Lung cancer remains the leading cause of cancer-related mortality worldwide and poses an especially severe burden in China. Low-dose computed tomography (LDCT) screening has significantly improved the detection rate of early-stage lung cancer; however, it has also introduced major challenges, including increased false-positive findings, overtreatment, difficulties in long-term follow-up, and regional disparities in healthcare resources. In recent years, the rapid development of artificial intelligence (AI), large language models (LLMs), the medical Internet of Things (MIoT), and Metaverse Medicine has driven the evolution of Medical AI Agents from simple imaging-assistance tools into novel digital medical entities capable of perception, reasoning, decision-making, execution, feedback, and continuous learning. Pulmonary Nodule Agents represent a new generation of digital medical systems built upon multimodal data integration, medical GPT technologies, knowledge graphs, and continuous digital healthcare frameworks. These systems can perform risk identification, dynamic stratification, pathway recommendation, long-term follow-up, and closed-loop management for pulmonary nodules, thereby enabling precision control of lung cancer risk throughout the entire clinical pathway. Currently, AI has evolved from traditional computer-aided detection (CAD) systems toward multi-agent collaborative architectures and is increasingly being integrated into advanced clinical scenarios, including digital multidisciplinary team (MDT) management, Hospital at Home, digital twins, and Metaverse Medicine. Drawing upon international research, consensus guidelines, and the BAIMGPT and PNapp 5A framework proposed by Professor Chunxue Bai’s team, this article systematically reviews the concepts, technological foundations, core architectures, clinical applications, real-world challenges, and future development trends of pulmonary nodule agents. Particular emphasis is placed on the role of AI in pulmonary nodule detection, risk stratification, multimodal integration, continuous follow-up, grassroots healthcare empowerment, and real-world governance. Evidence suggests that pulmonary nodule agents are not merely managing “nodules on imaging,” but rather the dynamic future risk of lung cancer in individual patients. In the future, such intelligent agents are expected to transform pulmonary nodule management from a paradigm of “detecting nodules” to one of “precision risk management,” ultimately advancing the vision of “preventive medicine by renowned physicians and universal healthcare enabled by metaverse medicine.”

Key Words: artificial intelligence; intelligent agent; pulmonary nodule; early lung cancer screening; medical GPT; multimodal integration

In ultrasound imaging, the pleura is visualized as echo reflection formed by the echoes of the interface between the pleura and the lung surface. Three major signs of pleura determines whether the patient has pneumothorax. In this paper, we propose a method to identify pleural line for the diagnosis of pneumothorax. Firstly, the gray threshold of ultrasonic image is properly classified by pre-experiment. Secondly, possible pleural line regions are identified based on threshold classification. Thirdly, the region of pleural line is identified based on the known characteristics of pleural line. The last step is to consider whether it is necessary to modify the threshold to accurately identify the pleural line region. Moreover, we tested 890 ultrasound samples, which included three categories: lung sliding, lung point, and lung sliding disappearance. Each category of samples was divided into two subsets, typical and atypical. The average identification rate reached 90.45%. According to the test results, the advantages and disadvantages of the proposed method as well as the further improvement direction were analyzed. This method for identifying pleural line can serve as the groundwork for developing automatic algorithm for diagnosing pneumothorax.

With the rapid development of artificial intelligence, deep learning technology has been widely applied across various fields. In the medical field, deep learning models, by analyzing medical images and clinical data, can automatically detect features of different types of lesions, such as polyps, ulcers, and cancers, thereby assisting physicians in early diagnosis of disease. This review provides an overview of recent progress in applying deep learning for disease diagnosis in various parts of the gastrointestinal tract, including the esophagus, stomach, small intestine, and colon. It also discusses the challenges and potential future directions for deep learning in this field.

Asthma management is shifting from the traditional model based predominantly on intermittent outpatient assessments toward a precision management model centered on continuous monitoring, dynamic early warning, stratified intervention, and outcome tracking. The Respiratory Internet of Things (Respiratory IoT), leveraging intelligent inhalers, home pulmonary function testing, pulse oximeter, wearable devices, environmental sensors, mobile terminals, and cloud platforms as its primary components, creates an integrated service network connecting hospital wards, outpatient clinics, community settings, and home environments. Research conducted by Professor Bai Chunxue’s team on metaverse medicine, new-quality productivity in medicine, and BAIMGPT (Bai’s Medical GPT) provides a theoretical and technical framework with distinct Chinese contextual characteristics，supporting the workflow of "multi-source sensing, intelligent analysis, digital human interaction, quality control, and closed-loop execution."This paper systematically reviews the application of the Respiratory IoT in inpatient asthma monitoring, chronic disease management, and hospital-community-home collaboration, summarizing research progress, practical challenges, and future directions for precision monitoring and closed-loop management in asthma, with the aim of informing the development of a continuous and integrated care system for respiratory diseases.

Key Words: asthma; respiratory internet of things; precision monitoring; closed-loop management; smart inhalers; digital medicine

Urinalysis is an essential diagnostic tool for urinary tract infections, kidney disease, diabetes, and other clinical conditions. Dipsticks, which allow for quick screening of urine specimens, are used in the clinic settings to identify the presence and concentration of labeled substances such as urine pH, urine protein, urine glucose, urine ketone, and urine nitrite. This paper reviews four urine dry chemical analysis methods, which are based on human eyes, integrating sphere, color sensors, and image sensors, respectively. The techniques of each method are also discussed.

Objective: To assess the impact of measuring central venous pressure (CVP) on the prognosis of patients with septic shock. Methods: Septic shock patients with and without CVP measurements were identified in the Medical Information Mart for Intensive Care IV database. The primary outcome was 28-day mortality, and a multivariate logistic regression model was used to analyze the association between CVP measurement and 28-day mortality in patients with septic shock. The results were validated using logistic regression after propensity score matching. Secondary outcomes were in-hospital mortality, 1-year mortality, incidence of acute kidney injury within the first 7 days in the intensive care unit (ICU), and length of stay in the ICU. After propensity score matching, logistic regression analysis was conducted to analyze the correlation between CVP measurements and secondary outcomes in patients with septic shock. Results: A total of 2966 patients were included, including 1219 patients whose CVP was measured within 24h after admission to the ICU. CVP measurement was found to be not correlated with 28-day mortality (odds ratio=0.978, 95% Confidence Interval 0.798-1.200, P=0.835). Analyzing the cohort after propensity score matching, CVP measurement was found to be associated with prolonged ICU stay (4.9 vs. 3.2 days; P<0.001). No statistical differences were found in the primary outcome and other secondary outcomes between those with CVP measurement and those not. Conclusion: CVP measurement is associated with prolonged ICU stay in patients with septic shock but not associated with mortality and incidence of acute kidney injury within 7 days. 

Background: Current prognostic research on sepsis-associated encephalopathy (SAE) predominantly focuses on elderly populations, while independent risk markers for young patients remain unclear. Objective: To investigate the prognostic value of red cell distribution width (RDW) for 30-day mortality in young SAE patients admitted to the intensive care unit (ICU). Methods: This retrospective cohort study analyzed 1,594 SAE patients (18-65 years) from the MIMIC-IV database. Using propensity score matching  (1:1 nearest-neighbor matching), 352 matched pairs were generated. RDW-mortality association was assessed through restricted cubic splines, multivariable Cox regression, and Kaplan-Meier analysis. Results: Among the 1,594 young patients with SAE analyzed, 144 subjects (9.0%) passed away within 30 days following their ICU admission. Non-survivors exhibited significantly higher baseline RDW than survivors (17.5±3.1 versus 14.9±2.4, P<0.001). Patients exhibiting elevated RDW showed higher rates of hepatic disorders, clotting dysfunction, and impaired kidney function (all P<0.001). RDW and 30-day post-ICU admission mortality were nonlinearly related. After matching (standardized mean difference <0.1 for all covariates), higher RDW values showed a notable association with greater mortality risk over a 30-day period (hazard ratio [HR]=2.7, 95% confidence interval [CI]: 1.4-5.3, P=0.003). Also, after comprehensive adjustment for covariates, each 1-unit increase in RDW was still associated with a 20% rise in the risk of death (HR=1.2, 95% CI: 1.1-1.4, P<0.001). Kaplan-Meier curves confirmed reduced 30-day survival in high-RDW group (log-rank test, P=0.002). Conclusions: Elevated RDW at ICU admission independently predicts 30-day mortality in young SAE patients.

Case analysis reporting is an important approach for general practitioners to transform clinical practice, diagnostic reasoning, follow-up observations, and reflective learning into sharable medical knowledge, with substantial clinical, educational, and research value. High-quality case analysis reporting can improve first-contact recognition, referral decisions, chronic disease management, and regional quality improvement, while also serving as an effective vehicle for case-based teaching, young physician training, and real-world evidence generation. However, in routine practice, general practitioners often face multiple barriers, including limited consultation time, incomplete data collection, weak diagnostic reasoning frameworks, insufficient standardized writing skills, difficulty in evidence retrieval, and low research conversion efficiency. Recent advances in generative artificial intelligence, large language models, natural language processing, multimodal AI, ambient clinical documentation tools, knowledge graphs, Internet of Things, and metaverse medicine have created new opportunities for empowering case analysis reporting in primary care. AI can support history taking, structured data extraction, reconstruction of disease timelines, problem representation, differential diagnosis prompting, evidence retrieval, case-based educational design, case repository development, and research transformation, thereby improving the completeness, standardization, interpretability, and reusability of case reports. Current studies suggest that AI has shown promising performance in complex diagnostic reasoning, clinical text generation, medical education, and documentation assistance. Nevertheless, real-world implementation in primary care remains constrained by hallucinations, bias, privacy risks, unclear accountability, limited external generalizability, and the potential erosion of clinicians’ independent reasoning ability. Looking forward, AI empowerment in primary care case analysis reporting should follow the principles of human-AI collaboration, physician leadership, factual verifiability, auditability, and gradual scenario-based deployment. The ultimate goal is not merely to help physicians write faster, but to build an intelligent case ecosystem that integrates clinical care, education, research, quality assurance, and regional knowledge sharing.

Key Words: artificial intelligence; general practitioners; case analysis reporting; large language models; medical education; real-world research