Establishing a Differential Diagnosis Method for Melanoma Using Blood Parameters Based on Machine Learning Technology

Establishing a Differential Diagnosis Method for Melanoma Using Blood Parameters Based on Machine Learning Technology | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Establishing a Differential Diagnosis Method for Melanoma Using Blood Parameters Based on Machine Learning Technology Qijiao Wang, Xia Liu, Ruihuan Zhang, Qi Zhang, Xinyao Zhang, Lei Li, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8003880/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objective: This study aims to develop and validate a differential diagnosis model for melanoma. This model will combine blood test results with machine learning techniques to provide an inexpensive, non-invasive method for early melanoma screening. Methods: Complete blood counts and biochemical indicators were retrospectively collectedfrom 4,534 patients for clinical data. Key variables were screened using the Recursive Feature Elimination (RFE) and Mutual Information (MI) methods. A predictive model was established using the XGBoost algorithm, and the model's performance was evaluated through 10-fold cross-validation and an independent test set. Results: The model achieved an area under the curve (AUC) of 0.996 on the training set and 0.975 on the test set, with respective accuracies of 0.981 and 0.926. MPV, DBIL, CK, Na, HDL-C, and LDH were identified as the most predictive indicators. Conclusion: This study confirms that conventional blood parameters combined with machinelearning technology can effectively distinguish melanoma, demonstrating high diagnostic value. This approach provides a feasible, noninvasive strategy for earlymelanoma identification. Melanoma Machine Learning Blood Biomarkers Differential Diagnosis Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction As the largest organ in the human body, skin cancer primarily encompasses two major categories: melanoma and non-melanoma skin cancers. Although melanoma accounts for only about 4% of skin cancer cases, it causes nearly 80% of skin cancer-related deaths, making it one of the most lethal types [1] . Globally, melanoma incidence and mortality rates are rising annually, making it a significant public health concern. In China, it ranks among the fastest-growing malignant tumors [2] . Common pathological subtypes include superficial spreading melanoma, nodular melanoma, malignant lentigo maligna melanoma, and acral lentiginous melanoma. Additional rare subtypes exist, such as epithelioid melanoma and fibroblastic melanoma [3] . The distribution of subtypes varies across ethnic groups: superficial spreading melanoma predominates in Caucasians, while acral lentiginous melanoma is more common in Asians and Africans [4] . Although Caucasians exhibit higher melanoma incidence, people of color demonstrate poorer melanoma-specific survival rates [5] . Exposure to ultraviolet radiation is the recognized primary risk factor for cutaneous melanoma [6, 7] . Early diagnosis is crucial for melanoma treatment, as localized surgery in early-stage patients reduces trauma, lowers medical costs, and significantly improves survival rates [8,9] . Melanomas confined to the epidermis carry an extremely low risk of death, and thin melanomas have a lower probability of metastasis. In contrast, advanced-stage patients are prone to recurrence and metastasis, resulting in high mortality rates [10-12] . The current gold standard for melanoma diagnosis remains invasive tissue biopsy, though this approach may lead to unnecessary surgeries [13] . Pathologists exhibit considerable subjective variation in histopathological interpretation [14] , suggesting potential for overdiagnosis and underdiagnosis [15] . Studies report misdiagnosis as a contributing factor to mortality in many melanoma patients [16, 17] . Dermoscopy, a commonly used adjunct tool, shows diagnostic efficacy significantly influenced by physician experience and lesion characteristics [18,19] . However, early-stage melanoma may lack specific dermoscopic features, making diagnosis challenging even with dermoscopy assistance [20, 21] . Blood tests and fluid biopsies, as important non-invasive methods for cancer diagnosis, have been widely applied in tumor diagnosis and treatment monitoring [22] . By detecting circulating tumor DNA [23] , protein biomarkers [24] , enzymes [25,26] , metabolites [27] , and antigens/antibodies [28,29] in plasma, they offer novel approaches for non-invasive tumor diagnosis and dynamic monitoring. Inflammatory markers and biochemical parameters within blood test results are also closely associated with the occurrence, staging, progression, survival, and prognosis of various cancers [30-33] . Furthermore, analyzing multiple blood parameters can assist in identifying tumor patients suitable for surgical resection. Dynamic monitoring of tumor-associated biomarkers in blood further enhances the accuracy of early diagnosis [34]. With the deepening application of machine learning technologies in the medical field [35] , they can uncover underlying patterns from complex data [36] . This supports disease prediction [37] , diagnosis [38] , prognosis assessment [39] , and risk evaluation [40] . Current research on the association between blood parameters and melanoma remains limited. This study aims to combine machine learning methods with the analysis of blood parameters in melanoma patients to explore their value in auxiliary diagnosis , providing new evidence for early screening and diagnosis. Methods and Materials Data Collection and Clinical Definitions Collect clinical data from 4,534 patients at Peking University Cancer Hospital between March 2015 and November 2022 for retrospective study. Patient demographic information extracted from electronic medical records included gender, age, date of visit, patient ID number, clinical diagnosis, department of visit, and laboratory test results for complete blood count and biochemistry (Table 1).The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Peking University Cancer Hospital (Approval No. KY202402). Written informed consent was obtained from all participants. Table 1 All parameter abbreviations and full names Abbreviation Full name WBC White Blood Cell Count LYMPH% Lymphocyte Ratio MONO% Monocyte Ratio NEUT% Neutrophil Ratio EOS% Eosinophil Ratio BASO% Basophil Ratio LYMPTH Lymphocyte Count MONO# Monocyte Count NEUT# Neutrophil Count EOS# Eosinophil Count BASO# Basophil Count MCV Mean Corpuscular Volume MCH Mean Corpuscular Hemoglobin MCHC Mean Corpuscular Hemoglobin Concentration PLT Platelet MPV Mean Platelet Volume PCT Platelet Crit K Potassium Na Sodium Cl Chlorine Ca Calcium P Inorganic Phosphorus GLU Glucose UREA Urea Cr Creatinine (Enzymatic Method UA Uric Acid TP Total Protein TBIL Total Bilirubin DBIL Direct Bilirubin GGT Y-Glutamyl Transferase CK Creatine Kinase LDH Lactate Dehydrogenase TG Triglyceride LDL-C Low-Density Lipoprotein Cholesterol HDL-C High-Density Lipoprotein Cholesterol TBA Total Bile Acid Inclusion and Exclusion Criteria for Melanoma Inclusion Criteria : 1. Patients diagnosed with melanoma who underwent surgical treatment at Peking University Cancer Hospital and possess complete pathological diagnosis results. 2. Patients who underwent complete blood count and biochemical testing at Peking University Cancer Hospital and have multiple pathological results; the most recent test data prior to surgery will be selected. Exclusion Criteria: 1. Patients with concomitant malignant diseases. 2. Patients lacking complete blood count and biochemical test parameters from Peking University Cancer Hospital. 3. Patients with clinical records indicating prior chemotherapy before surgery. Training and Validation of Predictive Models Data Partitioning Prior to data partitioning, we performed preprocessing on the dataset, including missing value handling, outlier detection, and removal of unknown data, to ensure data quality and enhance model training stability and effectiveness. The preprocessed dataset retained 4,534 patient records, which were randomly divided into training and testing sets at an 8:2 ratio. The training set was used for model construction and parameter optimization, while the testing set evaluated the model's generalization capability and performance, ensuring its stability in practical applications. Feature Selection Feature selection is the process of filtering key variables from the original feature set, aiming to reduce redundant information and noise interference while enhancing model training efficiency and prediction accuracy. The Recursive Feature Elimination (RFE) method and Mutual Information (MI) method are employed to determine the optimal variables for feature selection. RFE is an iterative model-based feature selection method that trains models multiple times and progressively removes less important features, ultimately retaining the optimal subset. Specifically, it first trains a random forest model using all features to output feature importance scores. It then sequentially removes the least important features, repeating the training and removal process with the remaining features until predetermined feature count or performance stopping criteria are met. The Mutual Information method, grounded in information theory, screens features by measuring the strength of dependency between features and the target variable, unrestricted by linear relationships. Its core involves calculating the mutual information value between feature X and target Y. This value indicates the reduction in uncertainty of Y given X; a higher value signifies a stronger association between them. During screening, features with higher mutual information values are typically retained. It is applicable to various data distributions and relationship types but cannot reflect the directionality of the association. Building the Predictive Model We employed the eXtreme Gradient Boosting (XGBoost) algorithm to construct the predictive model. As an efficient ensemble learning algorithm based on the gradient boosting framework, XGBoost achieves high-precision prediction by iteratively building multiple decision trees and combining their predictions through weighted aggregation. Its core mechanism employs a gradient descent strategy, generating a new decision tree in each iteration to fit the prediction residuals of the preceding model. By continuously stacking tree models, it optimizes overall performance. Simultaneously, it incorporates a regularization term to control tree complexity and prevent overfitting. Additionally, it supports parallel computing to accelerate training and possesses adaptive handling capabilities for missing values. Model Evaluation and Test Set To validate the model's generalization capability, we employed 10-fold cross-validation for model validation. The metrics for evaluating the prediction model include Accuracy, Recall, Specificity, Area Under the Receiver Operating Characteristic Curve (AUC), and F1-score. These metrics are calculated based on the True Negative (TN), True Positive (TP), False Negative (FN), and False Positive (FP) values from the confusion matrix, using the following formulas: Among these, the F1-score is a synthetic evaluation metric that measures the harmonic mean of a classification model's precision and recall. Precision is the ratio of correctly identified positives to all predicted positives: Precision = TP / (TP + FP). AUC is a crucial metric for evaluating classification model performance, quantifying the model's classification capability through the area under the ROC curve. It offers advantages such as threshold independence and insensitivity to class distribution, making it particularly suitable for imbalanced datasets. AUC ranges from 0 to 1, with values closer to 1 indicating superior model performance. Model Interpretation To enhance understanding of the model's operational principles and prediction outcomes, we employ the SHAP method to elucidate factors influencing model predictions. SHAP is a game-theoretic model interpretation technique that quantifies feature impact on predictions by calculating each feature's contribution value to the model output (i.e., SHAP values). The core concept of SHAP values originates from Shapley values, ensuring fair and consistent contribution allocation. Compared to traditional feature importance methods, SHAP not only explains global feature importance but also provides local explanations for individual sample predictions, helping users understand the model's decision-making basis in specific scenarios. Through visualization tools, users can intuitively analyze the relationship between features and model predictions, thereby enhancing the model's credibility and transparency. Statistical Analysis Model development and validation were conducted using Python 3.8.3 (library, sci-kit-learn). Descriptive analysis employed SPSS 22.0 (IBM, Chicago, IL, USA). Normally distributed continuous variables are presented as mean ± standard deviation (mean ± SD), while non-normally distributed data are shown as median (interquartile range [IQR]). The Kolmogorov-Smirnov test was used to assess data distribution patterns. Differences in group characteristics were analyzed using t-tests or Mann-Whitney U tests as appropriate, with P < 0.05 considered statistically significant. Results Original Variables A total of 2,000 melanoma cases were finalized as the positive experimental group, and 2,534 healthy screening participants were randomly selected as the negative control group（Table 2）. The mean age of melanoma patients was significantly higher than that of healthy examinees (56 years vs. 42 years, P<0.001), indicating age may be an important risk factor for melanoma. Additionally, no significant difference in gender distribution was observed between groups (male proportion: melanoma 49.5%, healthy examinees 49.1%). Among the 36 blood routine and biochemical indicators analyzed, four parameters—K, Ca, TBA, and MCH—showed no statistically significant differences between groups (P > 0.05). The remaining 35 indicators exhibited significant differences (P < 0.05). Inflammation-related markers showed significant differences between groups: melanoma patients exhibited markedly elevated WBC, NEUT%, and NEUT#, while LYMPH% and LYMPH# were significantly reduced. MPV and PCT were markedly decreased in melanoma patients, with no significant difference in PLT. Melanoma patients exhibited significantly elevated LDH, GLU, TG, and LDL-C levels, while HDL-C levels were significantly reduced. DBIL and GG were significantly elevated in melanoma patients. These differences suggest that melanoma patients may experience systemic inflammatory responses, metabolic disorders, and immune dysfunction, providing potential feature variables for subsequent model construction. Table 2 Comparison of Blood Parameters, Gender, and Age Between Melanoma Patients and Healthy Check-up Participants Label Group Number Average(Standard)/ Median(25 percent, 75 percent ） P Age Health 2534 42（35, 51） 0 Melanoma 2000 56（46, 64） K Health 2534 4.202（0.294） 0.214 Melanoma 2000 4.214（0.356） Na Health 2534 139.296（1.810） 0 Melanoma 2000 140.543（2.467） Cl Health 2534 103.701（1.949） 0 Melanoma 2000 103.339（2.949） Ca Health 2534 2.3753（0.086） 0.188 Melanoma 2000 2.3797（0.130） P Health 2534 1.1546（0.158） 0 Melanoma 2000 1.198（0.177） GLU Health 2534 5.4（5, 5.8） 0 Melanoma 2000 5.485（5.08, 6.13） UREA Health 2534 4.7（4, 5.5） 0 Melanoma 2000 4.87（4.04, 5.780） Cr Health 2534 67.2（58.7, 78.9） 0 Melanoma 2000 60（51, 71） UA Health 2534 318（265, 387） 0.001 Melanoma 2000 312（259, 373） TP Health 2534 73.8（71.2, 76.5） 0 Melanoma 2000 72.4（69.2, 75.4） TBIL Health 2534 13.3（10.7, 16.9） 0.002 Melanoma 2000 12.7（10.1, 16.1） DBIL Health 2534 2.4（2.582, 2.661） 0 Melanoma 2000 3.4（2.8, 4.4） GGT Health 2534 20.5（15, 33） 0 Melanoma 2000 25（17, 41） CK Health 2534 91（70, 131） 0 Melanoma 2000 62（45, 84） LDH Health 2534 168（151, 187） 0 Melanoma 2000 192.5（168, 187） TG Health 2534 1.12（0.76, 1.73） 0 Melanoma 2000 1.43（1.033, 2.06） LDL-C Health 2534 3.049（0.819） 0.001 Melanoma 2000 3.137（0.861） HDL-C Health 2534 1.484（O.415） 0 Melanoma 2000 1.209（0.317） TBA Health 2534 2.6（1.8, 3.9） 0.095 Melanoma 2000 2.6（1.5, 4.3） WBC Health 2534 5.885（5, 7） 0.001 Melanoma 2000 5.99（4.92, 7.18） LYMPH% Health 2534 33.578（7.254） 0 Melanoma 2000 26.743（9.304） MONO% Health 2534 6（5.1, 7.2） 0 Melanoma 2000 5.6（4.6, 6.7） NEUT% Health 2534 57.342（7.877） 0 Melanoma 2000 65.086（9.982） EOS% Health 2534 1.8（1.1, 2.8） 0 Melanoma 2000 1.4（0.7, 2.4） BASO% Health 2534 0.6（0.4, 0.8） 0 Melanoma 2000 0.4（0.3, 0.8） LYMPH# Health 2534 1.93（1.61, 2.33） 0 Melanoma 2000 1.52（1.21, 1.9) MONO# Health 2534 0.36（0.29, 0.44） 0 Melanoma 2000 0.33（0.26, 0.43） NEUT# Health 2534 3.37（2.71, 4.173） 0 Melanoma 2000 3.85（2.97, 4.908） EOS# Health 2534 0.1（0.06, 0.17） 0 Melanoma 2000 0.08（0.04, 0.14） BASO# Health 2534 0.03（0.02, 0.05） 0 Melanoma 2000 0.02（0.02, 0.03） MCV Health 2534 91.6（88.9, 94.1） 0 Melanoma 2000 91.1（88, 93.8） MCH Health 2534 30.2（29.2, 31.1） 0.24 Melanoma 2000 30.066（2.183） MCHC Health 2534 328.45（11.917） 0 Melanoma 2000 331.65（10.405） PLT Health 2534 250（215, 288） 0 Melanoma 2000 230（191, 276） MPV Health 2534 10.256（0.899） 0 Melanoma 2000 9.311（1.134） PCT Health 2534 0.2592（0.056） 0 Melanoma 2000 0.2205（0.063） Note: Following normality testing, normally distributed continuous variables are presented as mean ± standard deviation (SD), while non-normally distributed data are represented by median (interquartile range [IQR]); abbreviations are listed in Table 1. Feature Selection The study employed both RFE and MI methods for feature screening. First, RFE identified 27 variables from the original dataset (Figure. 1 A). followed by MI screening yielding 19 variables (Figure. 1 B). To obtain a more representative and robust feature set, the results from both methods were intersected, ultimately yielding 15 common variables (Figure. 1 C) for predictive model construction. These variables included Age, BASO#, LYMPH#, PCT, HDL-C, BASO%, LDH, LYMPH%, DBIL, NEUT%, CK, MPV, Na, Cr, and GLU. Although the XGBoost model itself is insensitive to multicollinearity, to avoid errors in model interpretation caused by highly linearly correlated features, the study further analyzed variable correlations by calculating correlation coefficients and plotting a heatmap. Results showed only two pairs of variables—LYMPH% and NEUT%, and BASO% and BASO#—exhibited significant linear relationships, while linear correlations among other variables were weak (Figure. 1D). Given the limited number of linearly correlated variables and their negligible impact on machine learning prediction performance, no further multicollinearity handling was performed in this study. Predictive Model Performance and Validation The training set comprises 3,627 samples, including 1,600 positive samples and 2,027 negative samples. The test set comprises 907 samples, including 400 positive samples and 507 negative samples. When constructing the XGBoost prediction model using the training set, the optimal hyperparameter combination was determined through 5-fold cross-validation using the GridSearchCV method within the defined parameter grid. The parameter grid included: n_estimators ranging from [80, 100, 120, 150], max_depth from [8, 10, 12, 14], and learning_rate from [0.05, 0.1, 0.2]. The validated optimal configuration was: n_estimators = 120, max_depth = 12, learning_rate = 0.1. Additionally, to address the imbalanced distribution of positive and negative samples in the training set, we adjusted the weights of different class samples via the class_weight parameter to balance category disparities. Simultaneously, regularization was applied to prevent certain features from being assigned excessive weights, thereby avoiding model over-reliance on noise or peculiar patterns within local training data. We comprehensively analyzed the model's performance on the test set using multiple evaluation metrics, ROC curves, and calibration plots. Table 2 details the model performance evaluation results based on 10-fold cross-validation for both the training and test sets, including accuracy, sensitivity, specificity, AUC, and F1-Score metrics, providing quantitative measures of model performance（Table 3）. The ROC curve visually depicts the model's classification performance at different thresholds. Both the training and test sets achieved AUC values exceeding 0.95, indicating excellent model performance (Figure. 2 A). Calibration plots assess the alignment between predicted and actual labels. Overall, minimal discrepancies were observed, further validating the model's reliability (Figure. 2 B). Table 3 Model Performance Table Dataset Accuracy rate Sensitivity specificity AUC F1-Score Training set 0.981 0.957 0.999 0.996 0.978 Test set 0.926 0.920 0.931 0.975 0.917 Assessment of Factors Influencing Malignant Melanoma To thoroughly analyze the contribution of each feature to the predictive model, this study employs the SHAP method to interpret the XGBoost model. SHAP values quantify the model's predictive logic, clearly illustrating the specific impact of each feature on the prediction outcome. Different colors represent the direction of a feature's contribution to the prediction: red indicates a positive contribution, while blue signifies a negative contribution. Specifically, a positive SHAP value indicates that the feature increases the predicted risk of malignant melanoma, while a negative value reflects its role in reducing the predicted risk. For the XGBoost model, Figure 3 displays the global feature importance. Among the 15 variables, the top six key predictive indicators are MPV, DBIL, CK, Na, HDL-C, and LDH (Figure 3). Among these, DBIL, Na, and LDH exhibit significant positive correlations with disease risk, meaning elevated levels of these indicators synergistically increase an individual's risk of malignant melanoma. Conversely, MPV, CK, and HDL-C demonstrate pronounced negative regulatory effects, with their increased values strongly correlated with reduced disease risk. In local explanations, SHAP waterfall plots visually demonstrate the magnitude and direction of each feature's impact on the model's final prediction. The baseline value (E[f(x)]) represents the average SHAP value across all sample predictions, serving as the model's starting reference point. Features increasing the prediction value relative to this baseline are highlighted in red, while those decreasing the prediction value are shown in blue. The final f(x) value represents the model output after integrating all feature contributions, constituting the final prediction for that sample. Figure 4 displays SHAP single-sample explanations for two samples, where the six features with smaller contributions are grouped under “Other features,” with their combined contribution presented as a single value. In Figure 4 A, MPV, LDH, HDL-C, LYMPH%, CK, DBIL, Age, and NEUT% collectively drive the predicted value upward, indicating a positive influence of these factors on the sample. Conversely, in Figure 4 B, Age, HDL-C, MPV, NEUT%, LDH, PCT, and BASO% collectively lower the predicted value, suggesting a negative impact of these features on the sample's outcome. Discussion This study successfully developed a machine learning model based on conventional blood parameters for the differential diagnosis of melanoma, demonstrating exceptional predictive performance (test set AUC = 0.975). This significantly outperforms the independent predictive capability of traditional single indicators for melanoma, which typically exhibit AUC values below 0.8 [41] . Furthermore, the XGBoost algorithm employed in this study excels at handling high-dimensional, nonlinear medical data. Its ensemble learning properties effectively prevent overfitting and enhance the model's generalization capability. Combined with interpretability analysis using SHAP values, we can not only evaluate overall feature importance but also provide local explanations for individual patient predictions. This significantly enhances the model's transparency and credibility in clinical decision-making [42] . From a mechanistic perspective, the key indicators identified in this study are closely associated with the pathophysiological processes of melanoma. Among these, MPV (Mean Platelet Volume) is a critical parameter reflecting the average size of platelets in the blood, indicating their production rate and stimulation level, and providing important insights into numerous pathological conditions [43] . Elevated MPV levels were observed in patients with gastric, breast, endometrial, thyroid, and lung cancers. In esophageal, ovarian, and colorectal cancers, the trend in MPV changes remains unclear. Conversely, reduced MPV was noted in renal cell and gallbladder cancers [44] . This decrease may relate to abnormal platelet consumption within the tumor microenvironment [45] . Conversely, certain hematologic malignancies stimulate excessive megakaryocyte production of new platelets, leading to elevated MPV [46] . HDL-C transports excess cholesterol from peripheral tissues back to the liver for metabolism and clearance via the reverse transport mechanism, thereby effectively counteracting atherosclerosis. Furthermore, HDL-C possesses multiple protective functions including anti-inflammatory, antioxidant, vascular endothelial protection, and anti-thrombotic effects. Dysfunction of HDL-C may increase cancer risk. Multiple large-scale retrospective studies indicate that a decline in HDL-C levels from the normal range significantly elevates the risk of gastric cancer, colorectal cancer, liver cancer, pancreatic cancer, gallbladder cancer, and biliary tract cancer [47,48] . This finding aligns with the prevalent low HDL-C levels observed in melanoma patients in this study, further supporting the association between low HDL-C and the development of various malignancies. Concurrently, this study found that sodium ions (Na⁺) and lactate dehydrogenase (LDH) levels were elevated in melanoma patients compared to healthy controls. Elevated serum sodium is not uncommon in cancer patients, as abnormal expression of sodium channels in tumor tissues increases intracellular sodium concentrations, thereby promoting invasion and metastasis. Studies indicate that sodium concentrations in breast cancer and its metastatic lesions are significantly higher than in adjacent normal tissue. Elevated sodium levels exert pro-inflammatory effects, suppress immune cell function, and promote resistance to chemotherapy [49] . LDH, a key glycolytic enzyme, exhibits elevated serum levels that not only indicate tumor burden and hypoxic necrosis but also promote metastasis through a lactate-mediated immunosuppressive microenvironment [44] . It is closely associated with tumor cell metabolic reprogramming, melanoma invasiveness, and poor prognosis [50] . This study also found significant differences in LDH levels between melanoma patient blood samples and negative samples (P<.001), consistent with previous research. This study has several limitations. First, all data originated from a single center, potentially introducing selection bias. Second, the retrospective study design cannot entirely eliminate confounding factors. Finally, the model has not been validated in prospective cohorts or external multicenter data, and its generalizability requires further confirmation. Conclusion The results above indicate that the selected blood markers are associated with melanoma to some extent. However, these markers do not demonstrate clear diagnostic value when considered individually. We can effectively distinguish specific diseases by harnessing the power of machine learning to integrate multiple test indicators and analyze the complex interrelationships among their patterns. Declarations Author Contributions: Conceptualization, Qijiao Wang, Xia Liu, andRuihuan Zhang; methodology, Qijiao Wang and Lei Li; software, Qi Zhang and Xinyao Zhang; validation, Zhijuan Guo, Shenyi Lian, and Ruihuan Zhang; project administration, Zhongwu Li; funding acquisition, Zhongwu Li. All authors have read and agreed to the published version of the manuscript. Funding: This work was supported by Capital’s Funds for Health Improvement and Research 2022–2-1024,the Natural Science Foundation of Inner Mongolia Autonomous Region of China (Grants No. 2024ZD31), Science and Technology Program of the Joint Fund of Science Research for the Public Hospital of Inner Mongolian Academy of Medical Sciences(2023GLLH0125). Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Peking University Cancer Hospital (Approval No. KY202402 and date of approval:2024.01). Informed Consent Statement: Written informed consent was obtained from all subjects involved in the study. Data Availability Statement: The data presented in this study are available on request from the corresponding author due to (e.g., privacy and ethical concerns). Conflicts of Interest: The authors declare no conflicts of interest. Acknowledgments: We sincerely thank our mentor, Professor Zhongwu Li, for his expert guidance. Our gratitude also goes to all teachers and colleagues who provided support and facilitation for this study References ZAREMBA A, ZIMMER L, GRIEWANK K G, et al. [Immunotherapy for malignant melanoma] [J]. Internist (Berl), 2020, 61(7): 669-675. SUNG H, FERLAY J, SIEGEL R L, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries [J]. CA Cancer J Clin, 2021, 71(3): 209-249. PUCKETT Y, WILSON A M, FARCI F, et al. Melanoma Pathology [M]. StatPearls. Treasure Island (FL); StatPearls PublishingCopyright © 2025, StatPearls Publishing LLC. 2025. MADANKUMAR R, GUMASTE P V, MARTIRES K, et al. Acral melanocytic lesions in the United States: Prevalence, awareness, and dermoscopic patterns in skin-of-color and non-Hispanic white patients [J]. J Am Acad Dermatol, 2016, 74(4): 724-730.e721. QIAN Y, JOHANNET P, SAWYERS A, et al. The ongoing racial disparities in melanoma: An analysis of the Surveillance, Epidemiology, and End Results database (1975-2016) [J]. J Am Acad Dermatol, 2021, 84(6): 1585-1593. DEWANE M E, KELSEY A, OLIVIERO M, et al. Melanoma on chronically sun-damaged skin: Lentigo maligna and desmoplastic melanoma [J]. J Am Acad Dermatol, 2019, 81(3): 823-833. MADDEN K, FORMAN S B, ELSTON D. Quantification of melanocytes in sun-damaged skin [J]. J Am Acad Dermatol, 2011, 64(3): 548-552. GARBE C, AMARAL T, PERIS K, et al. European consensus-based interdisciplinary guideline for melanoma. Part 1: Diagnostics - Update 2024 [J]. Eur J Cancer, 2025, 215: 115152. PETRIE T, SAMATHAM R, WITKOWSKI A M, et al. Melanoma Early Detection: Big Data, Bigger Picture [J]. J Invest Dermatol, 2019, 139(1): 25-30. NAIK P P. Cutaneous Malignant Melanoma: A Review of Early Diagnosis and Management [J]. World J Oncol, 2021, 12(1): 7-19. HASHIM P W, FRIEDLANDER P, GOLDENBERG G. Systemic Therapies for Late-stage Melanoma [J]. J Clin Aesthet Dermatol, 2016, 9(10): 36-40. GERSHENWALD J E, SCOLYER R A, HESS K R, et al. Melanoma staging: Evidence-based changes in the American Joint Committee on Cancer eighth edition cancer staging manual [J]. CA Cancer J Clin, 2017, 67(6): 472-492. FEIT N E, DUSZA S W, MARGHOOB A A. Melanomas detected with the aid of total cutaneous photography [J]. Br J Dermatol, 2004, 150(4): 706-714. ELDER D E, PIEPKORN M W, BARNHILL R L, et al. Pathologist characteristics associated with accuracy and reproducibility of melanocytic skin lesion interpretation [J]. J Am Acad Dermatol, 2018, 79(1): 52-59.e55. Pathologists' diagnosis of invasive melanoma and melanocytic proliferations: observer accuracy and reproducibility study [J]. Bmj, 2017, 358: j3798. BATAILLE V, SASIENI P, CURLEY R K, et al. Melanoma yield, number of biopsies and missed melanomas in a British teaching hospital pigmented lesion clinic: a 9-year retrospective study [J]. Br J Dermatol, 1999, 140(2): 243-248. CARLI P, NARDINI P, CROCETTI E, et al. Frequency and characteristics of melanomas missed at a pigmented lesion clinic: a registry-based study [J]. Melanoma Res, 2004, 14(5): 403-407. LINDSAY D, SOYER H P, JANDA M, et al. Cost-Effectiveness Analysis of 3D Total-Body Photography for People at High Risk of Melanoma [J]. JAMA Dermatol, 2025, 161(5): 482-489. FINK C, HAENSSLE H A. Non-invasive tools for the diagnosis of cutaneous melanoma [J]. Skin Res Technol, 2017, 23(3): 261-271. GERAMI P, BUSAM K, COCHRAN A, et al. Histomorphologic assessment and interobserver diagnostic reproducibility of atypical spitzoid melanocytic neoplasms with long-term follow-up [J]. Am J Surg Pathol, 2014, 38(7): 934-940. MARINO M L, CARRERA C, MARCHETTI M A, et al. Practice Gaps in Dermatology: Melanocytic Lesions and Melanoma [J]. Dermatol Clin, 2016, 34(3): 353-362. PHALLEN J, SAUSEN M, ADLEFF V, et al. Direct detection of early-stage cancers using circulating tumor DNA [J]. Sci Transl Med, 2017, 9(403). BETTEGOWDA C, SAUSEN M, LEARY R J, et al. Detection of circulating tumor DNA in early- and late-stage human malignancies [J]. Sci Transl Med, 2014, 6(224): 224ra224. COHEN J D, JAVED A A, THOBURN C, et al. Combined circulating tumor DNA and protein biomarker-based liquid biopsy for the earlier detection of pancreatic cancers [J]. Proc Natl Acad Sci U S A, 2017, 114(38): 10202-10207. HU X, CHEN R, WEI Q, et al. The Landscape Of Alpha Fetoprotein In Hepatocellular Carcinoma: Where Are We? [J]. Int J Biol Sci, 2022, 18(2): 536-551. HAMMARSTRöM S. The carcinoembryonic antigen (CEA) family: structures, suggested functions and expression in normal and malignant tissues [J]. Semin Cancer Biol, 1999, 9(2): 67-81. JOVé M, COLLADO R, QUILES J L, et al. A plasma metabolomic signature discloses human breast cancer [J]. Oncotarget, 2017, 8(12): 19522-19533. LACOMBE J, MANGé A, JARLIER M, et al. Identification and validation of new autoantibodies for the diagnosis of DCIS and node negative early-stage breast cancers [J]. Int J Cancer, 2013, 132(5): 1105-1113. DESMETZ C, BASCOUL-MOLLEVI C, ROCHAIX P, et al. Identification of a new panel of serum autoantibodies associated with the presence of in situ carcinoma of the breast in younger women [J]. Clin Cancer Res, 2009, 15(14): 4733-4741. FEST J, RUITER R, MULDER M, et al. The systemic immune-inflammation index is associated with an increased risk of incident cancer-A population-based cohort study [J]. Int J Cancer, 2020, 146(3): 692-698. NøST T H, ALCALA K, URBAROVA I, et al. Systemic inflammation markers and cancer incidence in the UK Biobank [J]. Eur J Epidemiol, 2021, 36(8): 841-848. PODNAR S, KUKAR M, GUNČAR G, et al. Diagnosing brain tumours by routine blood tests using machine learning [J]. Sci Rep, 2019, 9(1): 14481. WONG J Y Y, BASSIG B A, LOFTFIELD E, et al. White Blood Cell Count and Risk of Incident Lung Cancer in the UK Biobank [J]. JNCI Cancer Spectr, 2020, 4(2): pkz102. COHEN J D, LI L, WANG Y, et al. Detection and localization of surgically resectable cancers with a multi-analyte blood test [J]. Science, 2018, 359(6378): 926-930. SWANSON K, WU E, ZHANG A, et al. From patterns to patients: Advances in clinical machine learning for cancer diagnosis, prognosis, and treatment [J]. Cell, 2023, 186(8): 1772-1791. GREENER J G, KANDATHIL S M, MOFFAT L, et al. A guide to machine learning for biologists [J]. Nat Rev Mol Cell Biol, 2022, 23(1): 40-55. QI X, WANG S, FANG C, et al. Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants [J]. Redox Biol, 2025, 79: 103470. WU J, ZAN X, GAO L, et al. A Machine Learning Method for Identifying Lung Cancer Based on Routine Blood Indices: Qualitative Feasibility Study [J]. JMIR Med Inform, 2019, 7(3): e13476. YUAN Q, YE R, QIAN Y, et al. Machine learning and SHAP value interpretation for predicting the response to neoadjuvant chemotherapy and long-term clinical outcomes in Chinese female breast cancer [J]. Ann Med, 2025, 57(1): 2541316. LEI J, ZHAI J, ZHANG Y, et al. Supervised Machine Learning Models for Predicting Sepsis-Associated Liver Injury in Patients With Sepsis: Development and Validation Study Based on a Multicenter Cohort Study [J]. J Med Internet Res, 2025, 27: e66733. ZHANG J, YAO Y H, LI B G, et al. Prognostic value of pretreatment serum lactate dehydrogenase level in patients with solid tumors: a systematic review and meta-analysis [J]. Sci Rep, 2015, 5: 9800. LIANG D, WANG L, ZHONG P, et al. Perspective: Global Burden of Iodine Deficiency: Insights and Projections to 2050 Using XGBoost and SHAP [J]. Adv Nutr, 2025, 16(3): 100384. KORNILUK A, KOPER-LENKIEWICZ O M, KAMIŃSKA J, et al. Mean Platelet Volume (MPV): New Perspectives for an Old Marker in the Course and Prognosis of Inflammatory Conditions [J]. Mediators Inflamm, 2019, 2019: 9213074. DETOPOULOU P, PANOUTSOPOULOS G I, MANTOGLOU M, et al. Relation of Mean Platelet Volume (MPV) with Cancer: A Systematic Review with a Focus on Disease Outcome on Twelve Types of Cancer [J]. Curr Oncol, 2023, 30(3): 3391-3420. SUN Y, LI T, DING L, et al. Platelet-mediated circulating tumor cell evasion from natural killer cell killing through immune checkpoint CD155-TIGIT [J]. Hepatology, 2025, 81(3): 791-807. HE F, LARANJEIRA A B, KONG T, et al. Multiomic profiling reveals metabolic alterations mediating aberrant platelet activity and inflammation in myeloproliferative neoplasms [J]. J Clin Invest, 2024, 134(3). NAM S Y, JO J, CHO C M. A population-based cohort study of longitudinal change of high-density lipoprotein cholesterol impact on gastrointestinal cancer risk [J]. Nat Commun, 2024, 15(1): 2923. NAM S Y, JO J, JEON S W. Discrepant effect of high-density lipoprotein cholesterol on esophageal and gastric cancer risk in a nationwide cohort [J]. Gastric Cancer, 2024, 27(3): 451-460. JAMES A D, LESLIE T K, KAGGIE J D, et al. Sodium accumulation in breast cancer predicts malignancy and treatment response [J]. Br J Cancer, 2022, 127(2): 337-349. CLAPS G, FAOUZI S, QUIDVILLE V, et al. The multiple roles of LDH in cancer [J]. Nat Rev Clin Oncol, 2022, 19(12): 749-762. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8003880","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":549888483,"identity":"d9906191-c312-4df5-96ba-d9581eb4613f","order_by":0,"name":"Qijiao Wang","email":"","orcid":"","institution":"Inner Mongolia Medical University, China；Peking Universi ty Cancer Hospital (Inner Mongolia Campus) \u0026 Affiliated Canc er Hospital of Inner Mongolia Medical University, Inner Mongolia Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Qijiao","middleName":"","lastName":"Wang","suffix":""},{"id":549888485,"identity":"189bfb8b-2678-45ea-b2ca-77050b695bb4","order_by":1,"name":"Xia Liu","email":"","orcid":"","institution":"Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Department of Pathology, Peking University Cancer Hospital \u0026Institute","correspondingAuthor":false,"prefix":"","firstName":"Xia","middleName":"","lastName":"Liu","suffix":""},{"id":549888490,"identity":"9620b8fc-a785-4fe8-ad15-8e4dc01ac1aa","order_by":2,"name":"Ruihuan Zhang","email":"","orcid":"","institution":"The Inner Mongolia Medical Intelligent Diagnostics Big Data Research Institute, Inner Mongolia","correspondingAuthor":false,"prefix":"","firstName":"Ruihuan","middleName":"","lastName":"Zhang","suffix":""},{"id":549888492,"identity":"2f023c99-770a-4b9b-bb6f-dde9b966432f","order_by":3,"name":"Qi Zhang","email":"","orcid":"","institution":"Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Department of Pathology, Peking University Cancer Hospital \u0026Institute","correspondingAuthor":false,"prefix":"","firstName":"Qi","middleName":"","lastName":"Zhang","suffix":""},{"id":549888493,"identity":"62816a13-ad80-4398-af12-d96689f429e4","order_by":4,"name":"Xinyao Zhang","email":"","orcid":"","institution":"Inner Mongolia Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xinyao","middleName":"","lastName":"Zhang","suffix":""},{"id":549888500,"identity":"4b145a65-de4f-4454-bd2a-ce76f082566a","order_by":5,"name":"Lei Li","email":"","orcid":"","institution":"Inner Mongolia Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Li","suffix":""},{"id":549888504,"identity":"30db4599-a765-4074-bf7d-304565d2f5db","order_by":6,"name":"Zhijuan Guo","email":"","orcid":"","institution":"Inner Mongolia Medical University, China；Peking Universi ty Cancer Hospital (Inner Mongolia Campus) \u0026 Affiliated Canc er Hospital of Inner Mongolia Medical University, Inner Mongolia Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Zhijuan","middleName":"","lastName":"Guo","suffix":""},{"id":549888506,"identity":"1be214ef-e1a1-4b5d-87fe-f02557cf868b","order_by":7,"name":"Shenyi Lian","email":"","orcid":"","institution":"Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Department of Pathology, Peking University Cancer Hospital \u0026Institute","correspondingAuthor":false,"prefix":"","firstName":"Shenyi","middleName":"","lastName":"Lian","suffix":""},{"id":549888507,"identity":"4a34c9d6-3e03-4bf4-966c-6d11358348ae","order_by":8,"name":"Zhongwu Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuklEQVRIiWNgGAWjYDACCSB+YGCTAOYkFBCrJcEgLYGBDcwgWgvDYYgWBmK0yM/uMZNIKDifxy/fnfjhgQGDPL/YAfxaGOecAWoxuF0s2ca7GchgMJw5OwG/FmaJ3G0gLYkbjvFuAGkBsgloYYNoOQfSsvkHUVp4IFoOgLRsI84WCYn8zxYJBsmJM9tytwEZEoT9Ij8jLfHGhz92if3MZzff/FFhI88vTUALhq2kKR8Fo2AUjIJRgB0AALnsP+2Zf98FAAAAAElFTkSuQmCC","orcid":"","institution":"Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Department of Pathology, Peking University Cancer Hospital \u0026Institute","correspondingAuthor":true,"prefix":"","firstName":"Zhongwu","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2025-11-01 07:53:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8003880/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8003880/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96916138,"identity":"100e4a11-6c40-4b6b-918e-250f2941c7ac","added_by":"auto","created_at":"2025-11-27 14:08:05","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":673354,"visible":true,"origin":"","legend":"","description":"","filename":"EstablishingaDifferentialDiagnosisMethodforMelanomaUsingBloodParametersBasedonMachineLearningTechnology.docx2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/0162713033a26ae9dfbf7ca8.docx"},{"id":96787581,"identity":"c2598d73-d5cd-4a41-98b4-bd12a5dce7e9","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9420,"visible":true,"origin":"","legend":"","description":"","filename":"d234374b1ef74fba98826fe1321f635e.json","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/a59022f7e48a985847167756.json"},{"id":96787578,"identity":"d490aa7a-d7f4-4a43-8d4c-d9a61825f8cc","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":305678,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/2efaf0a4487cdbba6f7d875e.png"},{"id":96915095,"identity":"b31eb176-f8d9-4714-b366-ec1a7818f87f","added_by":"auto","created_at":"2025-11-27 14:06:51","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":91150,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/b87148e22ead81cabe9bfc63.png"},{"id":96787582,"identity":"aab98cc3-ec9b-41b2-b2dd-d0eed0589f98","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":128449,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/05c00910609d8c8faf56d4ff.png"},{"id":96787584,"identity":"68af4efe-8206-4a6b-9a3b-f2adcf607f44","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":84282,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/8611e1db7f98aeb901998da5.png"},{"id":96787585,"identity":"8ad561d4-bf3f-4322-a1bc-8bd4bfdaab34","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"wmf","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":708,"visible":true,"origin":"","legend":"","description":"","filename":"image1.wmf","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/bb8bb68248c2c6c7208a6980.wmf"},{"id":96787589,"identity":"de6c812c-b100-44e0-853a-3b3529d067aa","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"wmf","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":528,"visible":true,"origin":"","legend":"","description":"","filename":"image2.wmf","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/30447a44b8eb4bcb51bc6b98.wmf"},{"id":96787591,"identity":"6496a10c-8cb2-4a4b-a7aa-9aa3665c9191","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"wmf","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":532,"visible":true,"origin":"","legend":"","description":"","filename":"image3.wmf","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/14aad73b54a6c0ebf7456fd0.wmf"},{"id":96787588,"identity":"0ded1dc9-f928-4522-ad01-f6b634016103","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"wmf","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":688,"visible":true,"origin":"","legend":"","description":"","filename":"image4.wmf","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/a53037c40a0555dbdf71964c.wmf"},{"id":96787596,"identity":"1f912fad-edbf-4c2d-ba9e-a90db722ce81","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":65780,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/28bfd9985665e7ac5e24fd60.png"},{"id":96787583,"identity":"79253590-9f03-4048-b90d-ba2f283d4818","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27989,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/049228703380a6188c7abcb9.png"},{"id":96916174,"identity":"941c3a43-8b28-439d-be25-a2649a2570a6","added_by":"auto","created_at":"2025-11-27 14:08:08","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":31507,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/d0e80a35307e62959b172c6b.png"},{"id":96787592,"identity":"def65411-33b2-4355-8fe0-3b5646b82c5e","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22044,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/6f4caed0ab177e12165ecad4.png"},{"id":96787590,"identity":"93f0389e-087b-4c60-b550-b3a5fa6efacc","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1264,"visible":true,"origin":"","legend":"","description":"","filename":"Onlineimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/a511d1e0ab4fae07cae4ebb0.png"},{"id":96787593,"identity":"c52ab69b-e8d9-4f64-b6c4-8988e4aeef76","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":826,"visible":true,"origin":"","legend":"","description":"","filename":"Onlineimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/ac283fac3a8fa7b52a2caf00.png"},{"id":96787597,"identity":"29412ebe-4714-4a0d-96d3-918851c65626","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1056,"visible":true,"origin":"","legend":"","description":"","filename":"Onlineimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/1f80001bff87e123c99ddce9.png"},{"id":96787594,"identity":"2da657e7-4794-4e20-be82-c6a1075855db","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1660,"visible":true,"origin":"","legend":"","description":"","filename":"Onlineimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/06255f467a8ff5d778ef77d1.png"},{"id":96787577,"identity":"a22cfa77-79bf-43e4-9d35-faaf0a62120f","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":305678,"visible":true,"origin":"","legend":"\u003cp\u003eVisualization of the Dataset\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/901ba9b3cac9f5d6710de76c.png"},{"id":96787575,"identity":"aa66fbd0-259d-40a8-b4f3-f5bfc8ba2afb","added_by":"auto","created_at":"2025-11-26 06:27:02","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":91150,"visible":true,"origin":"","legend":"\u003cp\u003eROC Curve and Calibration Plot for Melanoma Model\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/e3ba9017bb69ae8f118d8383.png"},{"id":96916547,"identity":"dfccc8b2-190e-406d-b525-38ae732c58b6","added_by":"auto","created_at":"2025-11-27 14:08:41","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":128449,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP Summary Plot for Melanoma Model\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/a243873eac42c169a8d9763f.png"},{"id":96787587,"identity":"03a65d6b-7216-4c2d-8ab0-8431255b0660","added_by":"auto","created_at":"2025-11-26 06:27:03","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":84282,"visible":true,"origin":"","legend":"\u003cp\u003eSingle-sample interpretability plot of SHAP for melanoma model\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/65de269bb943bd773848d66f.png"},{"id":97126752,"identity":"da57d511-3bd1-44a0-a603-705bf028ff6b","added_by":"auto","created_at":"2025-12-01 08:25:19","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1519449,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8003880/v1/27212670-a30c-4683-a261-dfc543eb54b2.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Establishing a Differential Diagnosis Method for Melanoma Using Blood Parameters Based on Machine Learning Technology","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAs the largest organ in the human body, skin cancer primarily encompasses two major categories: melanoma and non-melanoma skin cancers. Although melanoma accounts for only about 4% of skin cancer cases, it causes nearly 80% of skin cancer-related deaths, making it one of the most lethal types \u003csup\u003e[1]\u003c/sup\u003e. Globally, melanoma incidence and mortality rates are rising annually, making it a significant public health concern. In China, it ranks among the fastest-growing malignant tumors \u003csup\u003e[2]\u003c/sup\u003e. Common pathological subtypes include superficial spreading melanoma, nodular melanoma, malignant lentigo maligna melanoma, and acral lentiginous melanoma. Additional rare subtypes exist, such as epithelioid melanoma and fibroblastic melanoma \u003csup\u003e[3]\u003c/sup\u003e. The distribution of subtypes varies across ethnic groups: superficial spreading melanoma predominates in Caucasians, while acral lentiginous melanoma is more common in Asians and Africans \u003csup\u003e[4]\u003c/sup\u003e. Although Caucasians exhibit higher melanoma incidence, people of color demonstrate poorer melanoma-specific survival rates \u003csup\u003e[5]\u003c/sup\u003e. Exposure to ultraviolet radiation is the recognized primary risk factor for cutaneous melanoma \u003csup\u003e[6, 7]\u003c/sup\u003e. Early diagnosis is crucial for melanoma treatment, as localized surgery in early-stage patients reduces trauma, lowers medical costs, and significantly improves survival rates \u003csup\u003e[8,9]\u003c/sup\u003e. Melanomas confined to the epidermis carry an extremely low risk of death, and thin melanomas have a lower probability of metastasis. In contrast, advanced-stage patients are prone to recurrence and metastasis, resulting in high mortality rates \u003csup\u003e[10-12]\u0026nbsp;\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe current gold standard for melanoma diagnosis remains invasive tissue biopsy, though this approach may lead to unnecessary surgeries \u003csup\u003e[13]\u003c/sup\u003e. Pathologists exhibit considerable subjective variation in histopathological interpretation \u003csup\u003e[14]\u003c/sup\u003e, suggesting potential for overdiagnosis and underdiagnosis \u003csup\u003e[15]\u003c/sup\u003e. Studies report misdiagnosis as a contributing factor to mortality in many melanoma patients\u003csup\u003e\u0026nbsp;[16, 17]\u003c/sup\u003e. Dermoscopy, a commonly used adjunct tool, shows diagnostic efficacy significantly influenced by physician experience and lesion characteristics \u003csup\u003e[18,19]\u003c/sup\u003e. However, early-stage melanoma may lack specific dermoscopic features, making diagnosis challenging even with dermoscopy assistance \u003csup\u003e[20, 21]\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eBlood tests and fluid biopsies, as important non-invasive methods for cancer diagnosis, have been widely applied in tumor diagnosis and treatment monitoring \u003csup\u003e[22]\u003c/sup\u003e. By detecting circulating tumor DNA \u003csup\u003e[23]\u003c/sup\u003e, protein biomarkers \u003csup\u003e[24]\u003c/sup\u003e, enzymes\u003csup\u003e[25,26]\u003c/sup\u003e, metabolites\u003csup\u003e[27]\u003c/sup\u003e, and antigens/antibodies\u003csup\u003e[28,29]\u003c/sup\u003e in plasma, they offer novel approaches for non-invasive tumor diagnosis and dynamic monitoring. Inflammatory markers and biochemical parameters within blood test results are also closely associated with the occurrence, staging, progression, survival, and prognosis of various cancers \u003csup\u003e[30-33]\u003c/sup\u003e. Furthermore, analyzing multiple blood parameters can assist in identifying tumor patients suitable for surgical resection. Dynamic monitoring of tumor-associated biomarkers in blood further enhances the accuracy of early diagnosis \u003csup\u003e[34].\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eWith the deepening application of machine learning technologies in the medical field \u003csup\u003e[35]\u003c/sup\u003e, they can uncover underlying patterns from complex data \u003csup\u003e[36]\u003c/sup\u003e. This supports disease prediction \u003csup\u003e[37]\u003c/sup\u003e, diagnosis \u003csup\u003e[38]\u003c/sup\u003e, prognosis assessment \u003csup\u003e[39]\u003c/sup\u003e, and risk evaluation \u003csup\u003e[40]\u003c/sup\u003e. Current research on the association between blood parameters and melanoma remains limited. This study aims to combine machine learning methods with the analysis of blood parameters in melanoma patients to explore their value in auxiliary diagnosis , providing new evidence for early screening and diagnosis.\u003c/p\u003e"},{"header":"Methods and Materials","content":"\u003cp\u003e\u003cstrong\u003eData Collection and Clinical Definitions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCollect clinical data from 4,534 patients at Peking University Cancer Hospital between March 2015 and November 2022 for retrospective study. Patient demographic information extracted from electronic medical records included gender, age, date of visit, patient ID number, clinical diagnosis, department of visit, and laboratory test results for complete blood count and biochemistry (Table 1).The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Peking University Cancer Hospital (Approval No. KY202402). Written informed consent was obtained from all participants.\u003c/p\u003e\n\u003cp\u003eTable 1 \u0026nbsp;All parameter abbreviations and full names\u003c/p\u003e\n\u003cdiv align=\"center\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"99%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAbbreviation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFull name\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eWBC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eWhite Blood Cell Count\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eLYMPH%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eLymphocyte Ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eMONO%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eMonocyte Ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eNEUT%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eNeutrophil Ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eEOS%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eEosinophil Ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eBASO%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eBasophil Ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eLYMPTH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eLymphocyte Count\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eMONO#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eMonocyte Count\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eNEUT#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eNeutrophil Count\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eEOS#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eEosinophil Count\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eBASO#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eBasophil Count\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eMCV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eMean Corpuscular Volume\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eMCH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eMean Corpuscular Hemoglobin\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eMCHC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eMean Corpuscular Hemoglobin Concentration\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003ePLT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003ePlatelet\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eMPV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eMean Platelet Volume\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003ePCT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003ePlatelet Crit\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eK\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003ePotassium\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eNa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eSodium\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eCl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eChlorine\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eCa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eCalcium\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eInorganic Phosphorus\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eGLU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eGlucose\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eUREA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eUrea\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eCr\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eCreatinine (Enzymatic Method\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eUA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eUric Acid\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eTP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eTotal Protein\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eTBIL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eTotal Bilirubin\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eDBIL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eDirect Bilirubin\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eGGT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eY-Glutamyl Transferase\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eCK\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eCreatine Kinase\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eLDH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eLactate Dehydrogenase\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eTG\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eTriglyceride\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eLDL-C\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eLow-Density Lipoprotein Cholesterol\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eHDL-C\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eHigh-Density Lipoprotein Cholesterol\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eTBA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eTotal Bile Acid\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eInclusion and Exclusion Criteria for Melanoma\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInclusion Criteria : 1. Patients diagnosed with melanoma who underwent surgical treatment at Peking University Cancer Hospital and possess complete pathological diagnosis results. 2. Patients who underwent complete blood count and biochemical testing at Peking University Cancer Hospital and have multiple pathological results; the most recent test data prior to surgery will be selected.\u003c/p\u003e\n\u003cp\u003eExclusion Criteria: 1. Patients with concomitant malignant diseases. 2. Patients lacking complete blood count and biochemical test parameters from Peking University Cancer Hospital. 3. Patients with clinical records indicating prior chemotherapy before surgery.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTraining and Validation of Predictive Models\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Partitioning\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePrior to data partitioning, we performed preprocessing on the dataset, including missing value handling, outlier detection, and removal of unknown data, to ensure data quality and enhance model training stability and effectiveness. The preprocessed dataset retained 4,534 patient records, which were randomly divided into training and testing sets at an 8:2 ratio. The training set was used for model construction and parameter optimization, while the testing set evaluated the model\u0026apos;s generalization capability and performance, ensuring its stability in practical applications.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFeature Selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFeature selection is the process of filtering key variables from the original feature set, aiming to reduce redundant information and noise interference while enhancing model training efficiency and prediction accuracy. The Recursive Feature Elimination (RFE) method and Mutual Information (MI) method are employed to determine the optimal variables for feature selection. RFE is an iterative model-based feature selection method that trains models multiple times and progressively removes less important features, ultimately retaining the optimal subset. Specifically, it first trains a random forest model using all features to output feature importance scores. It then sequentially removes the least important features, repeating the training and removal process with the remaining features until predetermined feature count or performance stopping criteria are met. The Mutual Information method, grounded in information theory, screens features by measuring the strength of dependency between features and the target variable, unrestricted by linear relationships. Its core involves calculating the mutual information value between feature X and target Y. This value indicates the reduction in uncertainty of Y given X; a higher value signifies a stronger association between them. During screening, features with higher mutual information values are typically retained. It is applicable to various data distributions and relationship types but cannot reflect the directionality of the association.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBuilding the Predictive Model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe employed the eXtreme Gradient Boosting (XGBoost) algorithm to construct the predictive model. As an efficient ensemble learning algorithm based on the gradient boosting framework, XGBoost achieves high-precision prediction by iteratively building multiple decision trees and combining their predictions through weighted aggregation. Its core mechanism employs a gradient descent strategy, generating a new decision tree in each iteration to fit the prediction residuals of the preceding model. By continuously stacking tree models, it optimizes overall performance. Simultaneously, it incorporates a regularization term to control tree complexity and prevent overfitting. Additionally, it supports parallel computing to accelerate training and possesses adaptive handling capabilities for missing values.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel Evaluation and Test Set\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo validate the model\u0026apos;s generalization capability, we employed 10-fold cross-validation for model validation. The metrics for evaluating the prediction model include Accuracy, Recall, Specificity, Area Under the Receiver Operating Characteristic Curve (AUC), and F1-score. These metrics are calculated based on the True Negative (TN), True Positive (TP), False Negative (FN), and False Positive (FP) values from the confusion matrix, using the following formulas:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"https://myfiles.space/user_files/58895_8739fc6c57c1c19a/58895_custom_files/img1764061519.png\" width=\"511\" height=\"355\"\u003e\u003c/p\u003e\n\u003cp\u003eAmong these, the F1-score is a synthetic evaluation metric that measures the harmonic mean of a classification model\u0026apos;s precision and recall. Precision is the ratio of correctly identified positives to all predicted positives: Precision = TP / (TP + FP).\u003c/p\u003e\n\u003cp\u003eAUC is a crucial metric for evaluating classification model performance, quantifying the model\u0026apos;s classification capability through the area under the ROC curve. It offers advantages such as threshold independence and insensitivity to class distribution, making it particularly suitable for imbalanced datasets. AUC ranges from 0 to 1, with values closer to 1 indicating superior model performance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel Interpretation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo enhance understanding of the model\u0026apos;s operational principles and prediction outcomes, we employ the SHAP method to elucidate factors influencing model predictions. SHAP is a game-theoretic model interpretation technique that quantifies feature impact on predictions by calculating each feature\u0026apos;s contribution value to the model output (i.e., SHAP values). The core concept of SHAP values originates from Shapley values, ensuring fair and consistent contribution allocation. Compared to traditional feature importance methods, SHAP not only explains global feature importance but also provides local explanations for individual sample predictions, helping users understand the model\u0026apos;s decision-making basis in specific scenarios. Through visualization tools, users can intuitively analyze the relationship between features and model predictions, thereby enhancing the model\u0026apos;s credibility and transparency.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eModel development and validation were conducted using Python 3.8.3 (library, sci-kit-learn). Descriptive analysis employed SPSS 22.0 (IBM, Chicago, IL, USA). Normally distributed continuous variables are presented as mean \u0026plusmn; standard deviation (mean \u0026plusmn; SD), while non-normally distributed data are shown as median (interquartile range [IQR]). The Kolmogorov-Smirnov test was used to assess data distribution patterns. Differences in group characteristics were analyzed using t-tests or Mann-Whitney U tests as appropriate, with P \u0026lt; 0.05 considered statistically significant.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eOriginal Variables\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 2,000 melanoma cases were finalized as the positive experimental group, and 2,534 healthy screening participants were randomly selected as the negative control group（Table 2）. \u0026nbsp;The mean age of melanoma patients was significantly higher than that of healthy examinees (56 years vs. 42 years, P\u0026lt;0.001), indicating age may be an important risk factor for melanoma. Additionally, no significant difference in gender distribution was observed between groups (male proportion: melanoma 49.5%, healthy examinees 49.1%). Among the 36 blood routine and biochemical indicators analyzed, four parameters\u0026mdash;K, Ca, TBA, and MCH\u0026mdash;showed no statistically significant differences between groups (P \u0026gt; 0.05). The remaining 35 indicators exhibited significant differences (P \u0026lt; 0.05). Inflammation-related markers showed significant differences between groups: melanoma patients exhibited markedly elevated WBC, NEUT%, and NEUT#, while LYMPH% and LYMPH# were significantly reduced. MPV and PCT were markedly decreased in melanoma patients, with no significant difference in PLT. Melanoma patients exhibited significantly elevated LDH, GLU, TG, and LDL-C levels, while HDL-C levels were significantly reduced. DBIL and GG were significantly elevated in melanoma patients. These differences suggest that melanoma patients may experience systemic inflammatory responses, metabolic disorders, and immune dysfunction, providing potential feature variables for subsequent model construction.\u003c/p\u003e\n\u003cp\u003eTable 2 Comparison of Blood Parameters, Gender, and Age Between Melanoma Patients and Healthy Check-up Participants\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" summary='{\"row\":{\"emphasizeNum\":1,\"emphasizeType\":2,\"originStyle\":{\"bBoldFont\":false,\"bValid\":true,\"backgroundColor\":\"#ffffff\",\"borders\":[{\"borderColor\":\"#000000\",\"borderType\":-4,\"lineStyle\":0,\"lineWidth\":0},{\"borderColor\":\"#000000\",\"borderType\":-3,\"lineStyle\":1,\"lineWidth\":4},{\"borderColor\":\"#000000\",\"borderType\":-2,\"lineStyle\":0,\"lineWidth\":0},{\"borderColor\":\"#000000\",\"borderType\":-1,\"lineStyle\":1,\"lineWidth\":4}],\"fontColor\":\"#000000\"}}}' width=\"546\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLabel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGroup\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAverage(Standard)/\u003c/strong\u003e\u003cstrong\u003eMedian(25 percent, 75 percent\u003c/strong\u003e\u003cstrong\u003e）\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e42（35, 51）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e56（46, 64）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eK\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e4.202（0.294）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.214\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e4.214（0.356）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eNa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e139.296（1.810）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e140.543（2.467）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eCl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e103.701（1.949）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e103.339（2.949）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eCa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e2.3753（0.086）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.188\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e2.3797（0.130）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.1546（0.158）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.198（0.177）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eGLU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e5.4（5, 5.8）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e5.485（5.08, 6.13）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eUREA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e4.7（4, 5.5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e4.87（4.04, 5.780）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eCr\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e67.2（58.7, 78.9）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e60（51, 71）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eUA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e318（265, 387）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e312（259, 373）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eTP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e73.8（71.2, 76.5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e72.4（69.2, 75.4）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eTBIL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e13.3（10.7, 16.9）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e12.7（10.1, 16.1）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eDBIL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e2.4（2.582, 2.661）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e3.4（2.8, 4.4）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eGGT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e20.5（15, 33）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e25（17, 41）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eCK\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e91（70, 131）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e62（45, 84）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eLDH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e168（151, 187）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e192.5（168, 187）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eTG\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.12（0.76, 1.73）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.43（1.033, 2.06）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eLDL-C\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e3.049（0.819）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e3.137（0.861）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eHDL-C\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.484（O.415）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.209（0.317）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eTBA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e2.6（1.8, 3.9）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.095\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e2.6（1.5, 4.3）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eWBC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e5.885（5, 7）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e5.99（4.92, 7.18）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eLYMPH%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e33.578（7.254）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e26.743（9.304）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eMONO%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e6（5.1, 7.2）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e5.6（4.6, 6.7）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eNEUT%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e57.342（7.877）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e65.086（9.982）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eEOS%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.8（1.1, 2.8）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.4（0.7, 2.4）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eBASO%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.6（0.4, 0.8）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.4（0.3, 0.8）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eLYMPH#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.93（1.61, 2.33）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e1.52（1.21, 1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eMONO#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.36（0.29, 0.44）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.33（0.26, 0.43）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eNEUT#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e3.37（2.71, 4.173）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e3.85（2.97, 4.908）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eEOS#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.1（0.06, 0.17）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.08（0.04, 0.14）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eBASO#\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.03（0.02, 0.05）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.02（0.02, 0.03）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eMCV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e91.6（88.9, 94.1）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e91.1（88, 93.8）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eMCH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e30.2（29.2, 31.1）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.24\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e30.066（2.183）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eMCHC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e328.45（11.917）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e331.65（10.405）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003ePLT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e250（215, 288）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e230（191, 276）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003eMPV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e10.256（0.899）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e9.311（1.134）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003ePCT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eHealth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2534\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.2592（0.056）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003eMelanoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e2000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 242px;\"\u003e\n \u003cp\u003e0.2205（0.063）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNote: Following normality testing, normally distributed continuous variables are presented as mean \u0026plusmn; standard deviation (SD), while non-normally distributed data are represented by median (interquartile range [IQR]); abbreviations are listed in Table 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFeature Selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study employed both RFE and MI methods for feature screening. First, RFE identified 27 variables from the original dataset (Figure. 1 A). followed by MI screening yielding 19 variables (Figure. 1 B). To obtain a more representative and robust feature set, the results from both methods were intersected, ultimately yielding 15 common variables (Figure. 1 C) for predictive model construction. These variables included Age, BASO#, LYMPH#, PCT, HDL-C, BASO%, LDH, LYMPH%, DBIL, NEUT%, CK, MPV, Na, Cr, and GLU. Although the XGBoost model itself is insensitive to multicollinearity, to avoid errors in model interpretation caused by highly linearly correlated features, the study further analyzed variable correlations by calculating correlation coefficients and plotting a heatmap. Results showed only two pairs of variables\u0026mdash;LYMPH% and NEUT%, and BASO% and BASO#\u0026mdash;exhibited significant linear relationships, while linear correlations among other variables were weak (Figure. 1D). Given the limited number of linearly correlated variables and their negligible impact on machine learning prediction performance, no further multicollinearity handling was performed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePredictive Model Performance and Validation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe training set comprises 3,627 samples, including 1,600 positive samples and 2,027 negative samples. The test set comprises 907 samples, including 400 positive samples and 507 negative samples. When constructing the XGBoost prediction model using the training set, the optimal hyperparameter combination was determined through 5-fold cross-validation using the GridSearchCV method within the defined parameter grid. The parameter grid included: n_estimators ranging from [80, 100, 120, 150], max_depth from [8, 10, 12, 14], and learning_rate from [0.05, 0.1, 0.2]. The validated optimal configuration was: n_estimators = 120, max_depth = 12, learning_rate = 0.1. Additionally, to address the imbalanced distribution of positive and negative samples in the training set, we adjusted the weights of different class samples via the class_weight parameter to balance category disparities. Simultaneously, regularization was applied to prevent certain features from being assigned excessive weights, thereby avoiding model over-reliance on noise or peculiar patterns within local training data. We comprehensively analyzed the model\u0026apos;s performance on the test set using multiple evaluation metrics, ROC curves, and calibration plots. Table 2 details the model performance evaluation results based on 10-fold cross-validation for both the training and test sets, including accuracy, sensitivity, specificity, AUC, and F1-Score metrics, providing quantitative measures of model performance（Table 3）. The ROC curve visually depicts the model\u0026apos;s classification performance at different thresholds. Both the training and test sets achieved AUC values exceeding 0.95, indicating excellent model performance (Figure. 2 A). Calibration plots assess the alignment between predicted and actual labels. Overall, minimal discrepancies were observed, further validating the model\u0026apos;s reliability (Figure. 2 B).\u003c/p\u003e\n\u003cp\u003eTable 3 Model Performance Table\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"549\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDataset\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy rate\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSensitivity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003especificity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1-Score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 88px;\"\u003e\n \u003cp\u003eTraining set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.981\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.957\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.999\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.996\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e0.978\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 88px;\"\u003e\n \u003cp\u003eTest set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.926\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.920\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.931\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.975\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e0.917\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eAssessment of Factors Influencing Malignant Melanoma\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo thoroughly analyze the contribution of each feature to the predictive model, this study employs the SHAP method to interpret the XGBoost model. SHAP values quantify the model\u0026apos;s predictive logic, clearly illustrating the specific impact of each feature on the prediction outcome. Different colors represent the direction of a feature\u0026apos;s contribution to the prediction: red indicates a positive contribution, while blue signifies a negative contribution. Specifically, a positive SHAP value indicates that the feature increases the predicted risk of malignant melanoma, while a negative value reflects its role in reducing the predicted risk. For the XGBoost model, Figure 3 displays the global feature importance. Among the 15 variables, the top six key predictive indicators are MPV, DBIL, CK, Na, HDL-C, and LDH (Figure 3). Among these, DBIL, Na, and LDH exhibit significant positive correlations with disease risk, meaning elevated levels of these indicators synergistically increase an individual\u0026apos;s risk of malignant melanoma. Conversely, MPV, CK, and HDL-C demonstrate pronounced negative regulatory effects, with their increased values strongly correlated with reduced disease risk.\u003c/p\u003e\n\u003cp\u003eIn local explanations, SHAP waterfall plots visually demonstrate the magnitude and direction of each feature\u0026apos;s impact on the model\u0026apos;s final prediction. The baseline value (E[f(x)]) represents the average SHAP value across all sample predictions, serving as the model\u0026apos;s starting reference point. Features increasing the prediction value relative to this baseline are highlighted in red, while those decreasing the prediction value are shown in blue. The final f(x) value represents the model output after integrating all feature contributions, constituting the final prediction for that sample. Figure 4 displays SHAP single-sample explanations for two samples, where the six features with smaller contributions are grouped under \u0026ldquo;Other features,\u0026rdquo; with their combined contribution presented as a single value. In Figure 4 A, MPV, LDH, HDL-C, LYMPH%, CK, DBIL, Age, and NEUT% collectively drive the predicted value upward, indicating a positive influence of these factors on the sample. Conversely, in Figure 4 B, Age, HDL-C, MPV, NEUT%, LDH, PCT, and BASO% collectively lower the predicted value, suggesting a negative impact of these features on the sample\u0026apos;s outcome.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study successfully developed a machine learning model based on conventional blood parameters for the differential diagnosis of melanoma, demonstrating exceptional predictive performance (test set AUC = 0.975). This significantly outperforms the independent predictive capability of traditional single indicators for melanoma, which typically exhibit AUC values below 0.8 \u003csup\u003e[41]\u003c/sup\u003e. Furthermore, the XGBoost algorithm employed in this study excels at handling high-dimensional, nonlinear medical data. Its ensemble learning properties effectively prevent overfitting and enhance the model\u0026apos;s generalization capability. Combined with interpretability analysis using SHAP values, we can not only evaluate overall feature importance but also provide local explanations for individual patient predictions. This significantly enhances the model\u0026apos;s transparency and credibility in clinical decision-making \u003csup\u003e[42]\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eFrom a mechanistic perspective, the key indicators identified in this study are closely associated with the pathophysiological processes of melanoma. Among these, MPV (Mean Platelet Volume) is a critical parameter reflecting the average size of platelets in the blood, indicating their production rate and stimulation level, and providing important insights into numerous pathological conditions \u003csup\u003e[43]\u003c/sup\u003e. Elevated MPV levels were observed in patients with gastric, breast, endometrial, thyroid, and lung cancers. In esophageal, ovarian, and colorectal cancers, the trend in MPV changes remains unclear. Conversely, reduced MPV was noted in renal cell and gallbladder cancers \u003csup\u003e[44]\u003c/sup\u003e. This decrease may relate to abnormal platelet consumption within the tumor microenvironment \u003csup\u003e[45]\u003c/sup\u003e. Conversely, certain hematologic malignancies stimulate excessive megakaryocyte production of new platelets, leading to elevated MPV \u003csup\u003e[46]\u003c/sup\u003e. HDL-C transports excess cholesterol from peripheral tissues back to the liver for metabolism and clearance via the reverse transport mechanism, thereby effectively counteracting atherosclerosis. Furthermore, HDL-C possesses multiple protective functions including anti-inflammatory, antioxidant, vascular endothelial protection, and anti-thrombotic effects. Dysfunction of HDL-C may increase cancer risk. Multiple large-scale retrospective studies indicate that a decline in HDL-C levels from the normal range significantly elevates the risk of gastric cancer, colorectal cancer, liver cancer, pancreatic cancer, gallbladder cancer, and biliary tract cancer \u003csup\u003e[47,48]\u003c/sup\u003e. This finding aligns with the prevalent low HDL-C levels observed in melanoma patients in this study, further supporting the association between low HDL-C and the development of various malignancies.\u003c/p\u003e\n\u003cp\u003eConcurrently, this study found that sodium ions (Na⁺) and lactate dehydrogenase (LDH) levels were elevated in melanoma patients compared to healthy controls. Elevated serum sodium is not uncommon in cancer patients, as abnormal expression of sodium channels in tumor tissues increases intracellular sodium concentrations, thereby promoting invasion and metastasis. Studies indicate that sodium concentrations in breast cancer and its metastatic lesions are significantly higher than in adjacent normal tissue. Elevated sodium levels exert pro-inflammatory effects, suppress immune cell function, and promote resistance to chemotherapy\u003csup\u003e[49]\u003c/sup\u003e. LDH, a key glycolytic enzyme, exhibits elevated serum levels that not only indicate tumor burden and hypoxic necrosis but also promote metastasis through a lactate-mediated immunosuppressive microenvironment \u003csup\u003e[44]\u003c/sup\u003e. It is closely associated with tumor cell metabolic reprogramming, melanoma invasiveness, and poor prognosis \u003csup\u003e[50]\u003c/sup\u003e. This study also found significant differences in LDH levels between melanoma patient blood samples and negative samples (P\u0026lt;.001), consistent with previous research.\u003c/p\u003e\n\u003cp\u003eThis study has several limitations. First, all data originated from a single center, potentially introducing selection bias. Second, the retrospective study design cannot entirely eliminate confounding factors. Finally, the model has not been validated in prospective cohorts or external multicenter data, and its generalizability requires further confirmation.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe results above indicate that the selected blood markers are associated with melanoma to some extent. However, these markers do not demonstrate clear diagnostic value when considered individually. We can effectively distinguish specific diseases by harnessing the power of machine learning to integrate multiple test indicators and analyze the complex interrelationships among their patterns.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u003c/strong\u003e Conceptualization, Qijiao Wang, Xia Liu, andRuihuan Zhang; methodology, Qijiao Wang and Lei Li; software, Qi Zhang and Xinyao Zhang; validation, Zhijuan Guo, Shenyi Lian, and Ruihuan Zhang; project administration, Zhongwu Li; funding acquisition, Zhongwu Li. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u0026nbsp;\u003c/strong\u003eThis work was supported by Capital\u0026rsquo;s Funds for Health Improvement and Research 2022\u0026ndash;2-1024,the Natural Science Foundation of Inner Mongolia Autonomous Region of China (Grants No. 2024ZD31), Science and Technology Program of the Joint Fund of Science Research for the Public Hospital of Inner Mongolian Academy of Medical Sciences(2023GLLH0125).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInstitutional Review Board Statement:\u0026nbsp;\u003c/strong\u003eThe study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Peking University Cancer Hospital (Approval No. KY202402 and date of approval:2024.01).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed Consent Statement:\u0026nbsp;\u003c/strong\u003eWritten informed consent was obtained from all subjects involved in the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement:\u0026nbsp;\u003c/strong\u003eThe data presented in this study are available on request from the corresponding author due to (e.g., privacy and ethical concerns).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest:\u0026nbsp;\u003c/strong\u003eThe authors declare no conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments: \u003c/strong\u003eWe sincerely thank our mentor, Professor Zhongwu Li, for his expert guidance. Our gratitude also goes to all teachers and colleagues who provided support and facilitation for this study\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eZAREMBA A, ZIMMER L, GRIEWANK K G, et al. [Immunotherapy for malignant melanoma] [J]. Internist (Berl), 2020, 61(7): 669-675.\u003c/li\u003e\n\u003cli\u003eSUNG H, FERLAY J, SIEGEL R L, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries [J]. CA Cancer J Clin, 2021, 71(3): 209-249.\u003c/li\u003e\n\u003cli\u003ePUCKETT Y, WILSON A M, FARCI F, et al. Melanoma Pathology [M]. StatPearls. Treasure Island (FL); StatPearls PublishingCopyright \u0026copy; 2025, StatPearls Publishing LLC. 2025.\u003c/li\u003e\n\u003cli\u003eMADANKUMAR R, GUMASTE P V, MARTIRES K, et al. Acral melanocytic lesions in the United States: Prevalence, awareness, and dermoscopic patterns in skin-of-color and non-Hispanic white patients [J]. J Am Acad Dermatol, 2016, 74(4): 724-730.e721.\u003c/li\u003e\n\u003cli\u003eQIAN Y, JOHANNET P, SAWYERS A, et al. The ongoing racial disparities in melanoma: An analysis of the Surveillance, Epidemiology, and End Results database (1975-2016) [J]. J Am Acad Dermatol, 2021, 84(6): 1585-1593.\u003c/li\u003e\n\u003cli\u003eDEWANE M E, KELSEY A, OLIVIERO M, et al. Melanoma on chronically sun-damaged skin: Lentigo maligna and desmoplastic melanoma [J]. J Am Acad Dermatol, 2019, 81(3): 823-833.\u003c/li\u003e\n\u003cli\u003eMADDEN K, FORMAN S B, ELSTON D. Quantification of melanocytes in sun-damaged skin [J]. J Am Acad Dermatol, 2011, 64(3): 548-552.\u003c/li\u003e\n\u003cli\u003eGARBE C, AMARAL T, PERIS K, et al. European consensus-based interdisciplinary guideline for melanoma. Part 1: Diagnostics - Update 2024 [J]. Eur J Cancer, 2025, 215: 115152.\u003c/li\u003e\n\u003cli\u003ePETRIE T, SAMATHAM R, WITKOWSKI A M, et al. Melanoma Early Detection: Big Data, Bigger Picture [J]. J Invest Dermatol, 2019, 139(1): 25-30.\u003c/li\u003e\n\u003cli\u003eNAIK P P. Cutaneous Malignant Melanoma: A Review of Early Diagnosis and Management [J]. World J Oncol, 2021, 12(1): 7-19.\u003c/li\u003e\n\u003cli\u003eHASHIM P W, FRIEDLANDER P, GOLDENBERG G. Systemic Therapies for Late-stage Melanoma [J]. J Clin Aesthet Dermatol, 2016, 9(10): 36-40.\u003c/li\u003e\n\u003cli\u003eGERSHENWALD J E, SCOLYER R A, HESS K R, et al. Melanoma staging: Evidence-based changes in the American Joint Committee on Cancer eighth edition cancer staging manual [J]. CA Cancer J Clin, 2017, 67(6): 472-492.\u003c/li\u003e\n\u003cli\u003eFEIT N E, DUSZA S W, MARGHOOB A A. Melanomas detected with the aid of total cutaneous photography [J]. Br J Dermatol, 2004, 150(4): 706-714.\u003c/li\u003e\n\u003cli\u003eELDER D E, PIEPKORN M W, BARNHILL R L, et al. Pathologist characteristics associated with accuracy and reproducibility of melanocytic skin lesion interpretation [J]. J Am Acad Dermatol, 2018, 79(1): 52-59.e55.\u003c/li\u003e\n\u003cli\u003ePathologists\u0026apos; diagnosis of invasive melanoma and melanocytic proliferations: observer accuracy and reproducibility study [J]. Bmj, 2017, 358: j3798.\u003c/li\u003e\n\u003cli\u003eBATAILLE V, SASIENI P, CURLEY R K, et al. Melanoma yield, number of biopsies and missed melanomas in a British teaching hospital pigmented lesion clinic: a 9-year retrospective study [J]. Br J Dermatol, 1999, 140(2): 243-248.\u003c/li\u003e\n\u003cli\u003eCARLI P, NARDINI P, CROCETTI E, et al. Frequency and characteristics of melanomas missed at a pigmented lesion clinic: a registry-based study [J]. Melanoma Res, 2004, 14(5): 403-407.\u003c/li\u003e\n\u003cli\u003eLINDSAY D, SOYER H P, JANDA M, et al. Cost-Effectiveness Analysis of 3D Total-Body Photography for People at High Risk of Melanoma [J]. JAMA Dermatol, 2025, 161(5): 482-489.\u003c/li\u003e\n\u003cli\u003eFINK C, HAENSSLE H A. Non-invasive tools for the diagnosis of cutaneous melanoma [J]. Skin Res Technol, 2017, 23(3): 261-271.\u003c/li\u003e\n\u003cli\u003eGERAMI P, BUSAM K, COCHRAN A, et al. Histomorphologic assessment and interobserver diagnostic reproducibility of atypical spitzoid melanocytic neoplasms with long-term follow-up [J]. Am J Surg Pathol, 2014, 38(7): 934-940.\u003c/li\u003e\n\u003cli\u003eMARINO M L, CARRERA C, MARCHETTI M A, et al. Practice Gaps in Dermatology: Melanocytic Lesions and Melanoma [J]. Dermatol Clin, 2016, 34(3): 353-362.\u003c/li\u003e\n\u003cli\u003ePHALLEN J, SAUSEN M, ADLEFF V, et al. Direct detection of early-stage cancers using circulating tumor DNA [J]. Sci Transl Med, 2017, 9(403).\u003c/li\u003e\n\u003cli\u003eBETTEGOWDA C, SAUSEN M, LEARY R J, et al. Detection of circulating tumor DNA in early- and late-stage human malignancies [J]. Sci Transl Med, 2014, 6(224): 224ra224.\u003c/li\u003e\n\u003cli\u003eCOHEN J D, JAVED A A, THOBURN C, et al. Combined circulating tumor DNA and protein biomarker-based liquid biopsy for the earlier detection of pancreatic cancers [J]. Proc Natl Acad Sci U S A, 2017, 114(38): 10202-10207.\u003c/li\u003e\n\u003cli\u003eHU X, CHEN R, WEI Q, et al. The Landscape Of Alpha Fetoprotein In Hepatocellular Carcinoma: Where Are We? [J]. Int J Biol Sci, 2022, 18(2): 536-551.\u003c/li\u003e\n\u003cli\u003eHAMMARSTR\u0026ouml;M S. The carcinoembryonic antigen (CEA) family: structures, suggested functions and expression in normal and malignant tissues [J]. Semin Cancer Biol, 1999, 9(2): 67-81.\u003c/li\u003e\n\u003cli\u003eJOV\u0026eacute; M, COLLADO R, QUILES J L, et al. A plasma metabolomic signature discloses human breast cancer [J]. Oncotarget, 2017, 8(12): 19522-19533.\u003c/li\u003e\n\u003cli\u003eLACOMBE J, MANG\u0026eacute; A, JARLIER M, et al. Identification and validation of new autoantibodies for the diagnosis of DCIS and node negative early-stage breast cancers [J]. Int J Cancer, 2013, 132(5): 1105-1113.\u003c/li\u003e\n\u003cli\u003eDESMETZ C, BASCOUL-MOLLEVI C, ROCHAIX P, et al. Identification of a new panel of serum autoantibodies associated with the presence of in situ carcinoma of the breast in younger women [J]. Clin Cancer Res, 2009, 15(14): 4733-4741.\u003c/li\u003e\n\u003cli\u003eFEST J, RUITER R, MULDER M, et al. The systemic immune-inflammation index is associated with an increased risk of incident cancer-A population-based cohort study [J]. Int J Cancer, 2020, 146(3): 692-698.\u003c/li\u003e\n\u003cli\u003eN\u0026oslash;ST T H, ALCALA K, URBAROVA I, et al. Systemic inflammation markers and cancer incidence in the UK Biobank [J]. Eur J Epidemiol, 2021, 36(8): 841-848.\u003c/li\u003e\n\u003cli\u003ePODNAR S, KUKAR M, GUNČAR G, et al. Diagnosing brain tumours by routine blood tests using machine learning [J]. Sci Rep, 2019, 9(1): 14481.\u003c/li\u003e\n\u003cli\u003eWONG J Y Y, BASSIG B A, LOFTFIELD E, et al. White Blood Cell Count and Risk of Incident Lung Cancer in the UK Biobank [J]. JNCI Cancer Spectr, 2020, 4(2): pkz102.\u003c/li\u003e\n\u003cli\u003eCOHEN J D, LI L, WANG Y, et al. Detection and localization of surgically resectable cancers with a multi-analyte blood test [J]. Science, 2018, 359(6378): 926-930.\u003c/li\u003e\n\u003cli\u003eSWANSON K, WU E, ZHANG A, et al. From patterns to patients: Advances in clinical machine learning for cancer diagnosis, prognosis, and treatment [J]. Cell, 2023, 186(8): 1772-1791.\u003c/li\u003e\n\u003cli\u003eGREENER J G, KANDATHIL S M, MOFFAT L, et al. A guide to machine learning for biologists [J]. Nat Rev Mol Cell Biol, 2022, 23(1): 40-55.\u003c/li\u003e\n\u003cli\u003eQI X, WANG S, FANG C, et al. Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants [J]. Redox Biol, 2025, 79: 103470.\u003c/li\u003e\n\u003cli\u003eWU J, ZAN X, GAO L, et al. A Machine Learning Method for Identifying Lung Cancer Based on Routine Blood Indices: Qualitative Feasibility Study [J]. JMIR Med Inform, 2019, 7(3): e13476.\u003c/li\u003e\n\u003cli\u003eYUAN Q, YE R, QIAN Y, et al. Machine learning and SHAP value interpretation for predicting the response to neoadjuvant chemotherapy and long-term clinical outcomes in Chinese female breast cancer [J]. Ann Med, 2025, 57(1): 2541316.\u003c/li\u003e\n\u003cli\u003eLEI J, ZHAI J, ZHANG Y, et al. Supervised Machine Learning Models for Predicting Sepsis-Associated Liver Injury in Patients With Sepsis: Development and Validation Study Based on a Multicenter Cohort Study [J]. J Med Internet Res, 2025, 27: e66733.\u003c/li\u003e\n\u003cli\u003eZHANG J, YAO Y H, LI B G, et al. Prognostic value of pretreatment serum lactate dehydrogenase level in patients with solid tumors: a systematic review and meta-analysis [J]. Sci Rep, 2015, 5: 9800.\u003c/li\u003e\n\u003cli\u003eLIANG D, WANG L, ZHONG P, et al. Perspective: Global Burden of Iodine Deficiency: Insights and Projections to 2050 Using XGBoost and SHAP [J]. Adv Nutr, 2025, 16(3): 100384.\u003c/li\u003e\n\u003cli\u003eKORNILUK A, KOPER-LENKIEWICZ O M, KAMIŃSKA J, et al. Mean Platelet Volume (MPV): New Perspectives for an Old Marker in the Course and Prognosis of Inflammatory Conditions [J]. Mediators Inflamm, 2019, 2019: 9213074.\u003c/li\u003e\n\u003cli\u003eDETOPOULOU P, PANOUTSOPOULOS G I, MANTOGLOU M, et al. Relation of Mean Platelet Volume (MPV) with Cancer: A Systematic Review with a Focus on Disease Outcome on Twelve Types of Cancer [J]. Curr Oncol, 2023, 30(3): 3391-3420.\u003c/li\u003e\n\u003cli\u003eSUN Y, LI T, DING L, et al. Platelet-mediated circulating tumor cell evasion from natural killer cell killing through immune checkpoint CD155-TIGIT [J]. Hepatology, 2025, 81(3): 791-807.\u003c/li\u003e\n\u003cli\u003eHE F, LARANJEIRA A B, KONG T, et al. Multiomic profiling reveals metabolic alterations mediating aberrant platelet activity and inflammation in myeloproliferative neoplasms [J]. J Clin Invest, 2024, 134(3).\u003c/li\u003e\n\u003cli\u003eNAM S Y, JO J, CHO C M. A population-based cohort study of longitudinal change of high-density lipoprotein cholesterol impact on gastrointestinal cancer risk [J]. Nat Commun, 2024, 15(1): 2923.\u003c/li\u003e\n\u003cli\u003eNAM S Y, JO J, JEON S W. Discrepant effect of high-density lipoprotein cholesterol on esophageal and gastric cancer risk in a nationwide cohort [J]. Gastric Cancer, 2024, 27(3): 451-460.\u003c/li\u003e\n\u003cli\u003eJAMES A D, LESLIE T K, KAGGIE J D, et al. Sodium accumulation in breast cancer predicts malignancy and treatment response [J]. Br J Cancer, 2022, 127(2): 337-349.\u003c/li\u003e\n\u003cli\u003eCLAPS G, FAOUZI S, QUIDVILLE V, et al. The multiple roles of LDH in cancer [J]. Nat Rev Clin Oncol, 2022, 19(12): 749-762.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Melanoma, Machine Learning, Blood Biomarkers, Differential Diagnosis","lastPublishedDoi":"10.21203/rs.3.rs-8003880/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8003880/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study aims to develop and validate a differential diagnosis model for melanoma. This model will combine blood test results with machine learning techniques to provide an inexpensive, non-invasive method for early melanoma screening.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eComplete blood counts and biochemical indicators were retrospectively collectedfrom 4,534 patients for clinical data. Key variables were screened using the Recursive Feature Elimination (RFE) and Mutual Information (MI) methods. A predictive model was established using the XGBoost algorithm, and the model's performance was evaluated through 10-fold cross-validation and an independent test set.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe model achieved an area under the curve (AUC) of 0.996 on the training set and 0.975 on the test set, with respective accuracies of 0.981 and 0.926. MPV, DBIL, CK, Na, HDL-C, and LDH were identified as the most predictive indicators.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study confirms that conventional blood parameters combined with machinelearning technology can effectively distinguish melanoma, demonstrating high diagnostic value. This approach provides a feasible, noninvasive strategy for earlymelanoma identification.\u003c/p\u003e","manuscriptTitle":"Establishing a Differential Diagnosis Method for Melanoma Using Blood Parameters Based on Machine Learning Technology","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-26 06:26:57","doi":"10.21203/rs.3.rs-8003880/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f4add9b2-02a9-4839-91fe-52e2d5c52309","owner":[],"postedDate":"November 26th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-01T08:23:33+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-26 06:26:57","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8003880","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8003880","identity":"rs-8003880","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}