture limited geriatric domains and are incompletely studied in the context of contemporary therapies. We prospectively evaluated patient-reported comprehensive geriatric assessment (CGA) domains and their associations with treatment toxicity and patient-reported outcomes (PRO) in older adults with MM. Patients aged ≥60 years with newly diagnosed or relapsed/refractory MM initiating a new therapy completed electronic CGA and PRO questionnaires at baseline and 4 months. Outcomes included clinician- and patient-reported treatment toxicity, treatment adaptations, health care use, and health-related quality of life (HRQOL). A cumulative deficit frailty index (FI) was derived from CGA and HRQOL domains. Among 90 patients (mean age, 72 years), geriatric impairments were common, particularly physical function domains. High-grade nonhematologic toxicities were significantly associated with older age, comorbidities, patient-reported functional limitations, and anxiety. Toxicity rates on the PRO version of the Common Terminology Criteria for Adverse Events (CTCAE) were higher than on CTCAE, highlighting a discordance between physician- and patient-reported treatment toxicity. Frailty was present in 22% and associated with worse baseline HRQOL and increased high-grade nonhematologic toxicity but not treatment adaptations or health care use. Average CGA and HRQOL scores remained stable over 4 months, although individual trajectories varied. Patient-reported geriatric impairments are prevalent in older adults with MM and strongly associated with nonhematologic toxicities even in the contemporary treatment era. Physician-reported CTCAE may not fully reflect patients' symptom burden. Heterogeneity in individual geriatric trajectories during treatment emphasize the need for longitudinal assessment. This study was registered at www.clinicaltrials.gov as NCT05918185.

Abstract Measurable residual disease (MRD) is a powerful prognostic tool in multiple myeloma, but predictors of achieving MRD negativity after autologous stem cell transplantation (ASCT) are not well defined. We studied 814 patients who achieved at least a very good partial response after ASCT between January 2016 and January 2023 and underwent MRD assessment by next-generation flow cytometry. Overall, 387 (48%) achieved MRD negativity after ASCT. The presence of t(11;14) (odds ratio [OR], 2.1; 95% confidence interval [CI], 1.4–3.2; P&lt; .01) and pretransplant response less than complete remission (OR, 2.1; 95% CI, 1.2-3.7; P&lt; .01) independently predicted MRD positivity. Patients who were MRD positive had inferior progression-free survival (PFS; hazard ratio [HR], 1.99; 95% CI, 1.53-2.58; P&lt; .001) and overall survival (HR, 1.62; 95% CI, 1.10-2.28; P = .009). Despite lower MRD-negative rate, 5-year PFS was similar among t(11;14) MRD positive (58%) and patients who were MRD negative (63%; P = .12). In patients who were MRD negative, the 5-year PFS was 73% in those without ≥2 HRCA compared with 55% in those with ≥2 HRCA (HR 1.9, 95% CI 1.2–3.3; P = .01). In the MRD-positive cohort, 5-year PFS was 53% in patients without ≥2 HRCA vs 17% in those with ≥2 HRCA (HR, 2.4; 95% CI, 1.6-3.6; P&lt; .01). In addition, higher residual clonal plasma cell burden within MRD-positive disease was associated with inferior PFS (HR, 2.4; 95% CI, 1.2-6.9; P = .002). This study highlights that the presence of t(11;14) independently predicts MRD positivity after ASCT, although it does not significantly affect prognosis in this subgroup.

Abstract Immune checkpoint inhibitors (ICIs) can produce durable tumor control, yet many patients experience primary or acquired resistance, mixed lesion-level responses, or immune-related adverse events. Tissue and blood biomarkers provide biologically informative but spatially incomplete views of a heterogeneous and time-varying tumor–host system. Radiologic imaging complements these assays by repeatedly measuring whole-body tumor burden, organ-specific disease patterns, lesion-level heterogeneity, and treatment-related change. This critical narrative review evaluates imaging-centered artificial intelligence (AI) for precision cancer immunotherapy, including baseline and serial CT, MRI, and PET; radiomics and learned image representations; lesion-level and whole-body heterogeneity; toxicity assessment; and integration with pathology, genomics, circulating tumor DNA, laboratory measurements, and electronic health records. Representative cohorts range from early radiogenomic studies of approximately 100–200 patients to externally validated and longitudinal models involving hundreds or thousands of patients. Rather than treating all reported performance estimates as equally actionable, we appraise the clinical question, cohort design, endpoint, validation strategy, calibration, and intended clinical action. Current evidence supports noninvasive immune phenotyping, risk stratification, longitudinal monitoring, and multimodal biomarker integration. However, most models predict outcomes among patients already receiving ICIs and therefore remain prognostic under treatment rather than predictive of differential treatment benefit. Radiotherapy (RT) combined with ICI serves as a focused imaging use case because lesion identity, dose distribution, fractionation, sequence, and thoracic and normal-tissue exposure must be linked to defined actions. Existing patient-level AI studies estimate efficacy or toxicity after treatment selection; none has prospectively shown that an AI-selected RT strategy improves patient outcomes. Translation will require patient-level leakage control, calibration, geographic and temporal validation, robustness to missing modalities, and prospective evaluation of prespecified model-guided decisions.

Abstract Understanding how immune checkpoint inhibitors (ICIs) reshape human T-cell responses requires models that move beyond static transcriptomic snapshots and discrete cell-state classifications. Here, we present an integrative AI-enabled virtual cell framework that represents human CD8⁺ T-cell responses as dynamic and computable systems during ICI therapy. By integrating single-cell RNA sequencing with paired T-cell receptor sequencing within the C2S-scale foundation model, we construct a virtual representation of individual T cells, in which each cell is encoded by a unique functional identity that captures its transcriptional, signaling, and clonal characteristics. Using this framework, we identify a previously unrecognized dynamic latent effector state of CD8⁺ T cells characterized by intermediate expression of effector genes, distinct signaling activity, and ongoing clonal expansion. Across independent patient cohorts, the virtual cell model consistently indicates that ICI therapy mainly acts by unmasking pre-existing effector potential rather than inducing de novo effector differentiation. Notably, this latent effector population remains transcriptionally restrained despite active signaling and clonal expansion, revealing a hidden reservoir of antitumor immune capacity. More broadly, our study demonstrates how AI-enabled virtual cell modeling can reconstruct latent cellular states and their dynamic transitions from multidimensional single-cell data. By incorporating functional identity into virtual cell model, this framework uncovers biologically meaningful yet non-obvious T-cell effector program during cancer immunotherapy and provides a generalizable approach for studying immune dynamics in human disease.