Lung cancer, the malignancy with the highest global mortality rate, is currently undergoing a paradigm shift from precision medicine to intelligent medicine in its diagnostic and therapeutic models. Artificial intelligence (AI), leveraging its core advantages in multimodal data fusion and high-dimensional featureextraction, has deeply permeated the entire disease continuum of lung cancer screening, diagnosis, and individualized treatment. AI demonstrates significant value in improving patient outcomes and enhancing clinical efficiency, thereby reshaping the fundamental logic and clinical practice pathways of lung cancer management. This article reviews the latest advances of AI in real-world clinical diagnosis and treatment scenarios for lung cancer, with a focus on multimodal data fusion architectures, breakthrough applications of AI-assisted lung cancer diagnosis and treatment, and the practical barriers to clinical translation. It critically analyzes core challenges including data standardization, privacy protection, and ethical compliance. Finally, it envisions the future landscape of a nationwide intelligent ecosystem for lung cancer diagnosis and treatment—driven by federated learning and empowered by data elements—providing a reference for the standardized application and innovative development of AI in precision lung cancer care.

Ground-glass nodules (GGNs) are common imaging manifestation in the early screening of lung adenocarcinoma. With the widespread use of low-dose computed tomography (LDCT) in lung cancer screening, the detection rate of GGNs has significantly increased. According to the presence or absence of a solid component, GGNs are mainly classified into pure ground-glass nodules (pGGNs) and mixed ground-glassnodules (mGGNs), which differ in their natural course and biological behavior. In general, pGGNs tend to progress more slowly, whereas mGGNs are more likely to develop invasive features. The vast majority of pGGNs remain stable for years, but some pGGNs and mGGNs may show an increase in size or in the solid component. The "indolence" observed on the surface of GGNs hides complex genomic, metabolic, and immune changes, which are difficult to capture with traditional image-based data. In recent years, multi-omics analysis, radiomics, and artificial intelligence models have provided new tools for identifying high-risk GGNs. However, there is still controversy over the clinical generalizability, interpretability, and standardization of these models. Furthermore, there is no consensus on whether surgical resection is required. This article reviews the molecular mechanisms, metabolic, and immune microenvironment changes involved in the progression of GGNs, discusses the advantages and limitations of imaging prediction models, and combines domestic and international guidelines and survival studies to explore the controversial points in follow-up and surgical strategies, aiming to provide references for the personalized management of GGNs.

With the widespread application of low-dose computed tomography (CT), the detection rate of multiple lung cancers (MLCs) is gradually increasing. The diagnosis and treatment of MLCs have become a major challenge in clinical practice in thoracic surgery and oncology. In April 2025, the Lung Cancer Professional Committee of the China Anti-Cancer Association (CACA) organized multidisciplinary experts from both domestic and international fields to release the first edition of the CACA Expert Consensus on the Diagnosis and Treatment of Multiple Lung Cancers, providing systematic recommendations for the diagnostic system, molecularassessment strategies, and surgical and non-surgical management of MLCs. This article provides a detailed interpretation of the core content of this consensus and, by incorporating the latest research progress in the field, delves into the pathogenesis, precise diagnostic strategies, and individualized treatment pathways for multiple lung cancers, aiming to offer a more comprehensive reference for clinical practice.

To evaluate the clinical value of nebulized indocyanine green(ICG) combined with near-infrared fluorescence imaging for intraoperative detection of air leaks during pulmonary resection.

Based on plasma lipidomics combined with machine learning approaches, this study aimed to screen molecular biomarkers for the diagnosis of early-stage lung cancer in elderly patients and to evaluate their diagnostic performance.

In the tumor microenvironment, a bidirectional feedback regulation exists between aberrant angiogenesis and antitumor immunity, which provides a basis for the synergistic effects of antiangiogenic therapy and immunotherapy. The underlying mechanisms include relief of immunosuppression mediated by proangiogenic factors, induction of vascular normalization to reshape the immune microenvironment, and formation of a positive immune-vascular feedback loop. Based on these mechanisms, the combination of antiangiogenic therapy and immunotherapy has demonstrated synergistic efficacy in advanced non-small cell lung cancer (NSCLC) and is gradually being moved forward to the neoadjuvant setting. However, in the perioperative treatment of resectable lung cancer, this regimen still faces challenges, including primary resistance, a brief window of vascular normalization, and an unclear understanding of spatial interaction mechanisms. This review systematically elucidates the synergistic mechanisms of antiangiogenic agents combined with immunotherapy in NSCLC and the latest advances of this combination strategy in the neoadjuvant therapy setting, aiming to provide a reference for clinical practice.

Non-small cell lung cancer, characterized by high incidence and mortality rates, significantly threatens human health. Precisely assessing patient prognosis and implementing adaptive treatment strategies have emerged as pivotal issues in contemporary thoracic oncology. Postoperative minimal residual disease (MRD) detection through liquid biopsy has demonstrated substantial potential. Circulating tumor DNA (ctDNA) enables real-time dynamic monitoring of tumor alterations. With advancements in ctDNA detection technologies, ctDNA has become a core method for MRD assessment. In the context of postoperative surveillance, ctDNA detection facilitates more accurate prognostic stratification and early prediction of tumor recurrence. Regarding therapeutic interventions, ctDNA detection can predict the efficacy of neoadjuvant and adjuvant therapies. In the future, further elucidating the value of assessing ctDNA status in predicting patient prognosis and guiding drug therapies will contribute to the advancement of precision medicine and adaptive treatments.

Objective To investigate the predictive value of preoperative plasma total cholesterol levels for postoperative delirium(POD)following esophagectomy and to analyze its associated risk factors.Methods The complete perioperative electronic medical records of 187 patients with POD who underwent esophagectomy at the 960 Hospital of the People's Liberation Army between January 2019 and December 2023 were retrospectively analyzed.According to the occurrence of POD within one week postoperatively,the patients were categorized into a delirium group and a non-delirium group.The preoperative plasma total cholesterol levels were used to construct a receiver operating characteristic(ROC)curve for predicting delirium after esophagectomy.The area under the curve(AUC)and its 95%confidence interval(CI)were calculated.Additionally,the plasma total cholesterol level corresponding to the maximum Youden index(sensitivity+specificity-1)was determined as the optimal critical value.Results POD occurred in 24 of 187 patients who underwent esophageal cancer resection,with an incidence rate of 12.83%.Compared to the non-delirium group,the delirium group exhibited older age,lower intraoperative blood pressure,lower plasma total cholesterol and low-density lipoprotein levels,longer anesthesia duration,and a higher preva-lence of coronary heart disease,which was identified as a risk factor for POD.Dexmedetomidine administration was associated with a reduced occurrence of POD.The AUC,95%CI,cut-off value,sensitivity,and specificity of preop-erative plasma total cholesterol level for predicting POD were 0.761,0.657～0.865,≤4.11 mmol/L,70.8%,and 77.3%,respectively.Conclusions Among non-hyperlipidemic patients,low preoperative plasma total cholesterol levels and intraoperative hypotension are identified as independent risk factors for POD following esophageal cancer resection.Preoperative plasma total cholesterol levels exhibit moderate predictive value for the occurrence of POD.

Objective To investigate the predictive value of preoperative plasma total cholesterol levels for postoperative delirium(POD)following esophagectomy and to analyze its associated risk factors.Methods The complete perioperative electronic medical records of 187 patients with POD who underwent esophagectomy at the 960 Hospital of the People's Liberation Army between January 2019 and December 2023 were retrospectively analyzed.According to the occurrence of POD within one week postoperatively,the patients were categorized into a delirium group and a non-delirium group.The preoperative plasma total cholesterol levels were used to construct a receiver operating characteristic(ROC)curve for predicting delirium after esophagectomy.The area under the curve(AUC)and its 95%confidence interval(CI)were calculated.Additionally,the plasma total cholesterol level corresponding to the maximum Youden index(sensitivity+specificity-1)was determined as the optimal critical value.Results POD occurred in 24 of 187 patients who underwent esophageal cancer resection,with an incidence rate of 12.83%.Compared to the non-delirium group,the delirium group exhibited older age,lower intraoperative blood pressure,lower plasma total cholesterol and low-density lipoprotein levels,longer anesthesia duration,and a higher preva-lence of coronary heart disease,which was identified as a risk factor for POD.Dexmedetomidine administration was associated with a reduced occurrence of POD.The AUC,95%CI,cut-off value,sensitivity,and specificity of preop-erative plasma total cholesterol level for predicting POD were 0.761,0.657～0.865,≤4.11 mmol/L,70.8%,and 77.3%,respectively.Conclusions Among non-hyperlipidemic patients,low preoperative plasma total cholesterol levels and intraoperative hypotension are identified as independent risk factors for POD following esophageal cancer resection.Preoperative plasma total cholesterol levels exhibit moderate predictive value for the occurrence of POD.

Non-small cell lung cancer, characterized by high incidence and mortality rates, significantly threatens human health. Precisely assessing patient prognosis and implementing adaptive treatment strategies have emerged as pivotal issues in contemporary thoracic oncology. Postoperative minimal residual disease (MRD) detection through liquid biopsy has demonstrated substantial potential. Circulating tumor DNA (ctDNA) enables real-time dynamic monitoring of tumor alterations. With advancements in ctDNA detection technologies, ctDNA has become a core method for MRD assessment. In the context of postoperative surveillance, ctDNA detection facilitates more accurate prognostic stratification and early prediction of tumor recurrence. Regarding therapeutic interventions, ctDNA detection can predict the efficacy of neoadjuvant and adjuvant therapies. In the future, further elucidating the value of assessing ctDNA status in predicting patient prognosis and guiding drug therapies will contribute to the advancement of precision medicine and adaptive treatments.