发表机构
Dalian University of Technology; The Hong Kong Polytechnic University; Southwest University（大连理工大学; 香港理工大学; 西南大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对不完整的移动与可穿戴心理健康特征视图，提出 MC-TRCM 模型，在 DepreST-CAT 和 PSYCHE-D 数据集的 PHQ-9、GAD-7 等端点预测中，其回归与分类性能优于多数基准模型， ablation 验证了核心组件的有效性。
AI 中文摘要
公开的移动与可穿戴设备心理健康数据集通常提供汇总特征表，而非同步原始传感器流。在这些发布版本中，每个锚点对应一次调查或标签时间，可能结合手机或可穿戴设备的汇总数据、既往症状评分、人口统计学信息、临床变量及源可用性指标。我们提出了模态条件时间递归上下文模型（Modality-Conditioned Temporal Recursive Context Model, MC-TRCM），该模型将每个特征源保留为单独的 token，并将缺失性纳入输入上下文的一部分。观测到的源用数值和缺失性摘要进行编码，缺失的源使用学习到的缺失 token，数据集和任务嵌入对融合过程进行条件约束，递归预测头在验证选定的步骤上细化每个输出。我们使用参与者级划分和仅验证的模型选择，在 DepreST-CAT 和抑郁严重程度变化预测（Prediction of Severity Change-Depression, PSYCHE-D）的六个预定义端点上评估了 MC-TRCM。MC-TRCM 在 DepreST-CAT 的患者健康问卷-9（Patient Health Questionnaire-9, PHQ-9）和广泛性焦虑障碍-7（Generalized Anxiety Disorder-7, GAD-7）严重程度预测中实现了最低的平均绝对误差，较最佳表格基准分别提升了 0.181 和 0.217 个尺度点。分类端点表现出任务依赖的行为：MC-TRCM 达到了最佳四舍五入 GAD-7 类别的平衡准确率，在 PSYCHE-D 多类别预测中以 0.002 个平衡准确率点的数值最高，在 PHQ-9 类别和 PSYCHE-D 二分类预测中也接近最强基准。 ablation 实验支持特征级线性调制、缺失 token、缺失投影和递归细化，而校准和特征源控制则表征了端点行为。我们的代码可在该 https URL 获取。
英文摘要
Public mobile and wearable mental-health datasets often provide summarized feature tables rather than synchronized raw sensor streams. In these releases, each anchor corresponds to a survey or label time and may combine phone or wearable summaries, prior symptom scores, demographics, clinical variables, and source-availability indicators. We propose the Modality-Conditioned Temporal Recursive Context Model (MC-TRCM), which preserves each feature source as a separate token and incorporates missingness as part of the input context. Observed sources are encoded with values and missingness summaries, absent sources use learned absence tokens, dataset and task embeddings condition fusion, and a recursive prediction head refines each output over validation-selected steps. We evaluated MC-TRCM on six predefined endpoints from DepreST-CAT and Prediction of Severity Change-Depression (PSYCHE-D) using participant-level splits and validation-only model selection. MC-TRCM achieved the lowest mean absolute error on DepreST-CAT Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder-7 (GAD-7) severity, improving over the best tabular reference by 0.181 and 0.217 scale points. Classification endpoints showed task-dependent behavior: MC-TRCM matched the best rounded GAD-7 category balanced accuracy, was numerically highest by 0.002 balanced-accuracy points on PSYCHE-D multiclass prediction, and remained close to the strongest references on PHQ-9 category and PSYCHE-D binary prediction. Ablations support Feature-wise Linear Modulation, absence tokens, missingness projections, and recursive refinement, while calibration and feature-source controls characterize endpoint behavior. Our code is available at https://github.com/Botwwt/MC-TRCM.
CommentsAccepted to ICONIP 2026. Code is available at https://github.com/Botwwt/MC-TRCM
