Tag
This academic paper investigates using LLMs for zero-shot prediction of psychological well-being scores from spontaneous speech, evaluating 12 models and achieving high correlation with clinical metrics.