Megadose AI progress, ranked and analyzed.

Improving Mental Health Screening and Early Risk Detection in Spanish

· ArXiv · AI/CL/LG ·
Spanish mental-health screening models are being tuned to flag risk earlier in long social-media histories.

The paper introduces three Spanish foundational models adapted with mental-health domain pre-training. It also proposes Incremental Context Expansion, a relabeling method that marks the point where accumulated posts provide enough evidence for detection. On three Spanish benchmarks, the authors report better state-of-the-art performance with lower detection latency. The models are publicly available. ArXiv · AI/CL/LG's note

score 4

Categories: Model Releases, Research