Researchers reviewed 677 studies on conversational AI in mental health and found that research has increased sharply since 2020. Most studies focused on detection and intervention, while prevention and long-term maintenance received far less attention. Maintenance, which includes treatment continuity and follow-up care, represented only 3% of the studies reviewed.
The review also found that large language models have become a major AI technology in this area, especially in intervention and maintenance. The authors proposed a human-centered framework for future mental health AI tools, focused on emotional sensitivity, user-centered interaction, human-AI collaboration, and ethics and accountability.
Why It Matters
This review highlights a key gap in mental health AI: far less research has focused on what happens between therapy sessions and over the longer course of care.
That gap is directly relevant to Therapy Ally. As AI becomes more common in mental health, the need for structured, human-centered tools that complement therapy, support continuity, and keep clinicians appropriately involved is likely to become more important.