Performance of ChatGPT Across Different Versions in Medical Licensing Examinations Worldwide: Systematic Review and Meta-Analysis.

BACKGROUND: Over the past 2 years, researchers have used various medical licensing examinations to test whether ChatGPT (OpenAI) possesses accurate medical knowledge. The performance of each version of ChatGPT on the medical licensing examination in multiple environments showed remarkable differences. At this stage, there is still a lack of a comprehensive understanding of the variability in ChatGPT's performance on different medical licensing examinations. OBJECTIVE: In this study, we reviewed all studies on ChatGPT performance in medical licensing examinations up to March 2024. This review aims to contribute to the evolving discourse on artificial intelligence (AI) in medical education by providing a comprehensive analysis of the performance of ChatGPT in various environments. The insights gained from this systematic review will guide educators, policymakers, and technical experts to effectively and judiciously use AI in medical education. METHODS: We searched the literature published between January 1, 2022, and March 29, 2024, by searching query strings in Web of Science, PubMed, and Scopus. Two authors screened the literature according to the inclusion and exclusion criteria, extracted data, and independently assessed the quality of the literature concerning Quality Assessment of Diagnostic Accuracy Studies-2. We conducted both qualitative and quantitative analyses. RESULTS: A total of 45 studies on the performance of different versions of ChatGPT in medical licensing examinations were included in this study. GPT-4 achieved an overall accuracy rate of 81% (95% CI 78-84; P<.01), significantly surpassing the 58% (95% CI 53-63; P<.01) accuracy rate of GPT-3.5. GPT-4 passed the medical examinations in 26 of 29 cases, outperforming the average scores of medical students in 13 of 17 cases. Translating the examination questions into English improved GPT-3.5's performance but did not affect GPT-4. GPT-3.5 showed no difference in performance between examinations from English-speaking and non-English-speaking countries (P=.72), but GPT-4 performed better on examinations from English-speaking countries significantly (P=.02). Any type of prompt could significantly improve GPT-3.5's (P=.03) and GPT-4's (P<.01) performance. GPT-3.5 performed better on short-text questions than on long-text questions. The difficulty of the questions affected the performance of GPT-3.5 and GPT-4. In image-based multiple-choice questions (MCQs), ChatGPT's accuracy rate ranges from 13.1% to 100%. ChatGPT performed significantly worse on open-ended questions than on MCQs. CONCLUSIONS: GPT-4 demonstrates considerable potential for future use in medical education. However, due to its insufficient accuracy, inconsistent performance, and the challenges posed by differing medical policies and knowledge across countries, GPT-4 is not yet suitable for use in medical education. TRIAL REGISTRATION: PROSPERO CRD42024506687; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=506687.

Edge and dense attention U-net for atrial scar segmentation in LGE-MRI.

The segmentation of atrial scars in LGE-MRI images has huge potential value for clinical diagnosis and subsequent treatment. In clinical practice, atrial scars are usually manually calibrated by experienced experts, which is time-consuming and prone to errors. However, automatic segmentation also faces difficulties due to myocardial scars' small size and variable shape. The present study introduces a dual branch network, incorporating edge attention, and deep supervision strategy. Edge attention is introduced to fully utilize the spatial relationship between the scar and the atrium. Besides, dense attention is embedded in bottom layer to solve feature disappearance. At the same time, deep supervision accelerates the convergence of the model and improves segmentation accuracy. The experiments were conducted on the 2022 atrial and scar segmentation challenge dataset. The results demonstrate that the proposed method has achieved superior performance.

Bidirectional causal associations between aging and major mental disorders: A population-based study using the two-sample mendelian randomization method from the UK biobank (AM-SRNMA 002).

AIMS: While observational studies have suggested associations linking aging and mental disorders, the question of causality has remained unclear. This study aimed to explore the causal relationship between aging level and major mental disorders. METHODS: We utilized Two-Sample Mendelian randomization (2SMR) with mental disorders data and aging indicators information from an extensive genome-wide association study (GWAS) database. The GWAS database is a comprehensive resource that compiles genetic association data, encompassing a sample size of over 450,000 individuals. We employed five methods for 2SMR and single nucleotide polymorphisms were chosen as instrumental variables. RESULTS: Our analyses consistently supported a bidirectional causal association between the Frailty Index (FI) and Major Depressive Disorder (MDD). Furthermore, our findings indicated potential influences, such as Attention-Deficit/Hyperactivity Disorder (ADHD) and Bipolar Disorder (BD) affecting GrimAge, and Anxiety Disorder (AD) impacting Left Hand Grip Strength (LHGS). In contrast, we observed no significant correlations for other mental disorders on FI, Telomere Length (TL), GrimAge, Appendicular Lean Mass (ALM), and LHGS. In the reverse direction, FI showed a significant impact on the risk of MDD, AD, and ADHD, while LHGS affected the risk of MDD. Importantly, no significant associations were found between other factors and the risk of MDD, BD, AD, Schizophrenia (SZ), and ADHD. CONCLUSIONS: This 2SMR analysis has presented evidence for a bidirectional causal relationship between FI and MDD, while the relationship between ADHD, BD, and GrimAge should be more considered. Our study provides genetic evidence supporting a causal link between aging indicators and several mental illnesses.

The weight of weights: Causal impact of overweight on major depressive disorder and its disparate association with suicide attempts in over 450,000 individuals of European ancestry (AM-SRNMA 003).

AIM: Some studies and societal discussions have suggested a potential link between overweight and a higher risk of mental disorders and suicidal ideation. However, the causal relationships between these factors remain unclear. This study aims to assess the bidirectional causal associations between overweight and mental disorders, including suicidal attempts. METHOD: We conducted a Two-Sample Mendelian Randomization study utilizing data from over 450,000 individuals of European ancestry sourced from a comprehensive Genome-Wide Association Study database. We chose single-nucleotide polymorphisms as instrumental variables. RESULTS: Our analyses consistently supported a unidirectional causal association of overweight with the increased risk of major depressive disorder (MDD, ß = 0.152, se = 0.069, P = 0.027), bipolar affective disorders (BD, ß = 0.197, se = 0.092, P = 0.033), and attention-deficit / hyperactivity disorder (ADHD, ß = 0.308, se = 0.080, P = 1.366 × 10-4). We observed no significant causal relationships for the exposure of overweight to anxiety disorder (AD), manic episode (MA), panic disorder (PD), schizophrenia (SZ), substance use disorder (SUD), autism spectrum disorder (ASD), ever attempted suicide, recent thoughts of suicide or self-harm, and suicide or other intentional self-harm events. CONCLUSION: This study has provided evidence for the causal relationship between overweight and MDD, BD, ADHD, with no observed relationship between overweight and AD, MA, PD, SZ, SUD, ASD, ever attempted suicide, recent thoughts of suicide or self-harm, and suicide or other intentional self-harm events.

Oxidation enhances the toxicity of polyethylene microplastics to mouse eye: Perspective from in vitro and in vivo.

Microplastics (MPs) are ubiquitously dispersed in the environment, and undergoing the process of oxidation that alters their physical and chemical properties. Eyes, which directly interface with the external milieu, inevitably encounter MPs. Nonetheless, the ophthalmic toxicity of MPs towards organisms remains unclear. In this study, primary mouse corneal epithelial cells (MCECs), C57BL/6 mice, and CX3CrlGFP/+ mice were utilized to evaluate the toxicity and differences between oxidized low-density polyethylene MPs(modified-MPs)and low-density polyethylene MPs (virgin-MPs) on eyes. The results manifested that virgin-MPs and modified-MPs could be endocytosed by primary MCECs, resulting in a range of cellular damage. Furthermore, they could diminish tear secretion, increase intraocular pressure, and could be internalized into cornea and retina in mice, instigating a series of detrimental reactions. Importantly, modified-MPs exhibited heightened toxicity towards mouse eyes, seemingly due to oxidation enhances the interaction between virgin-MPs/modified-MPs and tissues/cells, and leading to the release of toxic substances increased. In conclusion, our discoveries demonstrate that oxidation exacerbates the harm of virgin-MPs to eyes, and are of great significance for evaluating the risk of MPs to ocular health.

HMMR triggers immune evasion of hepatocellular carcinoma by inactivation of phagocyte killing.

Hepatocellular carcinoma (HCC) acquires an immunosuppressive microenvironment, leading to unbeneficial therapeutic outcomes. Hyaluronan-mediated motility receptor (HMMR) plays a crucial role in tumor progression. Here, we found that aberrant expression of HMMR could be a predictive biomarker for the immune suppressive microenvironment of HCC, but the mechanism remains unclear. We established an HMMR-/- liver cancer mouse model to elucidate the HMMR-mediated mechanism of the dysregulated "don't eat me" signal. HMMR knockout inhibited liver cancer growth and induced phagocytosis. HMMRhigh liver cancer cells escaped from phagocytosis via sustaining CD47 signaling. Patients with HMMRhighCD47high expression showed a worse prognosis than those with HMMRlowCD47low expression. HMMR formed a complex with FAK/SRC in the cytoplasm to activate NF-κB signaling, which could be independent of membrane interaction with CD44. Notably, targeting HMMR could enhance anti-PD-1 treatment efficiency by recruiting CD8+ T cells. Overall, our data revealed a regulatory mechanism of the "don't eat me" signal and knockdown of HMMR for enhancing anti-PD-1 treatment.

SD-YOLOv8: An Accurate Seriola dumerili Detection Model Based on Improved YOLOv8.

Accurate identification of Seriola dumerili (SD) offers crucial technical support for aquaculture practices and behavioral research of this species. However, the task of discerning S. dumerili from complex underwater settings, fluctuating light conditions, and schools of fish presents a challenge. This paper proposes an intelligent recognition model based on the YOLOv8 network called SD-YOLOv8. By adding a small object detection layer and head, our model has a positive impact on the recognition capabilities for both close and distant instances of S. dumerili, significantly improving them. We construct a convenient S. dumerili dataset and introduce the deformable convolution network v2 (DCNv2) to enhance the information extraction process. Additionally, we employ the bottleneck attention module (BAM) and redesign the spatial pyramid pooling fusion (SPPF) for multidimensional feature extraction and fusion. The Inner-MPDIoU bounding box regression function adjusts the scale factor and evaluates geometric ratios to improve box positioning accuracy. The experimental results show that our SD-YOLOv8 model achieves higher accuracy and average precision, increasing from 89.2% to 93.2% and from 92.2% to 95.7%, respectively. Overall, our model enhances detection accuracy, providing a reliable foundation for the accurate detection of fishes.

Anti-BCMA CAR-T therapy for multiple myeloma with extramedullary disease: A case report and review of the literature.

INTRODUCTION: Multiple myeloma (MM) with extramedullary disease (EMD) is rare in clinical practice, and B cell maturation antigen (BCMA) CAR-T cell therapy is a novel therapy for hematologic malignancies. Very few reports have been published on the effect of CAR-T-cell therapy in MM with EMD. Here, we report a case of MM with extramedullary lesions treated with BCMA CAR-T therapy. CASE PRESENTATION: A 66-year-old female patient presented to our hospital with an enlarged left maxillary gingiva. DIAGNOSIS: Diagnosis of indolent MM stage III (DS staging) and stage III (ISS and R ISS) with extramedullary lesions. INTERVENTION: The patient underwent a clinical trial of humanized anti-BCMA CAR T cell therapy. RESULTS: Symptoms improved; left gingival hyperplasia and swelling resolved; left buccal mass resolved; and neck and submandibular masses resolved. Pathological examination of the exfoliated masses showed necrotic tissue. CONCLUSION: MM with extramedullary lesions often has limited treatment options, and traditional chemotherapy methods are ineffective; however, BCMA CAR-T cell therapy can significantly improve the symptoms of extramedullary lesions in MM.

Integrating forest restoration into land-use planning at large spatial scales.

Forest restoration is being scaled up globally, carrying major expectations of environmental and societal benefits. Current discussions on ensuring the effectiveness of forest restoration are predominantly focused on the land under restoration per se. But this focus neglects the critical issue that land use and its drivers at larger spatial scales have strong implications for forest restoration outcomes, through the influence of landscape context and, importantly, potential off-site impacts of forest restoration that must be accounted for in measuring its effectiveness. To ensure intended restoration outcomes, it is crucial to integrate forest restoration into land-use planning at spatial scales large enough to account for - and address - these larger-scale influences, including the protection of existing native ecosystems. In this review, we highlight this thus-far neglected issue in conceptualizing forest restoration for the delivery of multiple desirable benefits regarding biodiversity and ecosystem services. We first make the case for the need to integrate forest restoration into large-scale land-use planning, by reviewing current evidence on the landscape-level influences and off-site impacts pertaining to forest restoration. We then discuss how science can guide the integration of forest restoration into large-scale land-use planning, by laying out key features of methodological frameworks required, reviewing the extent to which existing frameworks carry these features, and identifying methodological innovations needed to bridge the potential shortfall. Finally, we critically review the status of existing methods and data to identify future research efforts needed to advance these methodological innovations and, more broadly, the effective integration of forest restoration design into large-scale land-use planning.

Redox Approaches to Carbene Generation in Catalytic Cyclopropanation Reactions.

Transition metal-catalyzed carbene transfer reactions have a century-old history in organic chemistry and are a primary method for the synthesis of cyclopropanes. Much of the work in this field has focused on the use of diazo compounds and related precursors, which can transfer a carbene fragment to a catalyst with concomitant loss of a stable byproduct. Despite the utility of this approach, there are persistent limitations in the scope of viable carbenes, most notably those lacking stabilizing substituents. By coupling carbene transfer chemistry with two-electron redox cycles, it is possible to expand the available starting materials that can be used as carbene precursors. In this Minireview, we discuss emerging catalytic reductive cyclopropanation reactions using either gem-dihaloalkanes or carbonyl compounds. This strategy is inspired by classic stoichiometric transformations, such as the Simmons-Smith cyclopropanation and the Clemmensen reduction, but instead entails the formation of a catalytically generated transition metal carbene or carbenoid. We also present recent efforts to generate carbenes directly from methylene (CR2H2) groups via a formal 1,1-dehydrogenation. These reactions are currently restricted to substrates containing electron-withdrawing substituents, which serve to facilitate deprotonation and subsequent oxidation of the anion.

Light-Driven Photocatalyst-Free Synthesis of ß, δ-Functionalized Ketones from Aldehydes.

The synthesis of ketones has been a long focus of chemistry research, on account of its unique reactivity. Herein, we report a simple light-driven photocatalyst-free synthesis of ß, δ-functionalized ketones from aldehydes, using inexpensive and commercially abundant feedstock chemicals. This reaction is enabled by the direct acyl radical generation via hydrogen atom transfer and the subsequent radical addition process, avoiding the need for prefunctionalized substrates and organometallic reagent.

Oxidized/unmodified-polyethylene microplastics neurotoxicity in mice: Perspective from microbiota-gut-brain axis.

Microplastics (MPs) are inevitably oxidized in the environment, and their potential toxicity to organisms has attracted wide attention. However, the neurotoxicity and mechanism of oxidized polyethylene (Ox-PE) MPs to organisms remain unclear. Herein, we prepared oxidized low-density polyethylene (Ox-LDPE) and established a model of MPs exposure by continuously orally gavage of C57BL/6 J mice with LDPE-MPs/Ox-LDPE-MPs for 28 days with or without oral administration of Lactobacillus plantarum DP189 and galactooligosaccharides (DP189&GOS). The experimental results indicated that LDPE-MPs or Ox-LDPE-MPs caused several adverse effects in mice, mainly manifested by behavioral changes, disruption of the intestinal and blood-brain barrier (BBB), and simultaneous oxidative stress, inflammatory reactions, and pathological damage in the brain and intestines. Brain transcriptomic analysis revealed that the cholinergic synaptic signaling pathways, which affect cognitive function, were significantly disrupted after exposure to LDPE-MPs or Ox-LDPE-MPs. Real-time quantitative polymerase chain reaction and Western Blotting results further demonstrated that the critical genes (Slc5a7, Chat and Slc18a3) and proteins (Chat and Slc18a3) in the cholinergic synaptic signaling pathway were significantly down-regulated after exposure to LDPE-MPs or Ox-LDPE-MPs. These alterations lead to reduced acetylcholine concentration, which causes cognitive dysfunction in mice. Importantly, the DP189&GOS interventions effectively mitigated the MPs-induced cognitive dysfunction and intestinal microbiota alteration, improved intestinal and BBB integrity, attenuated the oxidative stress and inflammatory response, and also saw a rebound in the release of acetylcholine. These results indicated that LDPE-MPs and Ox-LDPE-MPs exert neurotoxic effects on mice by inducing oxidative stress, inflammatory responses, and dysregulation of cholinergic signaling pathways in the mouse brain. That probiotic supplementation is effective in attenuating MPs-induced neurotoxicity in mice. Overall, this study reveals the potential mechanisms of neurotoxicity of LDPE-MPs and Ox-LDPE-MPs on mice and their improvement measures, necessary to assess the potential risks of plastic contaminants to human health.

Light-Driven Three-Component Carbonylation of Aryl Halides Using Abundant Metal Carbonyl.

Carbonyl compounds are widely found in various pharmaceutical intermediates and synthetic precursors. Herein we report a simple light-driven three-component aryl halide process for synthesizing a variety of carbonylation products, utilizing Co2(CO)8 as an abundant solid carbonyl source, in good to excellent yields. The products can easily be subjected to further functionalization in synthesis. Mechanism studies indicated that this reaction is enabled by aryl radical generation and the subsequent CO insertion, alkene insertion, and protonation process.

Accuracy and Usability of Smartphone-Based Distance Estimation Approaches for Visual Assistive Technology Development.

Goal: Distance information is highly requested in assistive smartphone Apps by people who are blind or low vision (PBLV). However, current techniques have not been evaluated systematically for accuracy and usability. Methods: We tested five smartphone-based distance-estimation approaches in the image center and periphery at 1-3 meters, including machine learning (CoreML), infrared grid distortion (IR_self), light detection and ranging (LiDAR_back), and augmented reality room-tracking on the front (ARKit_self) and back-facing cameras (ARKit_back). Results: For accuracy in the image center, all approaches had <±2.5 cm average error, except CoreML which had ±5.2-6.2 cm average error at 2-3 meters. In the periphery, all approaches were more inaccurate, with CoreML and IR_self having the highest average errors at ±41 cm and ±32 cm respectively. For usability, CoreML fared favorably with the lowest central processing unit usage, second lowest battery usage, highest field-of-view, and no specialized sensor requirements. Conclusions: We provide key information that helps design reliable smartphone-based visual assistive technologies to enhance the functionality of PBLV.

Exploiting Geometric Features via Hierarchical Graph Pyramid Transformer for Cancer Diagnosis Using Histopathological Images.

Cancer is widely recognized as the primary cause of mortality worldwide, and pathology analysis plays a pivotal role in achieving accurate cancer diagnosis. The intricate representation of features in histopathological images encompasses abundant information crucial for disease diagnosis, regarding cell appearance, tumor microenvironment, and geometric characteristics. However, recent deep learning methods have not adequately exploited geometric features for pathological image classification due to the absence of effective descriptors that can capture both cell distribution and gathering patterns, which often serve as potent indicators. In this paper, inspired by clinical practice, a Hierarchical Graph Pyramid Transformer (HGPT) is proposed to guide pathological image classification by effectively exploiting a geometric representation of tissue distribution which was ignored by existing state-of-the-art methods. First, a graph representation is constructed according to morphological feature of input pathological image and learn geometric representation through the proposed multi-head graph aggregator. Then, the image and its graph representation are feed into the transformer encoder layer to model long-range dependency. Finally, a locality feature enhancement block is designed to enhance the 2D local representation of feature embedding, which is not well explored in the existing vision transformers. An extensive experimental study is conducted on Kather-5K, MHIST, NCT-CRC-HE, and GasHisSDB for binary or multi-category classification of multiple cancer types. Results demonstrated that our method is capable of consistently reaching superior classification outcomes for histopathological images, which provide an effective diagnostic tool for malignant tumors in clinical practice.

Velocity-aware spatial-temporal attention LSTM model for inverse dynamic model learning of manipulators.

Introduction: An accurate inverse dynamics model of manipulators can be effectively learned using neural networks. However, further research is required to investigate the impact of spatiotemporal variations in manipulator motion sequences on network learning. In this work, the Velocity Aware Spatial-Temporal Attention Residual LSTM neural network (VA-STA-ResLSTM) is proposed to learn a more accurate inverse dynamics model, which uses a velocity-aware spatial-temporal attention mechanism to extract dynamic spatiotemporal features selectively from the motion sequence of the serial manipulator. Methods: The multi-layer perception (MLP) attention mechanism is adopted to capture the correlation between joint position and velocity in the motion sequence, and the state correlation between hidden units in the LSTM network to reduce the weight of invalid features. A velocity-aware state fusion approach of LSTM network hidden units' states is proposed, which utilizes variation in joint velocity to adapt to the temporal characteristics of the manipulator dynamic motion, improving the generalization and accuracy of the neural network. Results: Comparative experiments have been conducted on two open datasets and a self-built dataset. Specifically, the proposed method achieved an average accuracy improvement of 61.88% and 43.93% on the two different open datasets and 71.13% on the self-built dataset compared to the LSTM network. These results demonstrate a significant advancement in accuracy for the proposed method. Discussion: Compared with the state-of-the-art inverse dynamics model learning methods of manipulators, the modeling accuracy of the proposed method in this paper is higher by an average of 10%. Finally, by visualizing attention weights to explain the training procedure, it was found that dynamic modeling only relies on partial features, which is meaningful for future optimization of inverse dynamic model learning methods.

Digitalization and innovation in green ports: A review of current issues, contributions and the way forward in promoting sustainable ports and maritime logistics.

As a fundamental transportation mode, maritime logistics has become an indispensable component on a global scale. However, there are multiple drawbacks associated with ports operating in traditional ways, such as higher cost, lower efficiency and generating more environmental pollution. Digital technologies have been researched and implemented gradually in green ports, especially in data collection and real-time monitoring, and these advances help to promote higher work efficiency and reduce detrimental environmental impacts. It was found that green ports (e.g. ports of Raffina, Los Angeles, and Long Beach) generally perform better in energy conservation and pollutant emission reduction. However, considering the variability in the level of digitalization, there are challenges in achieving effective communications between individual ports. Therefore, to optimize and update green port practices, a systematic review is necessary to comprehensively analyze the beneficial contributions of green ports. This review adopted bibliometric analysis to examine the shipping framework focusing on green ports digitalization and innovation. After that, with regards to the bibliometric results, five aspects were analyzed, including environment, performance, policy, technology, and management. Besides, intelligent life-cycle management was systematically discussed to improve green ports and maritime logistics performance and sustainability in three aspects, namely waste discharge, shipping management system and green ports management. The findings revealed that green ports and maritime logistics require digital cooperation, transformation, and management to achieve sustainable development goals, including route selection and control of ships' numbers, weather prediction, and navigational effluent monitoring, albeit with some obstacles.

[Comparison of Quality of Life of the Patients Three Months after Uniportal 
and Multiportal Thoracoscopic Lobectomy].

BACKGROUND: The relationship between quality of life at three months after lung cancer surgery and different surgical approaches is remains unclear. This study aimed to compare the quality of life of patients three months after uniportal and multiportal thoracoscopic lobectomy. METHODS: Data from patients who underwent lung surgery at the Department of Thoracic Surgery, Sichuan Cancer Hospital between April 2021 and October 2021 were collected. The European Organization for Research and Treatment of Quality of Life Questionnaire-Core 30 (EORTC QLQ-C30) and Quality of Life Questionnaire-Lung Cancer 29 (EORTC QLQ-LC29) were used to collect quality of life data of the patients. Potential confounding factors in the baseline data were included in a multivariate regression model for adjustment, and the quality of life of the two groups three months postoperatively was compared with traditional clinical outcomes. RESULTS: A total of 130 lung cancer patients were included, with 57 males (43.8%) and 73 females (56.2%), and an average age of (57.1±9.5) yr. In the baseline data of the two groups, there was a statistical difference in the number of chest drainage tubes placed (P<0.001). After adjustment with the regression model, at three months postoperatively, there were no significant differences in all symptoms and functional status scores between the two groups (all P>0.05). The multiportal group had longer surgery time (120.0 min vs 85.0 min, P=0.001), postoperative hospital stay (6.0 d vs 4.0 d, P=0.020), and a higher incidence of early ≥ grade 2 complications (39.0% vs 10.1%, P=0.011) compared to the uniportal group. CONCLUSIONS: Patients undergoing uniportal and multiportal thoracoscopic lobectomy have similar quality of life at three months postoperatively. The uniportal group may have certain advantages in terms of traditional clinical outcome indicators such as operation time, postoperative hospital stay, and early postoperative complications.

Proteomics-driven noninvasive screening of circulating serum protein panels for the early diagnosis of hepatocellular carcinoma.

Early diagnosis of hepatocellular carcinoma (HCC) lacks highly sensitive and specific protein biomarkers. Here, we describe a staged mass spectrometry (MS)-based discovery-verification-validation proteomics workflow to explore serum proteomic biomarkers for HCC early diagnosis in 1002 individuals. Machine learning model determined as P4 panel (HABP2, CD163, AFP and PIVKA-II) clearly distinguish HCC from liver cirrhosis (LC, AUC 0.979, sensitivity 0.925, specificity 0.915) and healthy individuals (HC, AUC 0.992, sensitivity 0.975, specificity 1.000) in an independent validation cohort, outperforming existing clinical prediction strategies. Furthermore, the P4 panel can accurately predict LC to HCC conversion (AUC 0.890, sensitivity 0.909, specificity 0.877) with predicting HCC at a median of 11.4 months prior to imaging in prospective external validation cohorts (No.: Keshen 2018_005_02 and NCT03588442). These results suggest that proteomics-driven serum biomarker discovery provides a valuable reference for the liquid biopsy, and has great potential to improve early diagnosis of HCC.