Academy of Consultation-Liaison Psychiatry

Journal Urges Validating and Monitoring of AI Systems

‘Research and quality improvement projects are critical to refine AI in C-L Psychiatry and address its limitations’

Potentially dehumanizing aspects of AI-based care must be resisted, says the Academy’s journal editor in an editorial for its latest edition.

“Artificial intelligence should complement, not replace, the human elements of C-L Psychiatry, ensuring that compassion, clinical judgment, and ethical considerations remain central to our practice,” says Hochang Lee, MD, FACLP.

“Despite its vast potential, integrating AI into C-L Psychiatry presents several challenges,” he says. “AI systems often require access to sensitive patient data, raising significant privacy and security concerns. Additionally, algorithmic bias poses a risk, as AI models may perpetuate or even magnify existing disparities in mental health care. Furthermore, the potentially dehumanizing aspects of AI-based care must be resisted.”

To achieve this, C-L psychiatrists must take an active role in validating and monitoring AI systems used in health care settings. “Research and quality improvement projects on AI are critical to refine AI applications in C-L Psychiatry and address their limitations. By investing in rigorous studies and fostering interdisciplinary collaboration, the field of C-L Psychiatry can harness AI’s full potential to enhance patient outcomes while maintaining ethical integrity.

“AI will undoubtedly shape the future of C-L Psychiatry, but it is our responsibility to ensure that its integration aligns with the ACLP vision of achieving optimal health through integrated medical and psychiatric care.”

The full editorial in the Journal of the Academy of Consultation-Liaison Psychiatry (JACLP) is here.

Key C-L Psychiatry Areas Where AI is Currently being Developed and Implemented
JACLP highlights that more than 5,000 mental health mobile apps are available; the Food and Drug Administration has approved seven digital therapeutics for psychiatric disorders, and dozens more are in the pipeline. AI-driven algorithms can analyze vast amounts of electronic health record data to identify hospitalized patients who would benefit from psychiatric consultation. Machine learning models trained on historical data can also predict which medical or surgical patients are at risk for conditions such as delirium, depression, or anxiety. Virtual health assistants can conduct preliminary psychiatric interviews, collect detailed histories, and identify key symptoms. “These tools could expedite psychiatric evaluations, allowing C-L psychiatrists to be more efficient in busy hospital settings.” AI can also analyze verbal and nonverbal cues during patient interactions. For example, voice analysis algorithms can detect changes in tone, pitch, and speech patterns indicative of depression, mania, or psychosis. Similarly, computer vision technologies can assess facial expressions and body language, providing additional insights into a patient’s emotional state. AI can help C-L psychiatrists synthesize information from multiple sources—electronic health records, patient interviews, and diagnostic tests—to generate evidence-based treatment recommendations. “This capability is particularly valuable in C-L Psychiatry, where patients often present with complex medical and psychiatric comorbidities.” Machine learning can suggest personalized treatment plans by analyzing large datasets of patient outcomes, thereby predicting which interventions are most likely to be effective based on a patient’s unique characteristics, including medical history, genetic profile, and social determinants of health. Additionally, predictive analytics can estimate treatment adherence, potential side-effects, and relapse risks, allowing psychiatrists to tailor interventions more precisely and improve patient outcomes. Chatbots and virtual therapists provide immediate access to mental health support and deliver evidence-based therapies such as cognitive behavioral therapy and dialectical behavior therapy. AI can also enhance biofeedback and neuro-feedback therapies. Algorithms analyze real-time data from sensors monitoring brain activity, heart rate, and other physiological parameters. These data are then used to help patients regulate their emotional states and improve mental well-being.

Key C-L Psychiatry Areas Where AI is Currently being Developed and Implemented

JACLP highlights that more than 5,000 mental health mobile apps are available; the Food and Drug Administration has approved seven digital therapeutics for psychiatric disorders, and dozens more are in the pipeline.

AI-driven algorithms can analyze vast amounts of electronic health record data to identify hospitalized patients who would benefit from psychiatric consultation. Machine learning models trained on historical data can also predict which medical or surgical patients are at risk for conditions such as delirium, depression, or anxiety.
Virtual health assistants can conduct preliminary psychiatric interviews, collect detailed histories, and identify key symptoms. “These tools could expedite psychiatric evaluations, allowing C-L psychiatrists to be more efficient in busy hospital settings.” AI can also analyze verbal and nonverbal cues during patient interactions. For example, voice analysis algorithms can detect changes in tone, pitch, and speech patterns indicative of depression, mania, or psychosis. Similarly, computer vision technologies can assess facial expressions and body language, providing additional insights into a patient’s emotional state.
AI can help C-L psychiatrists synthesize information from multiple sources—electronic health records, patient interviews, and diagnostic tests—to generate evidence-based treatment recommendations. “This capability is particularly valuable in C-L Psychiatry, where patients often present with complex medical and psychiatric comorbidities.”
Machine learning can suggest personalized treatment plans by analyzing large datasets of patient outcomes, thereby predicting which interventions are most likely to be effective based on a patient’s unique characteristics, including medical history, genetic profile, and social determinants of health. Additionally, predictive analytics can estimate treatment adherence, potential side-effects, and relapse risks, allowing psychiatrists to tailor interventions more precisely and improve patient outcomes.
Chatbots and virtual therapists provide immediate access to mental health support and deliver evidence-based therapies such as cognitive behavioral therapy and dialectical behavior therapy. AI can also enhance biofeedback and neuro-feedback therapies. Algorithms analyze real-time data from sensors monitoring brain activity, heart rate, and other physiological parameters. These data are then used to help patients regulate their emotional states and improve mental well-being.

CLP 2025 BANNER

This year’s ACLP Annual Meeting in San Antonio, November 19-22, Innovation in C-L Psychiatry: Exploring the Promise and Pitfalls of New Approaches, will include plenary presentations on the potential clinical applications of AI and large language models.

« All 2025 ACLP News « April 2025 ACLP News