论文解读
咳嗽声加基本信息做结核初筛：双向互查与漏诊加罚如何分工
针对仅用咳嗽声会丢掉年龄症状等背景风险、简单拼接又学不到两者配合的问题，DeepGB-TB 用表格概率嵌入加一维卷积做两路表示、双向交叉注意力做相互提炼、漏诊加权损失压低假阴性，在 1105 例七国数据上报告 AUROC 0.903 与 F1 0.851，代价是依赖回顾性病例对照数据与特定预处理和调参条件。
语音生物标志物
论文解读
去掉语言可预测成分后跨语言帕金森检测为何能找回灵敏度
针对目标语言只有健康人语音时分类器把目标语言当成健康信号的问题，论文用健康人拟合的岭回归残差去除语言可预测成分，在 5 个骨干、3 个任务、3 个目标语言上把平均 F1 从 0.52 提升到 0.87，但语言信息并未完全消除且评估仍限于印欧语系。
论文解读
不对概率取平均，对名次取平均：用排序聚合绕开跨模态置信度量纲
针对声学与语义分类器后验概率量纲不可比的问题，该工作用训练集留一法分布下的百分位名次代替原始概率做平均，并在高置信度分歧时才允许手工语言特征随机森林覆写，在 ADReSS2020 取得 95.83% 准确率、在 ADReSSo2021 取得 90.14% 准确率，代价是需要保留训练侧分布与多组逻辑回归分支并做前向选择。
论文解读
线性拼接已够好时，注意力还能补什么：BioSync 的宽深融合与合成验证
BioSync 用多头自注意力建模模态间交互并并联线性拼接分支，在两个合成队列上以认知队列 AUC 0.928 和代谢队列准确率 0.764 为最强证据，代价是干净数据优势微弱且临床推断仍待真实队列验证。
论文解读
Speaker-Disentangled Remote Speech Detection of Asthma and COPD Exacerbations
医疗音频 | 7.5/10
论文解读
Can Large Language Models Imitate Human Speech for Clinical Assessment? LLM-Driven Data Augmentation for Cognitive Score Prediction
语音生物标志物 | 6/10
论文解读
A Benchmark for Early-stage Parkinson's Disease Detection from Speech
语音生物标志物 | 7.2/10
论文解读
PROCESS-2: A Benchmark Speech Corpus for Early Cognitive Impairment Detection
语音生物标志物 | 5.4/10
论文解读
Voice Biomarkers for Depression and Anxiety
语音生物标志物 | 1.0/10
论文解读
Mixed-Precision Information Bottlenecks for On-Device Trait-State Disentanglement in Bipolar Agitation Detection
语音生物标志物 | 8.0/10
论文解读
Neck-Learn: Attention-Based Multiple Instance Learning and Ensemble Framework for Ecological Momentary Assessment
语音生物标志物 | 7.0/10
论文解读
Predicting Upcoming Stuttering Events from Three-Second Audio: Stratified Evaluation Reveals Severity-Selective Precursors, and the Model Deploys Fully On-Device
语音生物标志物 | 7.0/10
论文解读
A Toolkit for Detecting Spurious Correlations in Speech Datasets
模型评估 | 7.0/10
论文解读
Recurrence-Based Nonlinear Vocal Dynamics as Digital Biomarkers for Depression Detection from Conversational Speech
语音生物标志物 | 6.5/10
论文解读
A Bimodal Approach for Detecting Fatigue Using Speech and Personal Assessments in College Students
A Bimodal Approach for Detecting Fatigue Using Speech and Personal Assessments in College Students
论文解读
A Consistent Learning Depression Detection Framework Integrating Multi-View Attention
语音生物标志物 | 6.5/10
论文解读
An Anomaly-Aware and Audio-Enhanced Dual-Pathway Framework for Alzheimer’s Disease Progression Classification
语音生物标志物 | 7.0/10
论文解读
An Unsupervised Alignment Feature Fusion System for Spoken Language-Based Dementia Detection
语音生物标志物 | 7.0/10
论文解读
CMSA-Mamba: Hierarchical State Space Modeling for Audio-Based Depression Detection
语音生物标志物 | 7.0/10
论文解读
Conditional Diffusion Models for Mental Health-Preserving Voice Conversion
语音转换 | 8.0/10