People at risk of suicide tend to be isolated and cannot share their thoughts. For this reason, suicidal ideation monitoring becomes a hard task. Therefore, people at risk of suicide need to be monitored in a manner capable of identifying if and when they have a suicidal ideation, enabling professionals to perform timely interventions. This study aimed to develop the Boamente tool, a solution that collects textual data from users’ smartphones and identifies the existence of suicidal ideation. The solution has a virtual keyboard mobile application that passively collects user texts and sends them to a web platform to be processed. The platform classifies texts using natural language processing and a deep learning model to recognize suicidal ideation, and the results are presented to mental health professionals in dashboards. Text classification for sentiment analysis was implemented with different machine/deep learning algorithms. A validation study was conducted to identify the model with the best performance results. The BERTimbau Large model performed better, reaching a recall of 0.953 (accuracy: 0.955; precision: 0.961; F-score: 0.954; AUC: 0.954). The proposed tool demonstrated an ability to identify suicidal ideation from user texts, which enabled it to be experimented with in studies with professionals and their patients.

Artificial Intelligence (AI)-based systems have been proposed to aid Mental Health Professionals (MHPs) in various tasks, including the prevention of suicide by identifying Suicidal Ideation (SI). However, these systems may lack transparency and thereby create mistrust among MHPs. Explainable Artificial Intelligence (XAI) methods can elucidate how features influence system predictions, aiding MHPs in understanding them. This exploratory study aims to investigate how MHPs’ trust is influenced by AI explanations (educational intervention and XAI methods) and other factors (professional background, knowledge of AI and computing, and reported system misclassification).We conducted an experiment using Boamente, an AI-powered clinical decision support system designed to assist MHPs in suicide prevention. Boamente identifies SI in Brazilian Portuguese texts typed on smartphones by leveraging a Large Language Model (LLM) for analysis. The results demonstrate that professional background, knowledge of AI and computing, and educational intervention had no statistically significant effect on trust. In contrast, trust was affected by factors such as LLM prediction explanations, the quality of explanations, and reported misclassification. Therefore, providing prediction explanations to understand the inner workings of AI models led MHPs to be more critical in relation to predictions, while there was an overtrust on MHPs when no explanations were provided. Furthermore, disagreement with LLM classifications and perceptions of system vulnerabilities also affected trust.

Artificial intelligence can detect suicidal ideation manifestations in texts. Studies demonstrate that BERT-based models achieve better performance in text classification problems. Large language models (LLMs) answer free-text queries without being specifically trained. This work aims to compare the performance of three variations of BERT models and LLMs (Google Bard, Microsoft Bing/GPT-4, and OpenAI ChatGPT-3.5) for identifying suicidal ideation from nonclinical texts written in Brazilian Portuguese. A dataset labeled by psychologists consisted of 2,691 sentences without suicidal ideation and 1,097 with suicidal ideation, of which 100 sentences were selected for testing. We applied data preprocessing techniques, hyperparameter optimization, and hold-out cross-validation for training and testing BERT models. When evaluating LLMs, we used zero-shot prompting engineering. Each test sentence was labeled if it contained suicidal ideation, according to the chatbot’s response. Bing/GPT-4 achieved the best performance, with 98% across all metrics. Fine-tuned BERT models outperformed the other LLMs: BERTimbau-Large performed the best with a 96% accuracy, followed by BERTimbau-Base with 94%, and BERT-Multilingual with 87%. Bard performed the worst with 62% accuracy, whereas ChatGPT-3.5 achieved 81%. The high recall capacity of the models suggests a low misclassification rate of at-risk patients, which is crucial to prevent missed interventions by professionals. However, despite their potential in supporting suicidal ideation detection, these models have not been validated in a patient monitoring clinical setting. Therefore, caution is advised when using the evaluated models as tools to assist healthcare professionals in detecting suicidal ideation.

This study aimed to compare the performance of various Large Language Models (LLMs) in answering multiple-choice questions from the last six editions (2017 to 2024) of Revalida exam. The evaluation focused on models capable of processing content in Brazilian Portuguese (PT-BR), including open-source models, namely LLaMA 3.1 (8B parameters), Qwen 2.5 (7B parameters), and their reasoning-oriented distilled variants based on the DeepSeek-R1 architecture, as well as open-access commercial models such as GPT-3.5, GPT-4o, and Gemini. After evaluating the models’ accuracy against the official answer keys, GPT-4o emerged as the top-performing model, achieving an average accuracy of 63.85%. Next, GPT-4o was prompted to justify its answers to the 2024 exam, and its explanations were independently reviewed by three licensed physicians. The evaluators reported full agreement with the clinical reasoning presented, indicating the model’s ability to produce coherent and medically relevant justifications. Lastly, justifications generated by GPT-4o for correctly answered questions from previous exams (2017–2023) were compiled into a knowledge base, which was then used to enhance GPT-4o through retrieval-augmented generation and to fine-tune LLaMA 3.1, leading to measurable performance improvements on the 2024 exam. Despite promising performance, these models still demonstrate variability in responses, hallucinations, and limited reliability in high-stakes contexts. As such, their outputs should always be reviewed by qualified professionals, and human expertise remains essential in clinical decision-making and medical education scenarios, considering the PT-BR language. However, the observed gains from integrating prior exam content indicate that domain-specific adaptation strategies may help mitigate some of these limitations and enhance model alignment.