A review of the explainability and safety of conversational agents for mental health to identify avenues for improvement

The Potential of Gemini and GPTs for Structured Report Generation based on Free-Text 18F-FDG PET/CT Breast Cancer Reports

RATIONALE AND OBJECTIVE

To compare the performance of large language model (LLM) based Gemini and Generative Pre-trained Transformers (GPTs) in data mining and generating structured reports based on free-text PET/CT reports for breast cancer after user-defined tasks.

MATERIALS AND METHODS

Breast cancer patients (mean age, 50 years ± 11 [SD]; all female) who underwent consecutive 18F-FDG PET/CT for follow-up between July 2005 and October 2023 were retrospectively included in the study. A total of twenty reports from 10 patients were used to train user-defined text prompts for Gemini and GPTs, by which structured PET/CT reports were generated. The natural language processing (NLP) generated structured reports and the structured reports annotated by nuclear medicine physicians were compared in terms of data extraction accuracy and capacity of progress decision-making. Statistical methods, including chi-square test, McNemar test and paired samples t-test, were employed in the study.

RESULTS

The structured PET/CT reports for 131 patients were generated by using the two NLP techniques, including Gemini and GPTs. In general, GPTs exhibited superiority over Gemini in data mining in terms of primary lesion size (89.6% vs. 53.8%, p < 0.001) and metastatic lesions (96.3% vs 89.6%, p < 0.001). Moreover, GPTs outperformed Gemini in making decision for progress (p < 0.001) and semantic similarity (F1 score 0.930 vs 0.907, p < 0.001) for reports.

CONCLUSION

GPTs outperformed Gemini in generating structured reports based on free-text PET/CT reports, which is potentially applied in clinical practice.

DATA AVAILABILITY

The data used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Mental Health Applications of Generative AI and Large Language Modeling in the United States

(1) Background: Artificial intelligence (AI) has flourished in recent years. More specifically, generative AI has had broad applications in many disciplines. While mental illness is on the rise, AI has proven valuable in aiding the diagnosis and treatment of mental disorders. However, there is little to no research about precisely how much interest there is in AI technology. (2) Methods: We performed a Google Trends search for "AI and mental health" and compared relative search volume (RSV) indices of "AI", "AI and Depression", and "AI and anxiety". This time series study employed Box-Jenkins time series modeling to forecast long-term interest through the end of 2024. (3) Results: Within the United States, AI interest steadily increased throughout 2023, with some anomalies due to media reporting. Through predictive models, we found that this trend is predicted to increase 114% through the end of the year 2024, with public interest in AI applications being on the rise. (4) Conclusions: According to our study, we found that the awareness of AI has drastically increased throughout 2023, especially in mental health. This demonstrates increasing public awareness of mental health and AI, making advocacy and education about AI technology of paramount importance.

Applications of large language models in psychiatry: a systematic review

Background

With their unmatched ability to interpret and engage with human language and context, large language models (LLMs) hint at the potential to bridge AI and human cognitive processes. This review explores the current application of LLMs, such as ChatGPT, in the field of psychiatry.

Methods

We followed PRISMA guidelines and searched through PubMed, Embase, Web of Science, and Scopus, up until March 2024.

Results

From 771 retrieved articles, we included 16 that directly examine LLMs' use in psychiatry. LLMs, particularly ChatGPT and GPT-4, showed diverse applications in clinical reasoning, social media, and education within psychiatry. They can assist in diagnosing mental health issues, managing depression, evaluating suicide risk, and supporting education in the field. However, our review also points out their limitations, such as difficulties with complex cases and potential underestimation of suicide risks.

Conclusion

Early research in psychiatry reveals LLMs' versatile applications, from diagnostic support to educational roles. Given the rapid pace of advancement, future investigations are poised to explore the extent to which these models might redefine traditional roles in mental health care.

Recommendation endpoints and safety of an online self-triage for depression symptoms

INTRODUCTION

Online symptom checkers are a way to address patient concerns and potentially offload a burdened healthcare system. However, safety outcomes of self-triage are unknown, so we reviewed triage recommendations and outcomes of our institution's depression symptom checker.

METHODS

We examined endpoint recommendations and follow-up encounters seven days afterward during 2 December 2021 to 13 December 2022. Patients with an emergency department visit or hospitalization within seven days of self-triaging had a manual review of the electronic health record to determine if the visit was related to depression, suicidal ideation, or suicide attempt. Charts were reviewed for deaths within seven days of self-triage.

RESULTS

There were 287 unique encounters from 263 unique patients. In 86.1% (247/287), the endpoint was an instruction to call nurse triage; in 3.1% of encounters (9/287), instruction was to seek emergency care. Only 20.2% (58/287) followed the recommendations given. Of the 229 patients that did not follow the endpoint recommendations, 121 (52.8%) had some type of follow-up within seven days. Nearly 11% (31/287) were triaged to endpoints not requiring urgent contact and 9.1% (26/287) to an endpoint that would not need any healthcare team input. No patients died in the study period.

CONCLUSIONS

Most patients did not follow the recommendations for follow-up care although ultimately most patients did receive care within seven days. Self-triage appears to appropriately sort patients with depressed mood to emergency care. On-line self-triaging tools for depression have the potential to safely offload some work from clinic personnel.

Safety of Large Language Models in Addressing Depression

Background Generative artificial intelligence (AI) models, exemplified by systems such as ChatGPT, Bard, and Anthropic, are currently under intense investigation for their potential to address existing gaps in mental health support. One implementation of these large language models involves the development of mental health-focused conversational agents, which utilize pre-structured prompts to facilitate user interaction without requiring specialized knowledge in prompt engineering. However, uncertainties persist regarding the safety and efficacy of these agents in recognizing severe depression and suicidal tendencies. Given the well-established correlation between the severity of depression and the risk of suicide, improperly calibrated conversational agents may inadequately identify and respond to crises. Consequently, it is crucial to investigate whether publicly accessible repositories of mental health-focused conversational agents can consistently and safely address crisis scenarios before considering their adoption in clinical settings. This study assesses the safety of publicly available ChatGPT-3.5 conversational agents by evaluating their responses to a patient simulation indicating worsening depression and suicidality. Methodology This study evaluated ChatGPT-3.5 conversational agents on a publicly available repository specifically designed for mental health counseling. Each conversational agent was evaluated twice by a highly structured patient simulation. First, the simulation indicated escalating suicide risk based on the Patient Health Questionnaire (PHQ-9). For the second patient simulation, the escalating risk was presented in a more generalized manner not associated with an existing risk scale to assess the more generalized ability of the conversational agent to recognize suicidality. Each simulation recorded the exact point at which the conversational agent recommended human support. Then, the simulation continued until the conversational agent stopped entirely and shut down completely, insisting on human intervention. Results All 25 agents available on the public repository FlowGPT.com were evaluated. The point at which the conversational agents referred to a human occurred around the mid-point of the simulation, and definitive shutdown predominantly only happened at the highest risk levels. For the PHQ-9 simulation, the average initial referral and shutdown aligned with PHQ-9 scores of 12 (moderate depression) and 25 (severe depression). Few agents included crisis resources - only two referenced suicide hotlines. Despite the conversational agents insisting on human intervention, 22 out of 25 agents would eventually resume the dialogue if the simulation reverted to a lower risk level. Conclusions Current generative AI-based conversational agents are slow to escalate mental health risk scenarios, postponing referral to a human to potentially dangerous levels. More rigorous testing and oversight of conversational agents are needed before deployment in mental healthcare settings. Additionally, further investigation should explore if sustained engagement worsens outcomes and whether enhanced accessibility outweighs the risks of improper escalation. Advancing AI safety in mental health remains imperative as these technologies continue rapidly advancing.