Kui Jiang

Background: Lifestyle interventions for patients with prostate cancer have been shown to improve treatment adherence and quality of life. However, there remains a lack of large language models (LLMs) capable of delivering individualized and professional lifestyle recommendations under clearly defined medical safety boundaries and controlled evidence sources. Objective: This study aimed to develop and evaluate a supervised fine-tuned LLM—PCaPLMM_SFT (Prostate Cancer Patient Lifestyle Management Model via Supervised Fine-Tuning)—to support health literacy improvement and lifestyle self-management among patients with prostate cancer. Methods: We searched English-language literature primarily from PubMed (February 2015 to February 2025) to build a structured lifestyle management knowledge base covering diet, physical activity, weight management, medication adherence, and psychological support. We used a retrieval-augmented generation pipeline to generate patient-style question-answer (QA) pairs from retrieved knowledge slices. Bilingual English-Chinese QA data were generated from English-language source evidence through patient-oriented reformulation and retrieval-augmented generation–based answer generation, and independent English and Chinese test sets were constructed to assess bilingual QA performance. We trained Baichuan2-7B-Chat using a 2-stage strategy, consisting of continued pretraining, followed by supervised fine-tuning with low-rank adaptation. Model outputs were evaluated in 2 double-blind rounds by referee LLMs (Qwen3-Max and DeepSeek-R1) and compared with GPT-3.5-Turbo and the base Baichuan2-7B-Chat using 2500 queries across 5 lifestyle scenarios. Additionally, 3 domain experts conducted a blinded review of 50 QA samples (10 per scenario). We used the Mann-Whitney U test with effect size r , and Benjamini-Hochberg false discovery rate correction, and examined consistency using intraclass correlation coefficients. Results: Based on 2211 included publications, we constructed the PCaPLMM_SFT-Train dataset. The knowledge base yielded >150,000 structured knowledge slices. After 2 rounds of review, we obtained 42,330 single-turn QA pairs and 3008 multiturn dialogues, and the supervised fine-tuning phase used 45,338 structured QA samples. In the dual-round referee LLM assessment, PCaPLMM_SFT consistently outperformed Baichuan2-7B-Chat across dimensions and showed comparable or superior performance to GPT-3.5-Turbo across 5 lifestyle scenarios. Consistency analyses indicated moderate to good agreement between referee models across rounds, supporting the robustness of the comparative evaluation. Conclusions: PCaPLMM_SFT demonstrates the feasibility of constructing a medical lifestyle–focused LLM by integrating structured medical knowledge, QA-style training data, and a multilayer evaluation system. This framework provides a reproducible methodological foundation for evidence-based health education and lifestyle management and establishes groundwork for future evaluation in real-world health management settings.

Epilepsy is a neurological disorder characterized by transient cerebral dysfunction arising from abnormal cortical neuronal discharges, which can precipitate sudden muscle contractions, loss of consciousness, and convulsions. As one of the most prevalent brain disorders, it affects a substantial share of the global population. Electroencephalography (EEG) — the most widely used and extensively studied noninvasive modality for monitoring cerebral electrical activity — underpins clinical and computational approaches to epilepsy detection. In this work, we propose a Consistency-Regularized Multi-View Transfer Learning (CR-MVTL) algorithm for epileptic EEG analysis and AI-assisted detection. Compared with prevailing methods, CR-MVTL offers: (i) strong cross-view complementarity that promotes information sharing and improves recognition; (ii) high generalizability that enables rapid parameter adaptation when target-domain data are scarce; and (iii) dynamic view-weighting that attenuates weak views while amplifying informative ones. We evaluate CR-MVTL across 12 simulated multi-view transfer scenarios. Extensive experiments demonstrate absolute improvements in average accuracy of 7.25 and 6.41 percentage points in the two-view and three-view settings, respectively, over competitive baselines.

In recent years, the application of machine learning techniques for identifying epileptic electroencephalogram (EEG) signals has become increasingly prevalent. In this study, we propose a method called Multi-View Learning and Regularized Label Relaxation Linear Regression (MVRLR) for the recognition of epileptic EEG signals. To be specific, first, we generate multi-view data by employing three different feature extraction methods to capture the EEG features from different perspectives. Then, in the linear regression framework, we introduce a nonnegative label relaxation matrix to transform the strict binary label matrix into a relaxed variable matrix, allowing for more flexible label fitting and expanding the boundaries between different classes. Additionally, we utilize manifold learning to construct a class tightness graph, which serves as a regularization term to prevent overfitting. Finally, we employ Shannon entropy to quantify the overall uncertainty within the probability distribution. To validate the effectiveness of the proposed method, we conduct experiments using a multi-view epileptic EEG dataset and compare it with several other classification algorithms. The extensive experimental results demonstrate that the proposed MVRLR method achieves higher classification accuracy. By applying the proposed method, we can more accurately identify epileptic EEG signals, providing robust support for early diagnosis and personalized treatment of epilepsy.

Objectives: To achieve a rapid assessment of health literacy (HL) levels among diabetic patients. Methods: A questionnaire survey was conducted among diabetic patients from nine communities in Nantong City, Jiangsu Province, China, using convenient sampling. Based on the survey results, data from three communities were randomly selected as the test set, with the remaining data used as the training set. Feature selection was performed using recursive feature elimination. Predictive models were established and compared using logistic regression (LR), random forest (RF), and support vector machine (SVM). Calibration curves, decomposition plots, and partial dependence plots were drawn to evaluate and interpret the models. Results: In November 2023, a total of 802 valid questionnaires were received. Eight variables were selected for modeling: educational level, exercise habits, average monthly household income, dietary control, age, medication for blood sugar control, duration of diabetes, and number of cohabitants. The recall for LR in the three communities was 0.778, 0.800, and 0.862 [area under the curve (AUC): 0.810, 0.792, and 0.775]. For RF, the recall values were 0.879, 0.877, and 0.923 (AUC: 0.781, 0.710, and 0.710). For SVM, the recall values were 0.859, 0.862, and 0.877 (AUC: 0.813, 0.759, and 0.770). Model evaluation showed that as the data volume increased, the calibration curves became more ideal. Conclusions: As one of the few HL prediction models for diabetic patients in mainland China that is built based on multi-center survey data and evaluated through multi-center assessment, this model can quickly identify patients with insufficient HL using a small amount of objective personal information.

Purpose: This study is aimed at investigating health literacy (HL) among diabetes mellitus (DM) patients using a comprehensive, scientific, feasible, and suitable HL assessment indicator system tailored for the diabetic population in mainland China and systematically analyzing the factors influencing HL in this population.Methods: The Delphi expert consultation method was employed to initially draft and refine the Diabetes Health Literacy Evaluation Indicator System (DHLEIS). The reliability and representativeness of the indicator system were tested through metrics including the active coefficient, authority degree, and coordination degree. A HL survey questionnaire for diabetic patients was developed based on DHLEIS and administered to diabetic patients across five hospitals in Nantong and Yancheng cities, Jiangsu Province. The random forest method was utilized to deeply analyze the impact of various factors on HL and its four dimensions and to identify the core influencing factors.Results: Analysis of 707 questionnaires based on the DHLEIS revealed that nine factors—age, sex, body shape, income, exercise, education level, duration of DM, whether insulin is injected, and the number of cohabitants—significantly impact the HL levels. Among these, age, duration of DM, education level, and number of cohabitants were particularly influential across the four dimensions of health knowledge, awareness, behavior, and skills. Factors related to health knowledge and skills were the most significant contributors to overall HL.Conclusions: The multidimensional analysis of factors influencing HL offers valuable insights into characterizing varying levels of HL among diabetic patients. This approach supports targeted cognitive improvements and the effective enhancement of health skills, ultimately leading to better health outcomes.

Relevant studies published between January 2010 and June 2021 were identified through relevant databases, including the Science Citation Index Expanded (SCIE) database of Web of Science, PubMed, and Embase, in order to assess the effect of health literacy (HL) intervention on patients with diabetes. A total of 21 articles were eligible. The results showed that: (1) this review involved different HL assessment tools, most of which were self-designed scales and assessment tools focused on measuring functional HL. (2) The differences in glycosylated hemoglobin (HbA1c) (weighted mean difference [WMD] = −0.78, 95% confidence interval [CI]: −0.94, −0.62) and medication adherence (standardized mean difference [SMD] = 1.85, 95% CI: 0.19, 3.52) between the HL intervention group and the usual care group were statistically significant. There was no significant improvement in systolic blood pressure (SMD = −0.05, 95% CI: −0.34, 0.25). Furthermore, this review reported that self-efficacy (SMD = 0.85, 95% CI: 0.65, 1.04) was increased, and the level of HL was improved. In the assessments of risk of bias, 90% of the studies were classified as medium. The quality of the evidence of medication adherence was very low, and the reliability of the conclusions was not enough to confirm the effect of HL.

Background: Gestational diabetes mellitus (GDM) is a common endocrine metabolic disease, involving a carbohydrate intolerance of variable severity during pregnancy. The incidence of GDM-related complications and adverse pregnancy outcomes has declined, in part, due to early screening. Machine learning (ML) models are increasingly used to identify risk factors and enable the early prediction of GDM. Objective: The aim of this study was to perform a meta-analysis and comparison of published prognostic models for predicting the risk of GDM and identify predictors applicable to the models. Methods: Four reliable electronic databases were searched for studies that developed ML prediction models for GDM in the general population instead of among high-risk groups only. The novel Prediction Model Risk of Bias Assessment Tool (PROBAST) was used to assess the risk of bias of the ML models. The Meta-DiSc software program (version 1.4) was used to perform the meta-analysis and determination of heterogeneity. To limit the influence of heterogeneity, we also performed sensitivity analyses, a meta-regression, and subgroup analysis. Results: A total of 25 studies that included women older than 18 years without a history of vital disease were analyzed. The pooled area under the receiver operating characteristic curve (AUROC) for ML models predicting GDM was 0.8492; the pooled sensitivity was 0.69 (95% CI 0.68-0.69; P<.001; I2=99.6%) and the pooled specificity was 0.75 (95% CI 0.75-0.75; P<.001; I2=100%). As one of the most commonly employed ML methods, logistic regression achieved an overall pooled AUROC of 0.8151, while non–logistic regression models performed better, with an overall pooled AUROC of 0.8891. Additionally, maternal age, family history of diabetes, BMI, and fasting blood glucose were the four most commonly used features of models established by the various feature selection methods. Conclusions: Compared to current screening strategies, ML methods are attractive for predicting GDM. To expand their use, the importance of quality assessments and unified diagnostic criteria should be further emphasized.

Objective. To systematically evaluate the effects of health literacy intervention on health literacy level and glycolipid metabolism of people with diabetes in mainland China. Methods. A systematic review of journal articles discussing diabetes and health literacy was performed by searching PubMed, Embase, the Science Citation Index Expanded (SCIE) database of Web of Science, the China National Knowledge Infrastructure (CNKI) database, the Chinese Scientific and Technical Journals database (CQVIP), and the Wanfang database. Cochrane Effective Practice and Organization of Care Review Group (EPOC) standards were applied for quality assessment. A meta-analysis was performed using Stata 12.0 software. Results. A total of 44 articles, including seven controlled before-and-after trials (CBAs), 27 randomized controlled trials (RCTs), and 10 nonrandomized controlled trials (non-RCTs), were included. The results showed that (1) health literacy level in the intervention group was improved compared with the preintervention and the control group; (2) fasting plasma glucose (FPG) (standardized mean difference , 95% CI: −2.28, −1.42), 2-hour plasma glucose (2hPG) ( , 95% CI: −2.68, −1.68), and HbA1c (weighted mean difference , 95% CI: −1.48, −0.94) were significantly reduced in the intervention group; (3) total cholesterol (TC) ( , 95% CI: −0.64, −0.23) was significantly reduced in the intervention group, although there were no statistically significant differences for triglycerides (TG) ( , 95% CI: −0.73, 0.05), low-density lipoprotein cholesterol (LDL-C) ( , 95% CI: −0.46, 0.07), or high-density lipoprotein cholesterol (HDL-C) ( , 95% CI: −0.29, 0.17). Conclusion. Intervention based on health literacy can effectively improve health literacy levels and reduce glucose metabolism and TC level among people with diabetes mellitus, although it has no significant effect on TG, LDL-C, or HDL-C.

AIM: To verify the association between retinopathy, nephropathy, and periodontitis in type 2 diabetic (T2D) patients. METHODS: Several electronic databases were available for our comprehensive search including China National Knowledge Infrastructure (CNKI), Chinese VIP Information (VIP), Wanfang, Web of Science, ScienceDirect and PubMed and were queried for relevant citations (updated to Mar. 2019). RevMan was utilized to perform Meta analysis and publication bias detection. After evaluation of the methodological quality of included studies, a fixed or random effect model was utilized to analyze data from included studies. RESULTS: A total of eight articles were finally included in this Meta analysis. In all 3987 subjects, there were 1207 T2D patients accompanying with microvascular complications and 1734 patients with periodontitis as well. The Meta forest plot presented little heterogeneity of the eight studies (P<0.00001, I2=89%). The total effect demonstrated periodontitis was associated with overall microvascular complications (OR: 1.96, 95%CI: 1.67-2.30, Z=8.25, P<0.00001). Subgroup investigations among the studies in Asian (OR: 2.33, 95%CI: 1.91-2.85) and North American (OR: 1.42, 95%CI: 1.08-1.86) populations confirmed the existed association between retinopathy, nephropathy, and periodontitis. While the strength of such associations between periodontitis and diabetic microvascular complications were more obvious in the Asians than North Americans. All the results indicated that periodontitis was associated with diabetic retinopathy (OR: 3.77, 95%CI: 2.71-5.24), diabetic nephropathy (OR: 1.55, 95%CI: 1.24-1.94) in T2D patients. CONCLUSION: The periodontitis is associated with diabetic retinopathy, diabetic nephropathy among T2D patients and further large sample size clinical trials are in need to confirm the findings.

BACKGROUND Gestational diabetes mellitus (GDM) is a common endocrine metabolic disease, involving a carbohydrate intolerance of variable severity during pregnancy. The incidence of GDM-related complications and adverse pregnancy outcomes has declined, in part, due to early screening. Machine learning (ML) models are increasingly used to identify risk factors and enable the early prediction of GDM. OBJECTIVE The aim of this study was to perform a meta-analysis and comparison of published prognostic models for predicting the risk of GDM and identify predictors applicable to the models. METHODS Four reliable electronic databases were searched for studies that developed ML prediction models for GDM in the general population instead of among high-risk groups only. The novel Prediction Model Risk of Bias Assessment Tool (PROBAST) was used to assess the risk of bias of the ML models. The Meta-DiSc software program (version 1.4) was used to perform the meta-analysis and determination of heterogeneity. To limit the influence of heterogeneity, we also performed sensitivity analyses, a meta-regression, and subgroup analysis. RESULTS A total of 25 studies that included women older than 18 years without a history of vital disease were analyzed. The pooled area under the receiver operating characteristic curve (AUROC) for ML models predicting GDM was 0.8492; the pooled sensitivity was 0.69 (95% CI 0.68-0.69; P<.001; I2=99.6%) and the pooled specificity was 0.75 (95% CI 0.75-0.75; P<.001; I2=100%). As one of the most commonly employed ML methods, logistic regression achieved an overall pooled AUROC of 0.8151, while non–logistic regression models performed better, with an overall pooled AUROC of 0.8891. Additionally, maternal age, family history of diabetes, BMI, and fasting blood glucose were the four most commonly used features of models established by the various feature selection methods. CONCLUSIONS Compared to current screening strategies, ML methods are attractive for predicting GDM. To expand their use, the importance of quality assessments and unified diagnostic criteria should be further emphasized.