A Meaningful Machine Learning Model for Predicting Amputation Rate of Patients with Diabetic Foot

A Meaningful Machine Learning Model for Predicting Amputation Rate of Patients with Diabetic Foot | 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 A Meaningful Machine Learning Model for Predicting Amputation Rate of Patients with Diabetic Foot Zixuan Liu, Siyang Han, Lei Gao, Jiangning Wang, Qi Yao This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5716696/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 Background . Diabetic foot (DF) is a destructive complication of diabetes,which has high amputation rate and causes great social burden. With the development of artificial intelligence, there is an urgent need for an accurate prediction method for the prognosis of chronic diseases. Methods . This retrospective study collected the data of 149 patients with DF from January 2019 to December 2022. The data was divided into the test set and training set by 5-fold cross-validation. Then we used Lasso algorithm to screen relevant risk factors, and the predictive model was built with support vector mechanism(SVM) to forecast probability. Combining multiple methods such as receiver operating characteristic, calibration, and risk decision curve for multimodal analysis of the model. Results. Statistical analysis showed that there were statistically significant differences in CRP, PCT, WBC and Wagner grading between the training and testing sets (P＜0.05). According to the results of multiple factor analysis, CRP and Wagner rating 3-4 were the independent predictive factors for generalized amputation.The area under the receiver operating characteristic(ROC) curves of themodel were 0.89. This model’s calibration capability was 19.614 through Hosmer-Lemeshow test (p=0.012).When the threshold value was selected as 0.302, the constructed model was able to generate a net gain of 0.351. Conclusion . The predictive model we constructed can accurately estimate the rate of amputation during hospitalization in DFU patients. Our survey data suggested that C-reactive protein(CRP) and the Wagner grade of the affected foot might play a vital role in predicting diabetic foot amputation. Diabetic foot Amputation Maching learning Wagner grade Prediction model Intervention treatment Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 1. Introduction Over the past century, the prevalence of diabetes and its related complications has increased rapidly, making it a major cause of morbidity and mortality 1 . Diabetic foot (DF) is one of the most serious complications of diabetes. People with diabetes have a 25 percent lifetime risk of developing foot ulcers, and 14 to 24 percent require severe gangrene or minor lower limb amputation 2 . In addition, it is estimated that the five-year mortality rate after amputation ranges from 39–68%, which is equivalent to the life expectancy of individuals with invasive cancer or severe congestive heart failure 3 . Therefore, early and accurate prediction of amputation has guiding significance for improving the quality of life and survival rate of DF patients. The classification of the severity of diabetic foot infection has been the focus of international diabetic foot scholars 4 . Meggitt-Wagner classification, which is the most widely used rating system for evaluating the development of diabetic foot 5 .The higher the Wagner rating, the greater the likelihood of amputation, and the lower the cure and recovery rate 6 . University of Texas diabetic wound classification is an improvement on the Wagner scale, which associated wound depth with ischemic infection 7 . WiFi classification has been improved on the basis of the earlier classification, integrating ulcer area, ischemia index and foot infection degree 8 . It can assess the severity of diabetic foot patients from a multi-dimensional perspective, which is one of the most widely used classification systems at present. These methods are common tools for clinical diagnosis and long-term prediction of diabetic foot patients, but these are not the gold standard for clinical diagnosis. Moreover, the current classification system does not take into account the basic information of the patient, such as age, gender, medical history, which affects the accuracy and scientific nature of clinical decision-making to a certain extent. Risk prediction model refers to estimating the probability or risk of the existence (diagnostic model) or future occurrence (prognostic model) of a specific disease or condition through mathematical formulas 9 , 11 . Previous investigator used traditional statistical methods (eg. Multiple logistic regression analysis, COX proportional risk model) to predict the risk of amputation in DFU patients 10 . However, due to the diversity and unpredictability of the influencing factors, the prediction range of these methods is limited 12 , 13 . In recent years, with the continuous improvement of the understanding of medical big data and in-depth research on statistical methods, machine learning (ML) algorithm can predict the occurrence and prognosis of diseases 14 – 16 . This provides a new idea for our clinical treatment, and raises the efficiency of prediction.Support vector machine (SVM) is a supervised learning algorithm which can be used to solve regression analysis and binary classification problems. It reduces the error caused by empirical classification and increases the margin, also known as the maximum margin classifier 17 . At present, SVM has been widely used in the medical field, but it is mainly used in the prognosis assessment of cancer patients, and rarely used in the field of survival analysis, especially in the field of chronic diseases 18 – 20 . In summary, we intend to develop a machine learning model based on SVM algorithm that can predict the amputation rate of diabetic foot ulcers. In addition, we try to integrate clinical diagnosis and treatment with long-term prognosis decision-making, providing scientific guidance for clinical decision-making and nursing work of diabetic foot ulcer. 2. Material and methods 2.1 Inclusion and exclusion criteria for subjects We randomly selected 149 diabetic foot patients admitted to the department of Orthopedics Surgery of Beijing Shijitan Hospital Affiliated to Capital Medical University from January 2019 to December 2022 as the study subjects to carry out this retrospective cohort study. The study included patients who met the following criteria : a) Admitted to hospital with a diagnosis that met DFU's clinical diagnostic criteria b) Diabetic foot patients above Wagner level 1 c) Routine laboratory tests and auxiliary examinations have been completed after admission d) Surgical treatment has been performed during the visit e) The number of hospitalizations of the patient within the investigation range ≤ 2 times f) This study protocol was known to the patient, and the patient himself was informed and consented. We also excluded subjects according to the following criteria: Amputation cases with low nutritional indexes (albumin ≤ 30g/L). Amputation cases with severe insufficiency of lower limb blood supply (B-ultrasonography showed vascular stenosis ≥ 75%). Patients with other infectious diseases. Patients with malignant tumors. Patients younger than 18 years Patients transferred to other healthcare facilities during treatment The prognosis of the patients was split into two types according to the surgical method : (a) Amputation group; (b) Non-amputee group. According to the plane of amputation, amputation can be divided into minor amputation, which are considered to be below the ankle amputation, and severe amputation that above the ankle amputation. 2.2 Subject inclusion index We collected the basic information of 149 patients (including hospitalization number, number of admissions, date of hospitalization, gender, age), clinical data (eg. history of hyperlipidemia, Wagner grade of affected foot, wound location, prognosis) and auxiliary examination (C-reactive protein, procalcitonin, white blood cell, albumin, and degree of arteriosclerosis of lower extremity) as electronic medical records. The units of laboratory tests collected for each patient were the same, C-reactive protein was mg/L, procalcitonin was ng/mL, white blood cell was 10 9 /L, albumin was g/L, and blood vessel occlusion degree of lower limbs was referred to the results of color Doppler examination of both lower limbs. 2.3 Data preprocessing In order to ensure the accuracy and scientificity of input variables and reduce systematic errors, we first fit the prediction model. The Lasso(Least Absolute Shrinkage and Selection Operator)algorithm can initially screen the predictors and obtain a model with good performance and simplicity. We take the lambda value corresponding to the cross-validation error within 1 standard deviation of the minimum error as the optimal penalty coefficient of the model. Then, multivariate logistic regression analysis was carried out on the variables selected by LASSO regression using backward likelihood method to determine the final predictors and construct the nomogram model. 2.4 Statistical analyses Descriptive statistical analysis of the data was performed for each of the two groups separately.Besides, continuous variables were expressed as mean ± standard deviation.For the variables analyzed by the Kolmogorov-Smirnov normality test, we compared the significance of differences in qualitative features between groups using the Wilson rank sum test and Chi-square test, respectively. In addition, categorical variables are expressed as counts (n) and percentages (%). A P value < 0.05 was considered statistically significant.All the above calculations were performed using SPSS 21.0. 2.5 Model development This paper uses machine learning as the analysis method of prediction model, which includes input variables, establishment of mapping relationship and output function results. We screened patient data by inclusion and exclusion measures and used them as input variables to the model. However, due to too many related factors, it is necessary to use the lasso algorithm to preprocess the included data and conduct correlation analysis. In terms of modeling, the data is divided into training sets and test sets by using 5-fold cross-validation. Then LDA, K neighborhood, SVM and other mature models were constructed to predict condition of diabetic foot amputation. We used MATLAB2019A to analyze the data. After modeling, we drawed the corresponding confusion matrix to estimate the effectiveness of predictive model, and selected the model with the best prediction ability and the corresponding function. 2.6 Model evaluation Three evaluation measures were used to evaluate the performance of the model in each category: area under the subject operating characteristic curve (AUC), sensitivity, and specificity. The model calibration curve is drawn to observe the fitting results between the ideal model and the real results. Decision analysis curve can reflect the impact of the predictive model on the patient's net benefit.The nomogram of the prediction model contributes to analyze the risk weights of individual risk factors. The flowchart of our experiment is shown in Fig. 1 . 3. Results 3.1 Statistical test result As shown in Table 1,we used t test to analyze continuous variables and Chi-square test to analyze categorical variables. Among them, variables CRP, PCT and WBC of training set and test set did not conform to normality, so U test was used for analysis. According to the analysis results, in the training set, there were statistically significant differences between CRP, PCT, WBC and Wagner_rating in terms of generalized amputation scores (P＜0.05) β Odds Ratio(95%CI) P （Intercept） -5.1546 Angiosclerosis1 -17.8435 ＜0.001 0.994 Angiosclerosis2 0.3903 1.477[0.119,18.306] 0.761 CRP 0.0207 1.021[1.010,1.032] ＜0.001 Degree of occlusion 50 -16.5541 ＜0.001 0.994 Degree of occlusion 75 1.1200 3.065[0.367,25.597] 0.301 Degree of occlusion 99 -0.0199 0.980[0.277,3.473] 0.975 Wagner rating3 2.7226 15.220[1.535,150.867] 0.020 Wagner rating4 4.3520 77.630[8.152,739.265] ＜0.001 TABLE 2 Multivariate analysis of infection index in diabetic foot amputation patients Multivariate analysis results (Table 2) showed that CRP, Wagner_rating3 and Wagner_rating4 were independent predictors of generalized amputation (all P＜0.05). The degree of vascular sclerosis and vascular occlusion in one or both lower limbs did not affect the prognosis of amputation. 3.2 Determine input variable The noose coefficient contours for 10 texture features was shown in Figure 2. The horizontal coordinate was the penalty coefficient, and vertical coordinate was the gene coefficient. A coefficient profile was plotted against the log(λ) sequence, and vertical line was plotted at selected values that are cross-validated by a factor of 10, where the optimal λ yields four non-zero coefficients. This was basically consistent with the results of statistical analysis. LASSO regression was used to screen the relevant risk factors,and then we calculated the corresponding regression coefficients to improve the robustness of the model 21 . Figure 3 shown the cross verification curve of LASSO regression analysis, where the horizontal coordinate was the penalty coefficient and vertical coordinate represents the cross verification error. The smaller the value of the vertical axis, the better the LASSO fitting effect. Using the minimum criterion and one standard error of the minimum criterion (1-se criterion), the vertical dotted line was drawn at the optimal value. After 10-fold cross-verification, the λ value was 0.0276, and the log(λ) was 0.122(1-SE criterion). Meanwhile, the upper horizontal coordinate corresponding to this point was the number of variables that could be used for analysis. 3.3 Model performance Figure 4 was a confusion matrix, plotted by a predictive model built from the data set.The formulation of confusion matrix made the prediction result more intuitive and convenient for clinical workers to make judgment and decision 22 . As shown in the figure 3, we observed that in the green part, the predicted results were consistent with the actual amputation results, while in the pink part, the two were completely opposite.We roughly estimated that the prediction accuracy of the model is about 81.2% through the confusion matrix, which basically conforms to the prediction.The true positive rates of predicted amputation and non-amputation, respectively were 79% and 83%. We tried to use SVM method to build a prediction model, run the prediction model according to the customized function, calculate the accuracy of the model, and select the function with the highest accuracy as the modeling decision tree. The prediction accuracy of our function was up to 82.4%, which was basically in line with expectations. This proved that the prediction model we constructed had a certain guiding effect on clinical work. 3.4 Model evaluation Figure 5 was plotted based on whether generalized amputation or not, where the Area Under Curve (AUC) is 0.89. The closer the curve area is to 1, the higher the accuracy of the proof model. The maximum approximate entry index (Max(sensitivity + specificity -1)) under the ROC curve was supreme when the tangent point was 0.21, corresponding to 0.83. The calibration ability of the model was 19.614(p=0.012) through Hosmer-Lemeshow test. Figure 6(a) was the calibration curve of the model on the training set, Figure 6(b) was the calibration curve of the model on the test set. Calibration curves were drawn based on the agreement between the observed predicted risk of amputation and the actual results. The Y-axis represented the actual amputation outcome, and the X-axis represented the predicted risk of amputation. The gray diagonal line represented an ideal perfect model, while the solid black line represents the performance of the model, the closer the two lines were, the better the fit. 3.5 Multimodal analysis The nomogram of the prediction model based on whether generalized amputation or not was shown in Figure 7, integrating all independent predictors. FIGURE 8 shown the net benefit of the model to the patient as the threshold selection changes. When the threshold value was selected as 0.302 derived from the Jorden index, the model was able to generate a net gain of 0.351. 4. Discussion In this study, we developed an amputation prediction model that incorporates 10 baseline features to predict the probability of amputation in DF patients. The AUC and calibration capability of the prediction model were 0.89 and 19.614, respectively. At the same time, according to the decision analysis curve, when the threshold is 0.302, the net benefit is 0.351. This shows that the prediction model has a good clinical application prospect, and that the predicted probability of the model is in reliable agreement with the actual probability. Numerous studies have shown that the more severe the infection, the higher the amputation rate in DF patients 23 – 24 . In the prediction model of amputation prognosis of DF patients, a machine learning model with infection severity, lower limb blood supply and systemic nutrition as input variables was established. We found that in the patient cases we collected, the prognosis of patients with amputation was affected by multiple factors, such as severe infection, lower limb vascular occlusion, poor nutritional status, etc., which was basically consistent with clinical experience and related studies 25 – 26 . In addition, it is worth noting that hyperlipidemia also increases the risk of amputation to some extent. The results are meaningful and generally in line with our expectations. The Wagner classification system is widely used in clinical practice to assess the severity of foot in diabetic patients. It categorizes foot ulcers into six grades 27 . The classification of wounds is based on depth, extent and degree of infection, which are just what we need to observe. This additional information helps to improve the accuracy of the model in predicting the risk of amputation in DFU patients. We hope to build a prediction model for amputation risk and further improve the DFU classification system from multiple aspects 28 – 30 . Overall, incorporating the Wagner classification system into the predictive model allows for a more comprehensive assessment of the risk of amputation in DFU patients.It enhances the predictive power of the model and provides valuable clinical information for personalized treatment and management of diabetic foot disease. In addition, simple linear models (COX regression, multiple Logistic regression, COX proportional risk model) have certain limitations in assessing the prognosis of DFU patients due to the diversity and unpredictability of influencing factors 31 . Boyko et al. 32 first carried out a prospective cohort study on diabetic patients in 2006, applied Cox proportional risk model to screen independent influencing factors, and finally formed a scoring system model. The AUC of this model was 0.81, indicating good differentiation. But the study was not externally verified. In 2019, British scholar Heald 33 et al. analyzed diabetic patients through retrospective cohort study and Logistic multiple regression, finally included 5 risk factors, and built a calculation equation for the risk probability of diabetic foot based on the regression coefficients of each factor. The model AUC was 0.65. Although the model has certain clinical practicability, the accuracy rate is still low. The emergence of artificial intelligence provides new ideas and methods for diabetes risk prediction. In 2021, Peng et al. 34 constructed a model to predict the risk of diabetic foot amputation, and constructed the nomogram to visually compare the risk weights of each risk factor. The model AUC was 0.876. Deng et al. 35 used XGBoost algorithm and COX regression to evaluate the impact of hyperglycemic crisis and other risk factors on the mortality of DFU patients. The model’s AUC is 0.680. A prospective study by Lv et al. 36 established a DFU risk model based on risk factors and presented it in the form of the nomogram and web calculator. The AUC of its model was 0.741. Therefore, our study used ANN to establish the model. Machine learning algorithms adopt a multivariate, non-parametric approach that can use non-normal distributions and strongly correlated data to build robust models and identify complex patterns 37 – 38 . Compared with the statistical methods in previous relevant studies, the prediction model we constructed includes more relevant factors, has higher accuracy and more intuitive prediction results.We believe that our model is convincing for the guidance of clinical work. The combination of model prediction and clinical decision making is another feature of our experiment. In this study, we intend to construct a database of diabetic foot - diabetic foot amputation patients, study the real world of diabetic foot amputation patients, and validate predictive models to guide clinical decision making. Obviously, whether the predictive model based on retrospective data analysis can be applied to clinical practice needs to be further verified. Therefore, we hope to collect the information of newly included diabetic foot patients, compared the model prediction results, optimized the internal algorithm of the model, further screened the relevant risk factors, and understood the distribution of diabetic foot characteristics in the current survey population. Based on the relevant risk factors we have screened, we should proceed with the multi-disciplinary combined diagnosis and treatment of diabetic foot patients from the four aspects of nutrition, blood circulation, wound surface and infection. It is essential for the early prevention, timely intervention and scientific individualized treatment of DFU. 40 However, there are some limitations to our study. Although we use SVM algorithms to build models, we still lack sufficient clinical validation and cohort studies.Although we use SVM algorithms to build models, we still lack sufficient clinical validation and cohort studies. It is true that machine learning models primarily establish mapping relationships between input variables and predicted outcomes, rather than capturing direct causal relationships. Identifying causal relationships in complex medical conditions such as foot ulcers is challenging. Machine learning models can provide insights into the associations between patient characteristics and predicted outcomes, but further investigation and studies are needed to establish causality. It is essential to consider the timing and effectiveness of treatment interventions when interpreting the predictions made by the model. The fluctuation of the infection index of a single patient's second admission is influenced by the treatment measures. At the same time, we can not ignore that the decision of diabetic foot amputation surgery is affected by many aspects, such as the patient's economic status, wound status, and the subjective judgment of clinicians, and it is difficult to predict whether amputation is possible only through observational indicators. We want to develop a model that differ from the existing DFU taxonomy system and can be optimized and refined on the basis of the Wagner classification. In this way, our research constructs highly accurate and practical models to predict amputation rates in DFU patients and attempts to combine clinical care with long-term prognosis. 5. Conclusion In summary, we built an intelligent model which can be used to forecast the risk of inpatient amputation in DF patients and analyzed the real world of newly enrolled diabetic foot patients. Our experimental results show that the machine learning model not only has accurate predictive power, but also provides new ideas for the formulation of personalized treatment plans for patients. Declarations Conflict of interest: The authors declare no competing interests. Ethics approval: The study was notifed to the Ethics Committee of Beijing Shijitan Hospital Affiliated to Capital Medical University and it was waived from the formal approval due to its retrospective study design. Data availability: The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding author. Author contribution: Z.L and S. H contributed equally to this work, they are co-first authors. The frst draft of the manuscript was written by Z.L. S.H was involved in chart production. Q.Y., J.W. and L.G are the co-corresponding authors of this manuscript. Q.Y participated in putting forward constructive suggestions for the follow-up of the research. J. W. was responsible for guiding the construction of the model. L.G. was in charge of providing clinical cases. Funding information: This study was not supported by any specific grants from funding agencies in the public, commercial, or not-for-profit sectors. References Ibrahim, A. IDF Clinical Practice Recommendation on the Diabetic Foot: A Guide for Healthcare Professionals. Diabetes Res. Clin.Pract. 2017, 127, 285–287. Global disability burdens of diabetes-related lower-extremity complications in 1990 and 2016. Zhang Y, Lazzarini PA, McPhail SM, et al [J]. Diabetes Care. 2020;43:964–974. Volmer-Thole, M.; Lobmann, R. Neuropathy and Diabetic Foot Syndrome. Int. J. Mol. Sci. 2016, 17, 917. Monteiro-Soares M, Boyko EJ, Jeffcoate W, et al. Diabetic foot ulcer classifications: a critical review. Diabetes/Metab Res Rev. 2020;36:e3272. Diabetic podiatry Group, Peripheral Vascular Diseases Committee of Chinese Microcirculation Society. Expert Consensus on diabetic foot wound repair and treatment [J]. Chinese Journal of Diabetes, 2018, 10(5):305-309. Chinese Medical Association Diabetes Society, Chinese Medical Association Infectious Diseases Society, Chinese Medical Association Tissue Repair and Regeneration Society. Chinese Guidelines for the Prevention and Treatment of diabetic foot (2019 edition) (Ⅳ). Chinese Journal of Diabetes, 2019, 11(5):316-327. Li Zhihong, Guo Shuqin, Li Tingting, et al. Comparison of Wagner grading method for diabetic foot and University of TEXAS classification [J]. Chinese Journal of Diabetes, 2012,4 (8) : 469-73. International Diabetic Foot Working Group: Classification of diabetic foot ulcers Part of the International Diabetic Foot Working Group: International Guidelines for Diabetic Foot Management (2019) [J]. Infection, Inflammation and Repair, 2019, 20(4):231-237. Xie Xiaoran, Xu Rong. A systematic evaluation of risk prediction model for diabetic foot [J]. Chinese Journal of Nursing, 21, 56 (1) : 124-131 Heald A,Lunt M,Rutter M K,et al. Developing a foot ulcer risk model:what is needed to do this in a real-world primary care setting? [J]. Diabet Med,2019,36(11):1412-1416. Moons K G,Altman D G,Reitsma J B,et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis(TRIPOD):explanation and elaboration [J]. Ann Intern Med,2015,162(1):W1-73. Basu A, Manning WG, Mullahy J. Comparing alternative models: log vs cox proportional hazard? [J]. Health Econ. 2004;13(8):749-765. Wright RE. Logistic regression. Reading and Understanding Multivariate Statistics [J]. Washington DC: American Psychological Association; 1995:217-244. Nanda R, Nath A, Patel S, et al. Machine learning algorithm to evaluate risk factors of diabetic foot ulcers and its severity [J]. Med Biol Eng Comput, 2022, 60(8): 2349-2357. Zale A, Mathioudakis N. Machine learning models for inpatient glucose prediction[J]. Curr Diab Rep, 2022, 22(8): 353-364. Deng L, Xie P, Chen Y, et al. Impact of acute hyperglycemic crisis episode on survival in individuals with diabetic foot ulcer using a machine learning approach [J]. Front Endocrinol (Lausanne), 2022, 13: 974063. Brereton RG, Lloyd GR. Support vector machines for classification and regression [J]. Analyst, 2010, 135(2): 230-267. Ansari D，Nilsson J，Andersson R，et al. Artificial neural networks predict survival from pancreatic cancer after radical surgery [J]. Am J Surg，2013，205（1）：1-7. Amiri Z，Mohammad K，Mahmoudi M，et al. Assessing the effect of quantitative and qualitative predictors on gastric cancer individuals survival using hierarchical artifcial neural network models [J]. Iran Red Cresc Med J，2013，15（1）：42-48. Qiao G，Li J，Huang A，et al. Artificial neural networking model for the prediction of post -hepatectomy survival of patients with early hepatocellular carcinoma [J]. J Gastroenterol Hepatol，2014，29（12）：2014-2020. Zhang Lijin, Wei Xiayan, Lu Jiaqi, et al. Lasso regression: From interpretation to prediction [J]. Advances in Psychological Science, 2019,28(10) :1777-1788. Valero-Carreras Daniel, Alcaraz Javier, Landete Mercedes, et al. Comparing two SVM models through different metrics based on the confusion matrix [J]. Computers and Operations Research. Volume 152 , Issue . 2023 Musa IR，Ahmed MON，Sabir EI，et al. Factors associated with amputation among patients with diabetic foot ulcers in a Saudi population[J]. BMC Research Notes，2018，11（1）：260. Hu Huiping, Chen Mingwei. Analysis of related factors of diabetic foot ulcer healing and amputation [J]. Journal of Anhui Medical University, 2016,51 (11) : 1634-1637. Sun J-H, Tsai J-S, Huang C-H, et al. Risk factors for lower extremity amputation in diabetic foot disease categorized by Wagner classification[J]. Diabetes Res Clin Pract. 2012;95(3):358-363. Zhan LX, Branco BC, Armstrong DG, Mills JL Sr. The Society for Vascular Surgery lower extremity threatened limb classification system based on wound, ischemia, and foot infection(WIfI) correlates with risk of major amputation and time to wound healing[J]. J Vasc Surg. 2015;61(4):939-944. Mehraj M, Shah I. A review of Wagner classification and current concepts in management of diabetic foot. Int J Orthop Sci.2018;4(1):933-935. Lu S, Chen R, Wei W, Lu X. Understanding heart-failure patients EHR clinical features via SHAP interpretation of tree-based machine learning model predictions. arXiv preprint arXiv:210311254; 2021. Jeon BJ, Choi HJ, Kang JS, Tak MS, Park ES. Comparison of five systems of classification of diabetic foot ulcers and predictive factors for amputation[J]. Int Wound J. 2017;14(3):537-545. Santema TB, Lenselink EA, Balm R, Ubbink DT. Comparing the Meggitt-Wagner and the University of Texas wound classification systems for diabetic foot ulcers: inter-observer analyses[J].Int Wound J. 2016;13(6):1137-1141. Beckert S, Witte M, Wicke C, Königsrainer A, Coerper S. A new wound-based severity score for diabetic foot ulcers: a prospective analysis of 1,000 patients [J]. Diabetes Care. 2006;29(5):988-992. BOYKO E J,AHRONI J H,COHEN V,et al. Prediction of diabetic foot ulcer occurrence using commonly available clinical information:the seattle diabetic foot study [J]. Diabetes Care,2006,29(6):1202-1207. HEALD A,LUNT M,RUTTER M K,et al. Developing a foot ulcer risk model:what is needed to do this in a real-world primary care setting? [J]. Diabet Med, 2019,36(11):1412-1416. Bo P, Rui M, Yi L,et al. Development of Predictive Nomograms for Clinical Use to Quantify the Risk of Amputation in Patients with Diabetic Foot Ulcer[J]. Diabetes Res.2021,6621035(2021). Deng L, Xie P, Chen Y, et al. Impact of acute hyperglycemic crisis episode on survival in individuals with diabetic foot ulcer using a machine learning approach [J]. Front Endocrinol (Lausanne), 2022, 13: 974063. Lv, J. et al. Development and validation of a risk prediction model for foot ulcers in diabetic patients.[J]. Diabetes Res. 2023, 1199885 (2023). Ripatti S,Tikkanen E,Orho-Melander M,et al. A multilocus genetic risk score for coronary heart disease：Case -control and prospective cohort analyses［J］. The Lancet，2010，376（9750）：1393-1400. Silver M, Chen P, Li R,et al. Pathways-driven sparse regression identifies pathways and genes associated with high - density lipoprotein cholesterol in two asiancohorts［J］. PLoS Genetics，2013，9（11）：e1003939 Table 1 Table 1 is available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Table1.docx 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-5716696","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":396062276,"identity":"d45d7704-00b4-4ece-ade6-40bde1b78323","order_by":0,"name":"Zixuan Liu","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zixuan","middleName":"","lastName":"Liu","suffix":""},{"id":396062278,"identity":"33576871-3b25-481c-8459-4e57aa19500f","order_by":1,"name":"Siyang Han","email":"","orcid":"","institution":"Beijing Shijitan Hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Siyang","middleName":"","lastName":"Han","suffix":""},{"id":396062280,"identity":"80dc5f97-5e4e-4a17-9508-5c2bbf187f74","order_by":2,"name":"Lei Gao","email":"","orcid":"","institution":"Beijing Shijitan Hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Gao","suffix":""},{"id":396062286,"identity":"dc6ec519-acda-47e3-8124-4414108d6619","order_by":3,"name":"Jiangning Wang","email":"","orcid":"","institution":"Beijing Shijitan Hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiangning","middleName":"","lastName":"Wang","suffix":""},{"id":396062287,"identity":"cd171b44-8c07-4f5a-be0b-8ebf84f3722e","order_by":4,"name":"Qi Yao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwUlEQVRIiWNgGAWjYHCCNIYPBhI8/MzMhx8QrYVxRoWFnGQ7W5oBsVrYmHnOVBgbnOdRkCBKvcHxhGcPeNskEjcf5mEwYKixiSas5cyDdANJoJZth3kPPGA4lpbbQFDLjYQ0CUOwFr4EA8aGw0RqSQQ5rJnHQIJ4LQfOSBgbMBOrRfLMgzTJhgoJOYnDwEBOIMYvfMdz0qT/GNTx8PcfPvzgQ40NYS0KB3ISELwEXMqQgXxD+gFi1I2CUTAKRsFIBgCr+kIMb+ztHwAAAABJRU5ErkJggg==","orcid":"","institution":"Beijing Shijitan Hospital Affiliated to Capital Medical University","correspondingAuthor":true,"prefix":"","firstName":"Qi","middleName":"","lastName":"Yao","suffix":""}],"badges":[],"createdAt":"2024-12-26 14:23:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5716696/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5716696/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":72739932,"identity":"5cf200d0-c098-48f3-9c83-6c49e86ce20d","added_by":"auto","created_at":"2025-01-01 09:27:17","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":44598,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart of our experiment\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/6ea750981b9892c62f1bbbe2.png"},{"id":72739945,"identity":"7232c282-c23b-4b0a-b664-29c4239d1828","added_by":"auto","created_at":"2025-01-01 09:27:17","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":102073,"visible":true,"origin":"","legend":"\u003cp\u003eLASSO shrinkage coefficient path diagram\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/bf9de21008c471bde684dcc6.png"},{"id":72739935,"identity":"170e51ea-cdde-469a-8b9a-4582979a7c43","added_by":"auto","created_at":"2025-01-01 09:27:17","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":90635,"visible":true,"origin":"","legend":"\u003cp\u003eLASSO regression analysis cross-validation curve\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/caf228b1d5c8bef8cbe2f615.png"},{"id":72740810,"identity":"46b209cd-f1b4-424e-8b33-b7dcce2d5d46","added_by":"auto","created_at":"2025-01-01 09:35:17","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":27406,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion matrix drawn from prediction model\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/98e9b7ff7986dcd95a965521.png"},{"id":72739938,"identity":"199a2c24-b7c7-488f-9f48-15512f5180b3","added_by":"auto","created_at":"2025-01-01 09:27:17","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":21451,"visible":true,"origin":"","legend":"\u003cp\u003eAccuracy and calibration performance of the predictable model\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/90c80687394459c662753754.png"},{"id":72739954,"identity":"d1ac1de7-cd1e-4433-b779-62f18a02155d","added_by":"auto","created_at":"2025-01-01 09:27:18","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":48788,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curve of amputation prediction model\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/cbaa3f307d833bbbb0a2986d.png"},{"id":72739937,"identity":"92bf1f70-ec54-45bc-8ec0-a0b54c926012","added_by":"auto","created_at":"2025-01-01 09:27:17","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":33964,"visible":true,"origin":"","legend":"\u003cp\u003eA nomogram of the amputation prediction model\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/02f4d16dbcba358ae74a44cb.png"},{"id":72739940,"identity":"63354880-5c16-4f40-857d-884ec444d78c","added_by":"auto","created_at":"2025-01-01 09:27:17","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":32683,"visible":true,"origin":"","legend":"\u003cp\u003eDecision analysis curve of diabetic foot amputation\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/88c0c7c27a902c6dc06b0a52.png"},{"id":77664243,"identity":"97df158d-bf2a-4a93-bd4b-5ceef94cc93a","added_by":"auto","created_at":"2025-03-04 05:34:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":867991,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/b5018149-4f72-457e-9dff-dfaa4813475b.pdf"},{"id":72740812,"identity":"afdf6d6c-5463-46d0-8905-1310d0dbe332","added_by":"auto","created_at":"2025-01-01 09:35:17","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":21819,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.docx","url":"https://assets-eu.researchsquare.com/files/rs-5716696/v1/764ecfea416c1a7f530be0d5.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Meaningful Machine Learning Model for Predicting Amputation Rate of Patients with Diabetic Foot","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eOver the past century, the prevalence of diabetes and its related complications has increased rapidly, making it a major cause of morbidity and mortality\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Diabetic foot (DF) is one of the most serious complications of diabetes. People with diabetes have a 25 percent lifetime risk of developing foot ulcers, and 14 to 24 percent require severe gangrene or minor lower limb amputation\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. In addition, it is estimated that the five-year mortality rate after amputation ranges from 39\u0026ndash;68%, which is equivalent to the life expectancy of individuals with invasive cancer or severe congestive heart failure\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Therefore, early and accurate prediction of amputation has guiding significance for improving the quality of life and survival rate of DF patients.\u003c/p\u003e \u003cp\u003eThe classification of the severity of diabetic foot infection has been the focus of international diabetic foot scholars\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Meggitt-Wagner classification, which is the most widely used rating system for evaluating the development of diabetic foot\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e.The higher the Wagner rating, the greater the likelihood of amputation, and the lower the cure and recovery rate\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. University of Texas diabetic wound classification is an improvement on the Wagner scale, which associated wound depth with ischemic infection\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. WiFi classification has been improved on the basis of the earlier classification, integrating ulcer area, ischemia index and foot infection degree\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. It can assess the severity of diabetic foot patients from a multi-dimensional perspective, which is one of the most widely used classification systems at present. These methods are common tools for clinical diagnosis and long-term prediction of diabetic foot patients, but these are not the gold standard for clinical diagnosis. Moreover, the current classification system does not take into account the basic information of the patient, such as age, gender, medical history, which affects the accuracy and scientific nature of clinical decision-making to a certain extent.\u003c/p\u003e \u003cp\u003eRisk prediction model refers to estimating the probability or risk of the existence (diagnostic model) or future occurrence (prognostic model) of a specific disease or condition through mathematical formulas\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. Previous investigator used traditional statistical methods (eg. Multiple logistic regression analysis, COX proportional risk model) to predict the risk of amputation in DFU patients\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. However, due to the diversity and unpredictability of the influencing factors, the prediction range of these methods is limited\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. In recent years, with the continuous improvement of the understanding of medical big data and in-depth research on statistical methods, machine learning (ML) algorithm can predict the occurrence and prognosis of diseases\u003csup\u003e\u003cspan additionalcitationids=\"CR15\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. This provides a new idea for our clinical treatment, and raises the efficiency of prediction.Support vector machine (SVM) is a supervised learning algorithm which can be used to solve regression analysis and binary classification problems. It reduces the error caused by empirical classification and increases the margin, also known as the maximum margin classifier\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. At present, SVM has been widely used in the medical field, but it is mainly used in the prognosis assessment of cancer patients, and rarely used in the field of survival analysis, especially in the field of chronic diseases\u003csup\u003e\u003cspan additionalcitationids=\"CR19\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn summary, we intend to develop a machine learning model based on SVM algorithm that can predict the amputation rate of diabetic foot ulcers. In addition, we try to integrate clinical diagnosis and treatment with long-term prognosis decision-making, providing scientific guidance for clinical decision-making and nursing work of diabetic foot ulcer.\u003c/p\u003e"},{"header":"2. Material and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Inclusion and exclusion criteria for subjects\u003c/h2\u003e \u003cp\u003eWe randomly selected 149 diabetic foot patients admitted to the department of Orthopedics Surgery of Beijing Shijitan Hospital Affiliated to Capital Medical University from January 2019 to December 2022 as the study subjects to carry out this retrospective cohort study.\u003c/p\u003e \u003cp\u003eThe study included patients who met the following criteria :\u003c/p\u003e \u003cp\u003ea) Admitted to hospital with a diagnosis that met DFU's clinical diagnostic criteria\u003c/p\u003e \u003cp\u003eb) Diabetic foot patients above Wagner level 1\u003c/p\u003e \u003cp\u003ec) Routine laboratory tests and auxiliary examinations have been completed after admission\u003c/p\u003e \u003cp\u003ed) Surgical treatment has been performed during the visit\u003c/p\u003e \u003cp\u003ee) The number of hospitalizations of the patient within the investigation range\u0026thinsp;\u0026le;\u0026thinsp;2 times\u003c/p\u003e \u003cp\u003ef) This study protocol was known to the patient, and the patient himself was informed and consented.\u003c/p\u003e \u003cp\u003eWe also excluded subjects according to the following criteria:\u003c/p\u003e \u003cp\u003e \u003col style=\"list-style-type:lower-alpha;\"\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eAmputation cases with low nutritional indexes (albumin\u0026thinsp;\u0026le;\u0026thinsp;30g/L).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eAmputation cases with severe insufficiency of lower limb blood supply (B-ultrasonography showed vascular stenosis\u0026thinsp;\u0026ge;\u0026thinsp;75%).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePatients with other infectious diseases.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePatients with malignant tumors.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePatients younger than 18 years\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePatients transferred to other healthcare facilities during treatment\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eThe prognosis of the patients was split into two types according to the surgical method : (a) Amputation group; (b) Non-amputee group. According to the plane of amputation, amputation can be divided into minor amputation, which are considered to be below the ankle amputation, and severe amputation that above the ankle amputation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Subject inclusion index\u003c/h2\u003e \u003cp\u003eWe collected the basic information of 149 patients (including hospitalization number, number of admissions, date of hospitalization, gender, age), clinical data (eg. history of hyperlipidemia, Wagner grade of affected foot, wound location, prognosis) and auxiliary examination (C-reactive protein, procalcitonin, white blood cell, albumin, and degree of arteriosclerosis of lower extremity) as electronic medical records.\u003c/p\u003e \u003cp\u003eThe units of laboratory tests collected for each patient were the same, C-reactive protein was mg/L, procalcitonin was ng/mL, white blood cell was 10\u003csup\u003e9\u003c/sup\u003e/L, albumin was g/L, and blood vessel occlusion degree of lower limbs was referred to the results of color Doppler examination of both lower limbs.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Data preprocessing\u003c/h2\u003e \u003cp\u003eIn order to ensure the accuracy and scientificity of input variables and reduce systematic errors, we first fit the prediction model. The Lasso(Least Absolute Shrinkage and Selection Operator)algorithm can initially screen the predictors and obtain a model with good performance and simplicity. We take the lambda value corresponding to the cross-validation error within 1 standard deviation of the minimum error as the optimal penalty coefficient of the model. Then, multivariate logistic regression analysis was carried out on the variables selected by LASSO regression using backward likelihood method to determine the final predictors and construct the nomogram model.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Statistical analyses\u003c/h2\u003e \u003cp\u003eDescriptive statistical analysis of the data was performed for each of the two groups separately.Besides, continuous variables were expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation.For the variables analyzed by the Kolmogorov-Smirnov normality test, we compared the significance of differences in qualitative features between groups using the Wilson rank sum test and Chi-square test, respectively. In addition, categorical variables are expressed as counts (n) and percentages (%). A P value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant.All the above calculations were performed using SPSS 21.0.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Model development\u003c/h2\u003e \u003cp\u003eThis paper uses machine learning as the analysis method of prediction model, which includes input variables, establishment of mapping relationship and output function results. We screened patient data by inclusion and exclusion measures and used them as input variables to the model. However, due to too many related factors, it is necessary to use the lasso algorithm to preprocess the included data and conduct correlation analysis. In terms of modeling, the data is divided into training sets and test sets by using 5-fold cross-validation. Then LDA, K neighborhood, SVM and other mature models were constructed to predict condition of diabetic foot amputation. We used MATLAB2019A to analyze the data. After modeling, we drawed the corresponding confusion matrix to estimate the effectiveness of predictive model, and selected the model with the best prediction ability and the corresponding function.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6 Model evaluation\u003c/h2\u003e \u003cp\u003eThree evaluation measures were used to evaluate the performance of the model in each category: area under the subject operating characteristic curve (AUC), sensitivity, and specificity. The model calibration curve is drawn to observe the fitting results between the ideal model and the real results. Decision analysis curve can reflect the impact of the predictive model on the patient's net benefit.The nomogram of the prediction model contributes to analyze the risk weights of individual risk factors.\u003c/p\u003e \u003cp\u003eThe flowchart of our experiment is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e "},{"header":"3. Results","content":"\u003cp\u003e\u003cem\u003e3.1 Statistical test result\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAs shown in Table 1,we used t test to analyze continuous variables and Chi-square test to analyze categorical variables. Among them, variables CRP, PCT and WBC of training set and test set did not conform to normality, so U test was used for analysis. According to the analysis results, in the training set, there were statistically significant differences between CRP, PCT, WBC and Wagner_rating in terms of generalized amputation scores (P＜0.05)\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e\u0026beta;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003eOdds Ratio(95%CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003e（Intercept）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e-5.1546\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 280px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eAngiosclerosis1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e-17.8435\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;0.994\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eAngiosclerosis2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e0.3903\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e1.477[0.119,18.306]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e0.761\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eCRP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e0.0207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e1.021[1.010,1.032]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eDegree of occlusion 50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e-16.5541\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e0.994\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eDegree of occlusion 75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e1.1200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e3.065[0.367,25.597]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e0.301\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eDegree of occlusion 99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e-0.0199\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e0.980[0.277,3.473]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e0.975\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eWagner rating3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e2.7226\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e15.220[1.535,150.867]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e0.020\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 164px;\"\u003e\n \u003cp\u003eWagner rating4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e4.3520\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 162px;\"\u003e\n \u003cp\u003e77.630[8.152,739.265]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 119px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTABLE 2 \u0026nbsp;Multivariate analysis of infection index in diabetic foot amputation patients\u003c/p\u003e\n\u003cp\u003eMultivariate analysis results (Table 2) showed that CRP, Wagner_rating3 and Wagner_rating4 were independent predictors of generalized amputation (all P＜0.05). The degree of vascular sclerosis and vascular occlusion in one or both lower limbs did not affect the prognosis of amputation.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e3.2 Determine input variable\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eThe noose coefficient contours for 10 texture features was shown in Figure 2. The horizontal coordinate was the penalty coefficient, and vertical coordinate was the gene coefficient. A coefficient profile was plotted against the log(\u0026lambda;) sequence, and vertical line was plotted at selected values that are cross-validated by a factor of 10, where the optimal\u0026nbsp;\u0026lambda;\u0026nbsp;yields four non-zero coefficients. This was basically consistent with the results of statistical analysis. LASSO regression was used to screen the relevant risk factors,and then we calculated the corresponding regression coefficients to improve the robustness of the model\u003csup\u003e21\u003c/sup\u003e.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFigure 3 shown the cross verification curve of LASSO regression analysis, where the horizontal coordinate was the penalty coefficient and vertical coordinate represents the cross verification error. The smaller the value of the vertical axis, the better the LASSO fitting effect.\u003csup\u003e\u0026nbsp;\u003c/sup\u003eUsing the minimum criterion and one standard error of the minimum criterion (1-se criterion), the vertical dotted line was drawn at the optimal value. After 10-fold cross-verification, the \u0026lambda; value was 0.0276, and the log(\u0026lambda;) was 0.122(1-SE criterion). Meanwhile, the upper horizontal coordinate corresponding to this point was the number of variables that could be used for analysis.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e3.3 Model performance\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFigure 4 was a confusion matrix, plotted by a predictive model built from the data set.The formulation of confusion matrix made the prediction result more intuitive and convenient for clinical workers to make judgment and decision\u003csup\u003e22\u003c/sup\u003e.\u003csup\u003e\u0026nbsp;\u003c/sup\u003eAs shown in the figure 3, we observed that in the green part, the predicted results were consistent with the actual amputation results, while in the pink part, the two were completely opposite.We roughly estimated that the prediction accuracy of the model is about 81.2% through the confusion matrix, which basically conforms to the prediction.The true positive rates of predicted amputation and non-amputation, respectively were 79% and 83%.\u003c/p\u003e\n\u003cp\u003eWe tried to use SVM method to build a prediction model, run the prediction model according to the customized function, calculate the accuracy of the model, and select the function with the highest accuracy as the modeling decision tree. The prediction accuracy of our function was up to 82.4%, which was basically in line with expectations. This proved that the prediction model we constructed had a certain guiding effect on clinical work.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e3.4 Model evaluation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFigure 5 was plotted based on whether generalized amputation or not, where the Area Under Curve (AUC) is 0.89. The closer the curve area is to 1, the higher the accuracy of the proof model. The maximum approximate entry index (Max(sensitivity + specificity -1)) under the ROC curve was supreme when the tangent point was 0.21, corresponding to 0.83.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe calibration ability of the model was 19.614(p=0.012) through Hosmer-Lemeshow test. Figure 6(a) was the calibration curve of the model on the training set, Figure 6(b) was the calibration curve of the model on the test set.\u0026nbsp;Calibration curves were drawn based on the agreement between the observed predicted risk of amputation and the actual results. The Y-axis represented the actual amputation outcome, and the X-axis represented the predicted risk of amputation. The gray diagonal line represented an ideal perfect model, while the solid black line represents the performance of the model, the closer the two lines were, the better the fit.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e3.5 Multimodal analysis\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eThe nomogram of the prediction model based on whether generalized amputation or not was shown in Figure 7, integrating all independent predictors.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFIGURE 8 shown the net benefit of the model to the patient as the threshold selection changes. When the threshold value was selected as 0.302 derived from the Jorden index, the model was able to generate a net gain of 0.351.\u003c/em\u003e\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eIn this study, we developed an amputation prediction model that incorporates 10 baseline features to predict the probability of amputation in DF patients. The AUC and calibration capability of the prediction model were 0.89 and 19.614, respectively. At the same time, according to the decision analysis curve, when the threshold is 0.302, the net benefit is 0.351. This shows that the prediction model has a good clinical application prospect, and that the predicted probability of the model is in reliable agreement with the actual probability. Numerous studies have shown that the more severe the infection, the higher the amputation rate in DF patients\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. In the prediction model of amputation prognosis of DF patients, a machine learning model with infection severity, lower limb blood supply and systemic nutrition as input variables was established. We found that in the patient cases we collected, the prognosis of patients with amputation was affected by multiple factors, such as severe infection, lower limb vascular occlusion, poor nutritional status, etc., which was basically consistent with clinical experience and related studies \u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. In addition, it is worth noting that hyperlipidemia also increases the risk of amputation to some extent. The results are meaningful and generally in line with our expectations.\u003c/p\u003e \u003cp\u003eThe Wagner classification system is widely used in clinical practice to assess the severity of foot in diabetic patients. It categorizes foot ulcers into six grades\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. The classification of wounds is based on depth, extent and degree of infection, which are just what we need to observe. This additional information helps to improve the accuracy of the model in predicting the risk of amputation in DFU patients. We hope to build a prediction model for amputation risk and further improve the DFU classification system from multiple aspects\u003csup\u003e\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. Overall, incorporating the Wagner classification system into the predictive model allows for a more comprehensive assessment of the risk of amputation in DFU patients.It enhances the predictive power of the model and provides valuable clinical information for personalized treatment and management of diabetic foot disease.\u003c/p\u003e \u003cp\u003eIn addition, simple linear models (COX regression, multiple Logistic regression, COX proportional risk model) have certain limitations in assessing the prognosis of DFU patients due to the diversity and unpredictability of influencing factors\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. Boyko et al.\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e first carried out a prospective cohort study on diabetic patients in 2006, applied Cox proportional risk model to screen independent influencing factors, and finally formed a scoring system model. The AUC of this model was 0.81, indicating good differentiation. But the study was not externally verified. In 2019, British scholar Heald \u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e et al. analyzed diabetic patients through retrospective cohort study and Logistic multiple regression, finally included 5 risk factors, and built a calculation equation for the risk probability of diabetic foot based on the regression coefficients of each factor. The model AUC was 0.65. Although the model has certain clinical practicability, the accuracy rate is still low. The emergence of artificial intelligence provides new ideas and methods for diabetes risk prediction. In 2021, Peng et al.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003econstructed a model to predict the risk of diabetic foot amputation, and constructed the nomogram to visually compare the risk weights of each risk factor. The model AUC was 0.876. Deng et al.\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003eused XGBoost algorithm and COX regression to evaluate the impact of hyperglycemic crisis and other risk factors on the mortality of DFU patients. The model\u0026rsquo;s AUC is 0.680. A prospective study by Lv et al. \u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e established a DFU risk model based on risk factors and presented it in the form of the nomogram and web calculator. The AUC of its model was 0.741. Therefore, our study used ANN to establish the model. Machine learning algorithms adopt a multivariate, non-parametric approach that can use non-normal distributions and strongly correlated data to build robust models and identify complex patterns\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Compared with the statistical methods in previous relevant studies, the prediction model we constructed includes more relevant factors, has higher accuracy and more intuitive prediction results.We believe that our model is convincing for the guidance of clinical work.\u003c/p\u003e \u003cp\u003eThe combination of model prediction and clinical decision making is another feature of our experiment. In this study, we intend to construct a database of diabetic foot - diabetic foot amputation patients, study the real world of diabetic foot amputation patients, and validate predictive models to guide clinical decision making. Obviously, whether the predictive model based on retrospective data analysis can be applied to clinical practice needs to be further verified. Therefore, we hope to collect the information of newly included diabetic foot patients, compared the model prediction results, optimized the internal algorithm of the model, further screened the relevant risk factors, and understood the distribution of diabetic foot characteristics in the current survey population. Based on the relevant risk factors we have screened, we should proceed with the multi-disciplinary combined diagnosis and treatment of diabetic foot patients from the four aspects of nutrition, blood circulation, wound surface and infection. It is essential for the early prevention, timely intervention and scientific individualized treatment of DFU.\u003csup\u003e40\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eHowever, there are some limitations to our study. Although we use SVM algorithms to build models, we still lack sufficient clinical validation and cohort studies.Although we use SVM algorithms to build models, we still lack sufficient clinical validation and cohort studies. It is true that machine learning models primarily establish mapping relationships between input variables and predicted outcomes, rather than capturing direct causal relationships. Identifying causal relationships in complex medical conditions such as foot ulcers is challenging. Machine learning models can provide insights into the associations between patient characteristics and predicted outcomes, but further investigation and studies are needed to establish causality. It is essential to consider the timing and effectiveness of treatment interventions when interpreting the predictions made by the model. The fluctuation of the infection index of a single patient's second admission is influenced by the treatment measures. At the same time, we can not ignore that the decision of diabetic foot amputation surgery is affected by many aspects, such as the patient's economic status, wound status, and the subjective judgment of clinicians, and it is difficult to predict whether amputation is possible only through observational indicators. We want to develop a model that differ from the existing DFU taxonomy system and can be optimized and refined on the basis of the Wagner classification. In this way, our research constructs highly accurate and practical models to predict amputation rates in DFU patients and attempts to combine clinical care with long-term prognosis.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eIn summary, we built an intelligent model which can be used to forecast the risk of inpatient amputation in DF patients and analyzed the real world of newly enrolled diabetic foot patients. Our experimental results show that the machine learning model not only has accurate predictive power, but also provides new ideas for the formulation of personalized treatment plans for patients.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConflict of interest:\u0026nbsp;\u003c/strong\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval:\u0026nbsp;\u003c/strong\u003eThe study was notifed to the Ethics Committee of Beijing Shijitan Hospital Affiliated to Capital Medical University and it was waived from the formal approval due to its retrospective study design.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability:\u003c/strong\u003eThe original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding author.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contribution:\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eZ.L and S. H contributed equally to this work, they are co-first authors. The frst draft of the manuscript was written by Z.L. S.H was involved in chart production. Q.Y., J.W. and L.G are the co-corresponding authors of this manuscript. Q.Y participated in putting forward constructive suggestions for the follow-up of the research. J. W. was responsible for guiding the construction of the model. L.G. was in charge of providing clinical cases.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding information:\u003c/strong\u003eThis study was not supported by any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eIbrahim, A. IDF Clinical Practice Recommendation on the Diabetic Foot: A Guide for Healthcare Professionals. Diabetes Res. Clin.Pract. 2017, 127, 285\u0026ndash;287.\u003c/li\u003e\n\u003cli\u003eGlobal disability burdens of diabetes-related lower-extremity complications in 1990 and 2016. Zhang Y, Lazzarini PA, McPhail SM, et al [J]. Diabetes Care. 2020;43:964\u0026ndash;974.\u003c/li\u003e\n\u003cli\u003eVolmer-Thole, M.; Lobmann, R. Neuropathy and Diabetic Foot Syndrome. Int. J. Mol. Sci. 2016, 17, 917.\u003c/li\u003e\n\u003cli\u003eMonteiro-Soares M, Boyko EJ, Jeffcoate W, et al. Diabetic foot ulcer classifications: a critical review. Diabetes/Metab Res Rev. 2020;36:e3272.\u003c/li\u003e\n\u003cli\u003eDiabetic podiatry Group, Peripheral Vascular Diseases Committee of Chinese Microcirculation Society. Expert Consensus on diabetic foot wound repair and treatment [J]. Chinese Journal of Diabetes, 2018, 10(5):305-309.\u003c/li\u003e\n\u003cli\u003eChinese Medical Association Diabetes Society, Chinese Medical Association Infectious Diseases Society, Chinese Medical Association Tissue Repair and Regeneration Society. Chinese Guidelines for the Prevention and Treatment of diabetic foot (2019 edition) (Ⅳ). Chinese Journal of Diabetes, 2019, 11(5):316-327.\u003c/li\u003e\n\u003cli\u003eLi Zhihong, Guo Shuqin, Li Tingting, et al. Comparison of Wagner grading method for diabetic foot and University of TEXAS classification [J]. Chinese Journal of Diabetes, 2012,4 (8) : 469-73.\u003c/li\u003e\n\u003cli\u003eInternational Diabetic Foot Working Group: Classification of diabetic foot ulcers Part of the International Diabetic Foot Working Group: International Guidelines for Diabetic Foot Management (2019) [J]. Infection, Inflammation and Repair, 2019, 20(4):231-237.\u003c/li\u003e\n\u003cli\u003eXie Xiaoran, Xu Rong. A systematic evaluation of risk prediction model for diabetic foot [J]. Chinese Journal of Nursing, 21, 56 (1) : 124-131\u003c/li\u003e\n\u003cli\u003eHeald A,Lunt M,Rutter M K,et al. Developing a foot ulcer risk model:what is needed to do this in a real-world primary care setting? [J]. Diabet Med,2019,36(11):1412-1416.\u003c/li\u003e\n\u003cli\u003eMoons K G,Altman D G,Reitsma J B,et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis(TRIPOD):explanation and elaboration [J]. Ann Intern Med,2015,162(1):W1-73.\u003c/li\u003e\n\u003cli\u003eBasu A, Manning WG, Mullahy J. Comparing alternative models: log vs cox proportional hazard? [J]. Health Econ. 2004;13(8):749-765.\u003c/li\u003e\n\u003cli\u003eWright RE. Logistic regression. Reading and Understanding Multivariate Statistics [J]. Washington DC: American Psychological Association; 1995:217-244. \u003c/li\u003e\n\u003cli\u003eNanda R, Nath A, Patel S, et al. Machine learning algorithm to evaluate risk factors of diabetic foot ulcers and its severity [J]. Med Biol Eng Comput, 2022, 60(8): 2349-2357. \u003c/li\u003e\n\u003cli\u003eZale A, Mathioudakis N. Machine learning models for inpatient glucose prediction[J]. Curr Diab Rep, 2022, 22(8): 353-364.\u003c/li\u003e\n\u003cli\u003eDeng L, Xie P, Chen Y, et al. Impact of acute hyperglycemic crisis episode on survival in individuals with diabetic foot ulcer using a machine learning approach [J]. Front Endocrinol (Lausanne), 2022, 13: 974063.\u003c/li\u003e\n\u003cli\u003eBrereton RG, Lloyd GR. Support vector machines for classification and regression [J]. Analyst, 2010, 135(2): 230-267.\u003c/li\u003e\n\u003cli\u003eAnsari D，Nilsson J，Andersson R，et al. Artificial neural networks predict survival from pancreatic cancer after radical surgery [J]. Am J Surg，2013，205（1）：1-7.\u003c/li\u003e\n\u003cli\u003eAmiri Z，Mohammad K，Mahmoudi M，et al. Assessing the effect of quantitative and qualitative predictors on gastric cancer individuals survival using hierarchical artifcial neural network models [J]. Iran Red Cresc Med J，2013，15（1）：42-48.\u003c/li\u003e\n\u003cli\u003eQiao G，Li J，Huang A，et al. Artificial neural networking model for the prediction of post -hepatectomy survival of patients with early hepatocellular carcinoma [J]. J Gastroenterol Hepatol，2014，29（12）：2014-2020.\u003c/li\u003e\n\u003cli\u003eZhang Lijin, Wei Xiayan, Lu Jiaqi, et al. Lasso regression: From interpretation to prediction [J]. Advances in Psychological Science, 2019,28(10) :1777-1788.\u003c/li\u003e\n\u003cli\u003eValero-Carreras Daniel, Alcaraz Javier, Landete Mercedes, et al. Comparing two SVM models through different metrics based on the confusion matrix [J]. Computers and Operations Research. Volume 152 , Issue . 2023\u003c/li\u003e\n\u003cli\u003eMusa IR，Ahmed MON，Sabir EI，et al. Factors associated with amputation among patients with diabetic foot ulcers in a Saudi population[J]. BMC Research Notes，2018，11（1）：260.\u003c/li\u003e\n\u003cli\u003eHu Huiping, Chen Mingwei. Analysis of related factors of diabetic foot ulcer healing and amputation [J]. Journal of Anhui Medical University, 2016,51 (11) : 1634-1637. \u003c/li\u003e\n\u003cli\u003eSun J-H, Tsai J-S, Huang C-H, et al. Risk factors for lower extremity amputation in diabetic foot disease categorized by Wagner classification[J]. Diabetes Res Clin Pract. 2012;95(3):358-363.\u003c/li\u003e\n\u003cli\u003eZhan LX, Branco BC, Armstrong DG, Mills JL Sr. The Society for Vascular Surgery lower extremity threatened limb classification system based on wound, ischemia, and foot infection(WIfI) correlates with risk of major amputation and time to wound healing[J]. J Vasc Surg. 2015;61(4):939-944.\u003c/li\u003e\n\u003cli\u003eMehraj M, Shah I. A review of Wagner classification and current concepts in management of diabetic foot. Int J Orthop Sci.2018;4(1):933-935.\u003c/li\u003e\n\u003cli\u003eLu S, Chen R, Wei W, Lu X. Understanding heart-failure patients EHR clinical features via SHAP interpretation of tree-based machine learning model predictions. arXiv preprint arXiv:210311254; 2021.\u003c/li\u003e\n\u003cli\u003eJeon BJ, Choi HJ, Kang JS, Tak MS, Park ES. Comparison of five systems of classification of diabetic foot ulcers and predictive factors for amputation[J]. Int Wound J. 2017;14(3):537-545.\u003c/li\u003e\n\u003cli\u003eSantema TB, Lenselink EA, Balm R, Ubbink DT. Comparing the Meggitt-Wagner and the University of Texas wound classification systems for diabetic foot ulcers: inter-observer analyses[J].Int Wound J. 2016;13(6):1137-1141.\u003c/li\u003e\n\u003cli\u003eBeckert S, Witte M, Wicke C, K\u0026ouml;nigsrainer A, Coerper S. A new wound-based severity score for diabetic foot ulcers: a prospective analysis of 1,000 patients [J]. Diabetes Care. 2006;29(5):988-992.\u003c/li\u003e\n\u003cli\u003eBOYKO E J,AHRONI J H,COHEN V,et al. Prediction of diabetic foot ulcer occurrence using commonly available clinical information:the seattle diabetic foot study [J]. Diabetes Care,2006,29(6):1202-1207. \u003c/li\u003e\n\u003cli\u003eHEALD A,LUNT M,RUTTER M K,et al. Developing a foot ulcer risk model:what is needed to do this in a real-world primary care setting? [J]. Diabet Med, 2019,36(11):1412-1416. \u003c/li\u003e\n\u003cli\u003eBo P, Rui M, Yi L,et al. Development of Predictive Nomograms for Clinical Use to Quantify the Risk of Amputation in Patients with Diabetic Foot Ulcer[J]. Diabetes Res.2021,6621035(2021). \u003c/li\u003e\n\u003cli\u003eDeng L, Xie P, Chen Y, et al. Impact of acute hyperglycemic crisis episode on survival in individuals with diabetic foot ulcer using a machine learning approach [J]. Front Endocrinol (Lausanne), 2022, 13: 974063.\u003c/li\u003e\n\u003cli\u003eLv, J. et al. Development and validation of a risk prediction model for foot ulcers in diabetic patients.[J]. Diabetes Res. 2023, 1199885 (2023).\u003c/li\u003e\n\u003cli\u003eRipatti S,Tikkanen E,Orho-Melander M,et al. A multilocus genetic risk score for coronary heart disease：Case -control and prospective cohort analyses［J］. The Lancet，2010，376（9750）：1393-1400.\u003c/li\u003e\n\u003cli\u003eSilver M, Chen P, Li R,et al. Pathways-driven sparse regression identifies pathways and genes associated with high - density lipoprotein cholesterol in two asiancohorts［J］. PLoS Genetics，2013，9（11）：e1003939\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Table 1","content":"\u003cp\u003eTable 1 is available in the Supplementary Files section.\u003c/p\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":"Diabetic foot, Amputation, Maching learning, Wagner grade, Prediction model, Intervention treatment","lastPublishedDoi":"10.21203/rs.3.rs-5716696/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5716696/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cem\u003eBackground\u003c/em\u003e. Diabetic foot (DF) is a destructive complication of diabetes,which has high amputation rate and causes great social burden. With the development of artificial intelligence, there is an urgent need for an accurate prediction method for the prognosis of chronic diseases.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMethods\u003c/em\u003e. This retrospective study collected the data of 149 patients with DF from January 2019 to December 2022. The data \u0026nbsp;was divided into the test set and training set by 5-fold cross-validation. Then we used Lasso algorithm to screen relevant risk factors, and the predictive model was built with support vector mechanism(SVM) to forecast probability. Combining multiple methods such as receiver operating characteristic, calibration, and risk decision curve for multimodal analysis of the model.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eResults. \u003c/em\u003eStatistical analysis showed that there were statistically significant differences in CRP, PCT, WBC and Wagner grading between the training and testing sets (P＜0.05). According to the results of multiple factor analysis, CRP and Wagner rating 3-4 were the independent predictive factors for generalized amputation.The area under the receiver operating characteristic(ROC) curves of themodel were 0.89. This model’s calibration capability was 19.614 through Hosmer-Lemeshow test (p=0.012).When the threshold value was selected as 0.302, the constructed model was able to generate a net gain of 0.351.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eConclusion\u003c/em\u003e. The predictive model we constructed can accurately estimate the rate of amputation during hospitalization in DFU patients. Our survey data suggested that C-reactive protein(CRP) and the Wagner grade of the affected foot might play a vital role in predicting diabetic foot amputation.\u003c/p\u003e","manuscriptTitle":"A Meaningful Machine Learning Model for Predicting Amputation Rate of Patients with Diabetic Foot","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-01-01 09:27:12","doi":"10.21203/rs.3.rs-5716696/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":"3f3486b1-f7ea-481f-87c4-5800687e25b2","owner":[],"postedDate":"January 1st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-03-04T05:23:36+00:00","versionOfRecord":[],"versionCreatedAt":"2025-01-01 09:27:12","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5716696","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5716696","identity":"rs-5716696","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}