Background: The rapid development of artificial intelligence technology has enabled chatbots to increasingly promote health-related behaviors, addressing the high demand for human resources in traditional interventions. Several systematic reviews have been conducted in this area. However, the existing reviews have not focused on the rigorously designed randomized trials of the state-of-the-art chatbots (interacting with users through unconstrained natural language), thus calling for an updated review. Objective: We aimed to explore the effects of natural language processing (NLP) chatbot–based interventions on improving diet, physical activity, and tobacco smoking behaviors in the general population and to evaluate the chatbot use behaviors during the implementation process. Methods: We comprehensively searched 12 databases or registers for eligible studies published from January 1, 2010, until July 16, 2024, and obtained a total of 6301 studies. We included randomized controlled trials (RCTs) that used NLP-chatbots to promote diet, physical activity, or tobacco smoking behaviors among adults or children. Due to considerable heterogeneity across the included studies, we adopted the synthesis without meta-analysis guidelines and summarized the effectiveness of NLP chatbot–based interventions. We used the new evidence-mapping method (bubble plot) to visualize the results. We also described the results related to the changes in diet, physical activity, or tobacco smoking behaviors (eg, change of BMI and stage of change). To evaluate the implementation process of the intervention, we summarized users’ interaction with NLP-chatbots and their feelings (eg, satisfaction) about NLP-chatbot use. Additionally, we assessed the risk of bias of studies using the RoB 2.0 (Risk of Bias; The Cochrane Collaboration) tools. Results: We finally included 7 RCTs. Concerning dietary and physical activity behaviors, the effectiveness of NLP chatbot–based interventions was inconsistent among adults, while no evidence of effect was observed among children. Concerning tobacco smoking behaviors, the included studies showed consistent evidence of improving this behavior among adults. Regarding the risk of bias of the changes in diet, physical activity, and tobacco smoking behaviors, 2 of 3, 2 of 4, and 1 of 2 studies had a high risk of bias, respectively, while the remaining had a low risk of bias. Concerning the interactions with NLP-chatbots, studies showed an overall high percentage of general interaction between users and NLP-chatbots, but not a satisfactorily high percentage of interactions specific to health behaviors. Concerning feelings about NLP-chatbot use, users showed a positive impression of NLP-chatbot use, feeling it was useful, credible, and financially feasible. Conclusions: NLP chatbot–based interventions were beneficial for adults’ tobacco smoking behaviors, but no such evidence was found on diet or physical activity behaviors among adults or children. More RCTs with larger samples and lower risk of bias are urgently needed to enhance our findings in the future.

Background: Diagnosing and managing follicular thyroid neoplasms (FTNs) remains a significant challenge, as the malignancy risk cannot be determined until after diagnostic surgery. Objective: We aimed to use interpretable machine learning to predict the malignancy risk of FTNs preoperatively in a real-world setting. Methods: We conducted a retrospective cohort study at the Peking University Third Hospital in Beijing, China. Patients with postoperative pathological diagnoses of follicular thyroid adenoma (FTA) or follicular thyroid carcinoma (FTC) were included, excluding those without preoperative thyroid ultrasonography. We used 22 predictors involving demographic characteristics, thyroid sonography, and hormones to train 5 machine learning models: logistic regression, least absolute shrinkage and selection operator regression, random forest, extreme gradient boosting, and support vector machine. The optimal model was selected based on discrimination, calibration, interpretability, and parsimony. To address the highly imbalanced data (FTA:FTC ratio>5:1), model discrimination was assessed using both the area under the receiver operating characteristic curve and the area under the precision-recall curve (AUPRC). To interpret the model, we used Shapley Additive Explanations values and partial dependence and individual conditional expectation plots. Additionally, a systematic review was performed to synthesize existing evidence and validate the discrimination ability of the previously developed Thyroid Imaging Reporting and Data System for Follicular Neoplasm scoring criteria to differentiate between benign and malignant FTNs using our data. Results: The cohort included 1539 patients (mean age 47.98, SD 14.15 years; female: n=1126, 73.16%) with 1672 FTN tumors (FTA: n=1414; FTC: n=258; FTA:FTC ratio=5.5). The random forest model emerged as optimal, identifying mean thyroid-stimulating hormone (TSH) score, mean tumor diameter, mean TSH, TSH instability, and TSH measurement levels as the top 5 predictors in discriminating FTA from FTC, with the area under the receiver operating characteristic curve of 0.79 (95% CI 0.77‐0.81) and AUPRC of 0.40 (95% CI 0.37-0.44). Malignancy risk increased nonlinearly with larger tumor diameters and higher TSH instability but decreased nonlinearly with higher mean TSH scores or mean TSH levels. FTCs with small sizes (mean diameter 2.88, SD 1.38 cm) were more likely to be misclassified as FTAs compared to larger ones (mean diameter 3.71, SD 1.36 cm). The systematic review of the 7 included studies revealed that (1) the FTA:FTC ratio varied from 0.6 to 4.0, lower than the natural distribution of 5.0; (2) no studies assessed prediction performance using AUPRC in unbalanced datasets; and (3) external validations of Thyroid Imaging Reporting and Data System for Follicular Neoplasm scoring criteria underperformed relative to the original study. Conclusions: Tumor size and TSH measurements were important in screening FTN malignancy risk preoperatively, but accurately predicting the risk of small-sized FTNs remains challenging. Future research should address the limitations posed by the extreme imbalance in FTA and FTC distributions in real-world data.

Background: Motivational interviewing (MI) is an effective counseling approach for promoting health behavior change, but its scalability is constrained by the need for highly trained human counselors. Large language models (LLMs) may provide a scalable way to support MI counseling, but evidence remains limited, especially for Chinese MI resources and evaluations based on standardized MI fidelity frameworks. Objective: This study aimed to develop Chinese large language models for motivational interviewing (MI-LLMs) and evaluate whether MI-focused fine-tuning could improve their ability to generate counseling responses consistent with MI principles. Methods: We first curated 5 publicly available Chinese psychological counseling datasets and assessed sampled conversations in terms of comprehensiveness, professionalism, authenticity, and safety. The 2 highest-scoring datasets, CPsyCounD and PsyDTCorpus, were selected for MI-style data construction. Using GPT-4 with a structured MI-informed prompt, we transformed 2040 multiturn counseling conversations into MI-style dialogs. Among these, 2000 dialogs were used for training and 40 for testing. Three Chinese-capable open-source LLMs (Baichuan2-7B-Chat, ChatGLM-4-9B-Chat, and Llama-3-8B-Chinese-Chat-v2) were fine-tuned with low-rank adaptation on the training dataset and were referred to as MI-LLMs. Automatic evaluation was conducted on the testing dataset using Bilingual Evaluation Understudy–4 (BLEU-4) and Recall-Oriented Understudy for Gisting Evaluation (ROUGE) metrics. Manual evaluation was conducted using the Motivational Interviewing Treatment Integrity Coding Manual 4.2.1. Thirty simulated counseling dialogs generated by the MI-LLMs were compared with 30 real MI dialogs sampled from AnnoMI and translated into Chinese. Two trained graduate student raters coded global scores and behavior counts, from which summary scores were subsequently calculated. Results: In automatic evaluation, fine-tuning substantially improved BLEU-4 and ROUGE scores across all 3 models compared with the base models. In manual evaluation, the MI-LLMs achieved technical and relational global scores, as well as total MI-adherent ratios that approached those of real MI dialogs. The MI-LLM based on ChatGLM-4-9B-Chat showed the strongest overall global performance. However, MI-LLMs produced fewer complex reflections and had lower reflection-to-question ratios than real MI dialogs. Conclusions: This study provides preliminary evidence that MI focused fine-tuning can help Chinese LLMs acquire core counseling behaviors consistent with MI principles. It also offers a scalable approach for constructing MI style dialog resources in Chinese. Nevertheless, current MI-LLMs should be regarded as early-stage tools for supporting, rather than replacing human counselors. Future work should expand real MI training data and strengthen the complex reflective skills of MI-LLMs. Further studies are needed to evaluate their effectiveness, acceptability, and safety in health behavior change settings in the real world.

Background: Surgeons often face challenges in distinguishing between benign and malignant follicular thyroid neoplasms (FTNs), particularly small tumors, until diagnostic surgery is performed. Objective: This study aimed to identify the size-specific predictors for the malignancy risk of FTNs preoperatively. Methods: A retrospective cohort study was conducted at Peking University Third Hospital in Beijing, China, from 2012 to 2023. Patients with a postoperative pathological diagnosis of follicular thyroid adenoma (FTA) or follicular thyroid carcinoma (FTC) were included. FTNs were classified into small- and large-sized categories based on the cutoff value of the tumor diameter derived from spline regression, which indicated the turning point of malignancy risk. We identified the 5 most important predictors from 22 variables including demography, sonography, and hormones, using machine learning methods. We also calculated the odds ratios (OR) with 95% CI for these predictors in both small- and large-sized FTNs. Results: Altogether, we included 1494 FTNs, comprising 1266 FTAs and 228 FTCs. FTNs with a maximum diameter less than 3.0 cm were grouped as small-sized tumors (n=715), while those with larger diameters were categorized as large-sized tumors (n=779). In the small-sized group, tumors with macrocalcification (OR 2.90, 95% CI 1.50-5.60), those with peripheral calcification (OR 4.50, 95% CI 1.50-13.00), and those in younger patients (OR 1.33, 95% CI 1.05-1.69) showed a higher malignancy risk. In the large-sized group, tumors presenting with a nodule-in-nodule appearance (OR 3.30, 95% CI 1.30-7.90) exhibited a higher malignancy risk. In both groups, lower thyroid-stimulating hormone levels (OR 1.49, 95% CI 1.20-1.85 for small-sized FTNs; OR 1.61, 95% CI 1.37-1.96 for large-sized FTNs) and a larger mean diameter (OR 1.40, 95% CI 1.10-1.70 for small-sized FTNs; OR 1.50 95% CI 1.20-1.70 for large-sized FTNs) were associated with the malignancy risk of FTNs. Conclusion: This study identified size-specific predictors for malignancy risk in FTNs, highlighting the importance of stratified prediction based on tumor size.

BACKGROUND Motivational interviewing (MI) is a collaborative communication approach used to elicit autonomous motivation in health-related behavior change. Generative artificial intelligence (GenAI) provides new opportunities to deliver MI through conversational systems, but evidence on how these systems are designed, assessed, and translated into interventions remains fragmented. OBJECTIVE This scoping review aimed to characterize the current evidence on GenAI-MI chatbots across system design, safety measures, MI quality, user perceptions, and intervention outcomes. METHODS We conducted a scoping review in accordance with PRISMA-ScR. Studies published or publicly available from January 2015 to June 2, 2026 were identified through PubMed, Web of Science, Scopus, PsycINFO, DBLP, IEEE Xplore, ACM Digital Library, arXiv, and ACL Anthology. Eligible studies used generative AI to generate MI-related chatbot responses or counselor utterances. Data were extracted using a predefined extraction framework and synthesized descriptively. RESULTS Forty-seven reports comprising 48 studies were included. Twenty studies (41.7%) focused on system design without direct participant use, whereas 28 (58.3%) involved direct interaction with a GenAI-MI chatbot. Most systems were text based and disembodied, and 23 of 48 studies (47.9%) incorporated dynamic adaptation. Safety measures were unevenly reported, with privacy and data protection (20/48, 41.7%) and safety-oriented content generation (17/48, 35.4%) more commonly described than automated (7/48, 14.6%) or human (6/48, 12.5%) risk monitoring. Among studies involving direct participant use, 21 of 28 (75.0%) reported informed consent or user education. Thirty of 48 studies (62.5%) assessed MI quality using observer- or client-based evaluations. Existing observer and client evaluations generally suggested that GenAI-MI chatbots could produce MI-consistent interactions. User perceptions were generally favorable, particularly for empathy, usability, helpfulness, and intention to use, although measurement approaches were heterogeneous. Eighteen of 48 studies (37.5%) reported intervention outcomes, including applications in physical activity, smoking cessation, and alcohol or substance use. Most intervention studies involved a single session, and only 3 of 18 (16.7%) evaluated repeated use over 10 days to 4 weeks. Positive findings were reported more consistently for short-term motivation outcomes than for sustained behavioral or functional change. CONCLUSIONS Current evidence suggests that GenAI-MI chatbots can deliver MI-consistent interactions that are generally perceived favorably by users, but evidence supporting sustained behavioral or functional change remains limited. Future research should strengthen runtime safety monitoring, standardize MI quality assessment, and use longer-term comparative designs with behavioral and functional outcomes to determine whether short-term motivational changes could translate into meaningful intervention effects. CLINICALTRIAL OSF Registries xzw4g; https://osf.io/xzw4g

Background: Depression and obesity are major global health burdens, yet the co‐occurrence patterns and temporal trends of their comorbidity remain poorly characterised at the global level. Aims: To map the global patterns of co‐occurring depression and obesity, quantify temporal trends of depression, obesity and comorbidity, forecast the prevalence of comorbidity up to 2035 and compare social indicators across groups stratified by their combined temporal trends of depression and obesity. Methods: Data were sourced from the Global Burden of Disease Study 2023, the NCD Risk Factor Collaboration and the World Bank Open Data repository. Countries and territories ( n = 199) were classified into four co‐occurrence patterns based on the current prevalence of depression and obesity. Average annual percentage changes (AAPCs) were calculated to quantify temporal trends and define four trend groups. We forecast the prevalence of depression–obesity comorbidity across 199 countries and territories up to 2035 using an autoregressive integrated moving average model. The Kruskal–Wallis H test and Mann–Whitney U test were used to compare seven social indicators relevant to socioeconomic status and healthcare resources across these trend groups. Results: For the four co‐occurrence patterns, the consistently‐low‐prevalence group was primarily concentrated in Asia, the consistently‐high‐prevalence group in the Americas, the depression‐dominant group in Africa and the obesity‐dominant group in the Americas and Oceania. Concerning temporal trends, the prevalence of depression–obesity comorbidity increased globally (median AAPC: 4.66%; interquartile range [IQR]: 3.49%–6.30%). Notably, countries and territories in East Asia, Southeast Asia and sub‐Saharan Africa had higher comorbidity AAPCs despite their low current prevalence. Temporal trends were categorised into four patterns: 44 (22.11%) countries and territories were classified into the double‐fast‐growing group (Southeast Asia, sub‐Saharan Africa), 56 (28.14%) into the depression‐driven‐growing group (West Asia, North Africa, North America, Oceania), 56 (28.14%) into the obesity‐driven‐growing group (East Asia, Latin America and the Caribbean) and the remaining 43 (21.61%) into the double‐slow‐growing group (Europe, Australia and New Zealand). Predictive analysis indicated the median prevalence of depression–obesity comorbidity may increase to 833.32 (IQR: 532.07–1367.42) per 100 000 in 2030 and 869.35 (IQR: 544.11–1403.71) per 100 000 in 2035. Compared with the double‐slow‐growing group, the obesity‐driven and double‐fast‐growing groups showed poorer socioeconomic status and healthcare resources; the depression‐driven‐growing group showed poorer healthcare resources. Conclusions: The prevalence of depression–obesity comorbidity increased globally. Particular attention is warranted for countries and territories with currently low comorbidity prevalence but rapidly increasing trends. Our findings also highlight the need for accessible interventions for countries and territories with poorer socioeconomic status and healthcare resources.