280

Papers

9/9

Roadblocks Active

0

Connections

Career anxiety and depression among university students present significant challenges to mental health and academic achievement; Key behavioral markers of depression identified: avoidance of direct gaze, lower facial expressiveness, and social withdrawal

██████████ 0.8 depression-biomarkers Preprint

ReadSaveConnections

Flexible models (deep networks, gradient-boosted trees, kernel methods) are repeatedly matched or beaten by linear/logistic regression on biomedical tabular data with the same features; Measurement noise is a binding limit in biomedical prediction that cannot be overcome by more data, better architecture, or tuning

██████████ 0.7 depression-biomarkers Preprint

ReadSaveConnections

Meditation increases functional signal-to-noise ratio (f-SNR) through selective signal enhancement and noise decluttering; Deepening meditation practice reduces self-referential filtering and shifts neural activity toward a critical regime

██████████ 0.7 treatment-resistant-depression Preprint

ReadSaveConnections

Baseline-normalized hyperarousal trajectories differed significantly across conditions, with digital intervention group showing structured stabilization versus late-study escalation in physical-activity-only group; Digital intervention group demonstrated higher overall maintenance of symptom gains compared to physical-activity-only group during post-event follow-up

██████████ 0.8 digital-therapeutics Preprint

ReadSaveConnections

Safety guardrails in eight proprietary LLMs are reliable only for suicide and self-harm prevention across DSM-5 conditions; Eating disorders, substance use disorder, and major depressive disorder exhibit LLM safeguard failure rates up to 100%

██████████ 0.9 digital-therapeutics Preprint

ReadSaveConnections

Multi-agent audit framework significantly outperforms single-agent LLM baselines in clinical mental health screening; Mean Absolute Error for PHQ-8 depression severity prediction reduced from 5.35 to 5.02 using the proposed framework

██████████ 0.8 digital-therapeutics Preprint

ReadSaveConnections

Fine-tuned Qwen3.5-27B model predicts PHQ-9 total scores directly from AI mental health dialogue transcripts without additional clinical data; Model achieves MAE = 2.6, RMSE = 4.0, Pearson r = 0.80 on held-out test set of 842 users

██████████ 0.8 depression-biomarkers Preprint

ReadSaveConnections

Approximately 30% of college students meet insomnia criteria; Hyperarousal model of insomnia involves mutual reinforcement across cognitive, emotional, and physiological domains

██████████ 0.9 sleep-circadian-psychiatry Preprint

ReadSaveConnections

PsyBridge achieves 0.84 overall accuracy in multi-dimensional mental health assessment; Integration of cognitive and personality components reduces inconsistencies in moderate-risk prediction

██████████ 0.8 digital-therapeutics Preprint

ReadSaveConnections

Zero-shot cross-subject EEG decoding is achievable on large-scale datasets without subject-specific calibration; Transformer-based foundation models outperform convolutional neural networks and LSTM hybrids for generalizable EEG decoding

██████████ 0.8 depression-biomarkers Preprint

ReadSaveConnections