A predictive model for vascular complications of free flap transplantation based on machine learning

doi:10.21203/rs.3.rs-2619423/v1

Download PDF

Research Article

A predictive model for vascular complications of free flap transplantation based on machine learning

https://doi.org/10.21203/rs.3.rs-2619423/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objective: Exploring the risk factors for vascular complications after free flap transplantation and establishing a clinical auxiliary assessment tool for vascular complications in patients undergoing free flap transplantation based on machine learning methods.

Methods:Collecting patients who underwent free flap transplantation at a tertiary hospital in Guizhou Province from January 1, 2019 to December 31, 2021, using synthetic minority oversampling technique to oversample the training set, and constructing Logistic regression model, random forest model and neural network model three machine learning models and verifying them using the sampled data. Using the test set to evaluate the performance of the model through the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity and specificity indices.

Results: A total of 570 patients who underwent free flap transplantation were included in this study, of which 46 patients developed postoperative vascular complications. The neural network model performed best in the test set, with an AUC of 0.828. Multivariate logistic regression analysis showed that preoperative hemoglobin, preoperative fibrinogen, operation time, smoking history, number of anastomoses, and peripheral vascular injury were statistically significant independent risk factors for vascular complications after free flap transplantation. The top five predictive factors in the neural network were fibrinogen content, operation time, donor site, BMI, and platelet count.

Conclusion: The predictive model for vascular complications of free flap transplantation constructed in this study has good predictive ability, which can provide reference for medical personnel to take preventive measures to prevent vascular complications in high-risk patients undergoing free flap transplantation.

Machine learning

free flap

vascular crisis

risk factors

risk prediction model

With the continuous development of microsurgical technology, free flap transplantation has become one of the main treatments for repairing damaged wounds, reconstructing damaged parts and improving appearance^[1]. Studies have shown that the success rate of flap transplantation is increasing yearly, but there are still adverse complications after flap transplantation, of which the incidence of vascular complications is as high as 10% to 30%^[2]. Vascular complications refer to the phenomenon of vascular spasm or thrombosis after microsurgical anastomosis of small vessels, resulting in insufficient arterial perfusion or venous drainage of transplanted tissues, resulting in congestion or ischemia of tissues ^[3]. This can lead to the patient returning to the operating room, increasing the risk of flap transplantation failure, and thus increasing the patient's postoperative complications, hospital stay, economic burden and mental stress^{[4, 5]}. However, there are few reports of warning models with better predictive performance and clinical application potential. Therefore, this study constructed a risk warning model based on machine learning for vascular complications of free flap transplantation patients, providing technical tools and reference for clinical evaluation and treatment of free flap transplantation.

1. Study population

The data for this study were obtained from patients who underwent free flap transplantation at a tertiary hospital in Guizhou Province from January 1, 2019 to December 31, 2021. This study was approved by the Ethics Committee of Zunyi Medical University Affiliated Hospital (Approval No. KLLY-2021-086). Inclusion criteria: 1) patients who underwent free flap transplantation between January 1, 2019 and December 31, 2021; 2) patients aged 18-85 years. Exclusion criteria: 1) lack of complete clinical data; 2) exclusion of adjacent or pedicle tissue flaps. Dropout criteria: missing values of collected indicators exceed 30% After unified training, research personnel applied a unified baseline collection table, filled in the data collection table according to the real situation of the patient, and then checked the information in the collection table again to ensure the accuracy of the information.

2. Clinical data

The data collected from patients who underwent free flap transplantation at a tertiary hospital in Guizhou Province from January 1, 2019 to December 31, 2021 included preoperative variables such as gender, age, height, weight, BMI, drinking history, smoking history, diabetes, radiotherapy/chemotherapy history, ASA anesthesia score, preoperative biochemical indicators such as blood sugar, hemoglobin, white blood cell, prealbumin, platelet count, fibrinogen, prothrombin time, intraoperative variables such as wound area, venous anastomosis number, vascular anastomosis method, donor area, recipient area, intraoperative transfusion, postoperative day indicators such as postoperative infusion volume, postoperative microcirculation drug use, whether appropriate posture is given after surgery, pain score, surgeon and final outcome indicators.

3. Statistical analysis

Data were analyzed using SPSS 18.0 software. Metric data that conformed to the normal distribution were expressed as mean ± standard deviation (`x ± s ), and metric data that did not conform to the normal distribution were expressed as median and interquartile range M(P₂₅, P75). Count data were described by frequency and percentage, and intergroup differences were tested using chi-square test. The t-test was used for metric data that conformed to the normal distribution and homogeneity of variance in one-way analysis, and the rank sum test was used for metric data that did not conform to the normal distribution. Discrimination analysis was used to predict the performance of the model. Discrimination refers to the ability of the model to distinguish whether vascular complications occurred after free flap transplantation in patients, and the area under the ROC curve (AUC) was used to quantify it. Other related indicators included accuracy, sensitivity, and specificity. P<0.05 was considered statistically significant.

4. Model construction

Anaconda 3 (Python 3.9) and Python's third-party libraries such as Scikit-learn 0.24.2 were used to build the model. Since the ratio of vascular complications to non-complications in this study was an imbalanced data, Synthetic Minority Oversampling Technique (SMOTE) was used to sample the relatively fewer positive samples to achieve the goal of balanced data. All data were randomly divided into two independent datasets - training set and test set - in a 6:4 ratio . The training set was sampled by SMOTE and used for model training, and the test set was used for model performance verification. Logistic Regression (LR), Random Forest (RF) and Artificial Neural Network (ANN) machine learning models were established through Python's built-in machine learning library. For the Random Forest algorithm, the model optimal parameters were searched by random search (GridSearch) plus 5-fold cross-validation, and the final RF model was constructed. For the Neural Network model, 10 models were established using KerasTunner, and the model with the best parameters and score was selected as the final ANN model. Finally, the performance of the final models was compared. To increase the number of predictive factors included, this study used factors with P<0.1 as the input variables of the model.

1. Patient characteristics

In this study, a total of 600 free flap transplant patients were collected, of which 30 were excluded due to incomplete case data, resulting in 570 free flap transplant patients as the research objects. The average age of the study population was 49.5 years, with 408 males (70.0%) and 162 females (30.0%); 284 smokers (50.0%) and 286 non-smokers (50.0%). According to the surgical exploration, 46 cases were diagnosed with vascular complications, including 10 arterial complications, 33 venous complications, and 3 arteriovenous complications, with a rate of 8.07% of free flap vascular complications. Of the 570 free flaps, 20 cases had necrosis, and the success rate of free flap transplantation was 96.5%, which was consistent with the success rate of free flap transplantation reported in other literature.

2. Results of statistical analysis

2.1 Predictor screening results

Univariate analysis was performed on all vascular complications factors (Table 1 and Table 2), and 16 variables met the criteria of P<0.1, and all eligible variables were included: smoking history, body mass index, diabetes, preoperative platelet count, preoperative hemoglobin, preoperative fibrinogen, operation time, postoperative day infusion volume, peripheral vascular disease, preoperative chemotherapy, intraoperative transfusion, venous anastomosis number, arterial anastomosis method, flap type, donor site, recipient site. These variables were used as predictive factors and input into the model.

2.2 Risk factor analysis

Multivariate Logistic regression analysis was performed on the factors with P<0.05 in the univariate analysis (Table 3), and the results showed that preoperative hemoglobin, preoperative fibrinogen, operation time, smoking history, venous anastomosis number, and peripheral vascular disease were statistically significant (P<0.05), and were independent risk factors for the occurrence of postoperative vascular complications after free flap transplantation.

3. Model results

3.1 Model performance

In this study, three machine learning-based models were developed to analyze the potential risk factors related to free flap vascular complications. The dataset was split into training and test sets in a 6:4 ratio using the Scikit-learn dataset splitting method, with 317 patients in the training set, including 25 complication patients; and 228 patients in the test set, including 21 complication patients. The SMOTE algorithm was then used to sample the training set, and the minority class was oversampled with a ratio of 1:1 for the complication group and the non-complication group, resulting in a total of 317 complication cases. The sampled data was used to train the three models, and the test set was used to validate the models. Figure 2 is the confusion matrix of the model, displaying the prediction results of each model on the test set data. The results showed that the accuracy of the LR, RF, and ANN models were 78.5%!, 85.5%, and 78.1%, respectively; the sensitivity was 0.762, 0.571, and 0.857, respectively; the specificity was 0.787, 0.844, and 0.844, respectively; and the AUC values were 0.775, 0.728, and 0.828, respectively, as shown in Table 4 and Figure 1.

3.2 Model Variable Importance

Figure 3 shows the importance ranking of the predictive factors in the random forest model for predicting vascular complications of the flap. The factors were ranked according to the feature importance index Gini index(G) of the random forest model. The top five factors in the logistic regression model were operation time (G=0.194), fibrinogen content (G=0.169), platelet count (G=0.144), recipient site (G=0.129), and BMI (G=0.113); the top five variables in the random forest model were operation time (G=0.258), platelet count (G=0.153), preoperative fibrinogen content (G=0.129), BMI (G=0.115), and hemoglobin content (G=0.113); and the top five predictive factors in the neural network were fibrinogen content (G=0.111), operation time (G=0.103), donor site (G=0.102), BMI (G=0.09), and platelet count (G=0.07).

1 Factor associated with the development of vascular crisis

1.1 Haemoglobin

Haemoglobin is a standard objective clinical indicator to determine whether a patient is anaemic, and haemoglobin levels <130 g/L in men and <120 g/L in women are considered anaemic. Some studies have found that patients with perioperative anaemia are at an increased risk of flap graft failure in flap grafting, with a 0.5-fold decrease in flap failure for every 1g/dL increase in perioperative haemoglobin concentration.^{[6, 7]}. This is consistent with the findings of the present study, which showed that higher haemoglobin levels are a protective factor against the development of vascular complications. Haixuan Wu et al. ^[8] showed that low haemoglobin leads to an increased chance of complications after flap grafting. Therefore, they recommended a moderate increase in haemoglobin levels to maintain postoperative haemoglobin levels above ten g/dL to promote the early recovery of patients after flap grafting^{[8, 9]}.However, there are no specific guidelines for perioperative blood transfusion in patients with grafts. Other studies have suggested that perioperative blood transfusions may increase the risk of postoperative infections ^[10].

1.2 Fibrinogen

Fibrinogen has been considered essential for platelet coagulation, and studies have shown that blood in a hypercoagulable state increases the risk of thrombosis in patients after free flap grafting^[11]. The results of this study showed that patients with high preoperative fibrin levels were at increased risk of vascular crisis, with a 1.7-fold increase in the risk of flap vascular crisis for every 1 g/dL increase in patient fibrin concentration, suggesting that elevated fibrinogen concentrations can lead to increased blood viscosity, induce the development of atherosclerosis, and cause a slowing of blood circulation, increasing the risk of thrombosis^[12]. This is consistent with the view of Yang Lianghui et al. that high fibrinogen puts the blood in a hypercoagulable state and makes it more susceptible to thrombosis leading to the development of a crisis^[13].

1.3 Smoking

Smoking is an independent risk factor for the development of vascular crises. This study shows that vascular crisis occurs 4.43 times more frequently in smoke patients than in non-smokers. Smoking disrupts normal vascular physiology, causes vasospasm, and decreases blood flow to the surgical incision site^{[14, 15]}. Furthermore, nicotine in tobacco irritates the blood vessels, causing vasoconstriction, which reduces the blood supply to the recipient area, and does not provide adequate oxygen and nutrients to the skin flap after transplantation. Therefore, patients should be advised to quit smoking before surgery and be smoke-free in the ward ^[16].

1.4 Operation time

This study shows that prolonged operative time leads to an increased risk of vascular crisis after flap grafting and that the increased operative time may be the result of a combination of factors, including but not limited to the complexity of the procedure, the limited surgical experience of the surgeon^[17]. Ishimaru et al. ^[18] conducted a study reviewing national databases in Japan, where they analysed 2846 patients and found that longer operative times were associated with free flap failure. Additionally, Sanati-Mehrizy et al.^[17] a 2015 study of 2013 patients found an association between flap failure and operative time by univariate analysis.

1.5 Peripheral vascular disease

Peripheral vascular disease refers mainly to atherosclerotic stenosis or occlusive lesions of the lower extremity arteries that cause chronic or acute ischemic symptoms in the lower extremity, including asymptomatic atherosclerotic disease of the lower extremity, intermittent claudication, severe limb ischemia, and acute limb ischemia^[19]. Ishimaru et al. a study of head and neck flap grafting showed that patients with peripheral vascular disease were more likely to develop flap failure after flap grafting, consistent with the results of the present study^[18]. Peripheral vascular disease is mainly due to advanced age, diabetes mellitus, and hypertension, and other studies^[20] showed that diabetes mellitus is an independent risk factor for the vascular crisis, and it may be that diabetes mellitus causes microvascular damage in patients, which in turn increases the risk of thrombosis after flap grafting.

1.6 Number of venous anastomoses

Lee et al.^[21] a 2016 study retrospectively analysed patients with anterior femoral episcleral flap repair after oral cancer resection and found that although anastomosis of two veins took an average of approximately 30 minutes longer to operate than anastomosis of only one vein, the incidence of vascular crisis and venous vascular occlusion was significantly lower. According to the multifactorial analysis of this study, the number of venous anastomoses greater than one was a protective factor for the development of the vascular crisis. Some investigators suggest that anastomosis of two veins results in better flap return than one vein. If a patient has an obstruction of one vein, the other vein can continue to return venous blood, and two veins result in more efficient blood flow, reducing the risk of flap bruising and improving flap survival^[22]. This reduces the risk of flap stasis and improves flap survival.

2. Impact of unbalanced data

The cohort data for this study exemplify data imbalance, with significant differences in the number of minority classes (n = 46) and majority classes (n = 524). Models constructed based on imbalanced data cannot learn the typical characteristics of the minority class in full detail, resulting in predictive models developed using imbalanced data that often perform poorly when applied to new datasets. These models are often "overfitted", and overfitted models tend to underestimate the probability of events in low-risk patients and overestimate the probability of events in high-risk patients, which may influence clinical decision making^[23]. SMOTE is an oversampling technique that creates artificial samples based on actual samples in the dataset, but not identical, theoretically reducing the overfitting of the model^[24]. Since data imbalance can adversely affect the performance of predictive models, we improved the distribution of majority versus minority classes by using the SMOTE method to fit the model using the sampled training data making the model learn as much as possible about the pattern of occurrence of vascular crises.

3.Significance of each model factor

Integrating the order of importance of the variables of each model and the univariate analysis showed that the occurrence of free flap graft vascular crisis is a multifactor-driven adverse outcome. As seen in the variable importance rankings, the top 5 risk factors in the three models were roughly similar, with the main difference being that the models differed in the proportion of each predictor. We can find that time to surgery and fibrin content ranked high in importance in the three models, which corroborates with the risk factors identified by multifactorial logistic regression. Although some of the top-ranked factors were not consistent with the risk factors identified by multifactorial logistic regression, such as platelet count and BMI. Other studies reported risk factors including BMI, ischemic time, and platelet count^{[4, 25]}. For patients with excessive BMI, Sinha, S. et al^[25]suggested that adequate discretion should be exercised when performing free flap grafting in obese patients. In addition, prolonged ischemia time of free flaps and subsequent ischemia-reperfusion injury can increase the risk of postoperative complications and eventual flap graft failure, but this study did not include it due to too many missing values of ischemia time. Elevated platelets are a risk factor for thrombosis, and Stevens, M. N. et al. found a 2.67-fold increased risk of flap graft failure for every 1-unit rise in platelet count in patients with head and neck surgical free flap grafts, which is consistent with the findings of Kalmar, C. L. et al. who studied the association of thoracic microvascular repair failure with platelet counts in women^{[26, 27]}.

3. Neural network model has better predictive efficacy for free flap crises

For previous studies of the free flap graft crisis, most multifactorial aspects were analysed using traditional logistic regression-based methods^{[28, 29]} or decision tree algorithms to study flap graft failure without using neural network algorithms^[30]. It should be noted that neural network algorithms are more adaptive than machine learning algorithms such as logistic regression, decision trees, and support vector machines, and are capable of successive training when a large number of samples are updated, and their “black box” nature makes them more accessible for clinical staff to master^[31]. These algorithms’ “black box” makes them more accessible for clinical staff to master. In our study, the accuracy, sensitivity and specificity of the neural network model were 0.781, 0.857 and 0.773 respectively, with an AUC of 0.828 which shows that it has good recognition of whether vascular crisis occurs in postoperative patients and a certain degree of accuracy in predicting whether vascular crisis occurs in the future. Accuracy. We found a high accuracy (0.855) and specificity (0.884) to predict the occurrence of vascular crisis in free flaps using the random-sum model. This indicates that the random forest model can determine the non-occurrence of crisis. Still, its sensitivity is 0.571 and it does not yet have a high accuracy in determining the ability of patients to develop vascular crisis. The AUC value of the random forest model was 0.730 and the combined sensitivity and specificity showed that the model had some ability to identify the occurrence of vascular crisis in free flap grafts. The validity of the tree model constructed in this study was not significantly different from the results of predictive models demonstrated in other studies. As Shi YC et al.^[30] developed three machine learning models to predict microvascular reconstruction failure with AUC values ranging from 0.7 to 0.77, and O'Neill et al.^[32] used a random forest model to predict the occurrence of flap graft failure in breast microvascular reconstruction patients to obtain an AUC of 0.67 for the test set. The AUC (0.775) of the logistic regression was close to that of the random forest model, but there was a gap in prediction accuracy between the two.

4. Summary

Free flap grafting is a well-established technique for trauma repair, and timely identification and recognition of those at risk of postoperative complications can help improve the success rate of flap grafting^[33]. Therefore, this study predicts the occurrence of a vascular crisis in postoperative patients by building a risk model for the vascular crisis in free flap transplantation to detect the vascular crisis as early as possible and provide timely treatment to improve the success rate of flap transplantation. Machine learning methods are now widely used in disease prediction, aided diagnosis, and prognosis and have broad prospects. In this study, a commonly used machine learning algorithm was selected to construct a risk prediction model for the occurrence of vascular crisis after free flap grafting, which has an excellent predictive effect and a clinical reference value. This study is a single-centre retrospective study. Although the model achieved a high degree of discrimination, relatively few factors were included and limited cases were collected. The model needed to learn the complete pattern of the clinic’s free flaps of vascular crisis. Future multicenter and large-sample studies are needed to optimise and validate the model’s applicability.

In conclusion, the neural network model has good predictive properties for vascular crisis after free flap microvascular tissue reconstruction to help clinical workers identify high-risk groups as early as possible and provide appropriate management measures to reduce patient suffering and promote prognosis.

Ethics approval and consent to participate：This study was approved by the Ethics Committee of Zunyi Medical University Affiliated Hospital (Approval No. KLLY-2021-086). Consent to participate was provided by respondents by completing the survey, and the study was conducted in accordance with the Declaration of Helsinki

Consent for publication：Our manuscript does not contain data from any individual person, please state “Not applicable” in this section.

Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests: All financial and non-financial competing interests must be declared in this section.

Funding: No Funding.

Acknowledgements: "Not applicable" in this section

Institutional review board statement: This study was approved by the Ethics Committee of the Affiliated Hospital of Zunyi Medical University．

Informed consent statement: The data used in this study were not involved the patients’ privacy information, so the Ethics Committee of the Affiliated Hospital of Zunyi Medical University waived the informed consent.

Author's contribution：Jijin Yang and Xiaohua Wang are responsible for research design, research plan, data analysis, paper writing and paper revision; Yan Liang, Wenyan Long, Zhengang Wei, Liqin Lu and Wen Li responsible for research data entry, data checking, and paper revision. All authors have approved the submitted version and have consented to be held personally responsible for their own contributions and for ensuring that any queries related to the accuracy or authenticity of any portion of the work are thoroughly investigated, resolved, and the resolution documented in the literature. All authors have read and approved the final manuscript.

Miletta N R, Donelan M B, Hivnor C M. Management of trauma and burn scars: the dermatologist's role in expanding patient access to care [J]. Cutis, 2017, 100(1) : 18-20.
Lambert C, Creff G, Mazoue V, et al. Risk factors associated with early and late free flap complications in head and neck osseous reconstruction [J]. European archives of oto-rhino-laryngology : official journal of the European Federation of Oto-Rhino-Laryngological Societies (EUFOS) : affiliated with the German Society for Oto-Rhino-Laryngology - Head and Neck Surgery, 2022.
Laiyu， X, Lingli， P, Weihong， H, et al. Development and application of management information system for recognition and prevention of flap vascular crisis [J]. Chin J Nurs, 2022, 57(11) : 1291-6.
Las D E, de Jong T, Zuidam J M, et al. Identification of independent risk factors for flap failure: A retrospective analysis of 1530 free flaps for breast, head and neck and extremity reconstruction [J]. J Plast Reconstr Aesthet Surg, 2016, 69(7) : 894-906.
Lese I, Biedermann R, Constantinescu M, et al. Predicting risk factors that lead to free flap failure and vascular compromise: A single unit experience with 565 free tissue transfers [J]. J Plast Reconstr Aesthet Surg, 2021, 74(3) : 512-22.
Kim M J, Woo K J, Park B Y, et al. Effects of Transfusion on Free Flap Survival: Searching for an Optimal Hemoglobin Threshold for Transfusion [J]. Journal of reconstructive microsurgery, 2018, 34(8) : 610-5.
Hill J B, Patel A, Del Corral G A, et al. Preoperative anemia predicts thrombosis and free flap failure in microvascular reconstruction [J]. Annals of plastic surgery, 2012, 69(4) : 364-7.
Wu H, Liu F, Ji F, et al. Identification of Independent Risk Factors for Complications: A Retrospective Analysis of 163 Fibular Free Flaps for Mandibulofacial Reconstruction [J]. Journal of oral and maxillofacial surgery : official journal of the American Association of Oral and Maxillofacial Surgeons, 2018, 76(7) : 1571-7.
Nelson J A, Fischer J P, Chung C U, et al. Preoperative anemia impacts early postoperative recovery following autologous breast reconstruction [J]. J Plast Reconstr Aesthet Surg, 2014, 67(6) : 797-803.
Tewarie L, Moza A K, Khattab M A, et al. Effective Combination of Different Surgical Strategies for Deep Sternal Wound Infection and Mediastinitis [J]. Annals of thoracic and cardiovascular surgery : official journal of the Association of Thoracic and Cardiovascular Surgeons of Asia, 2019, 25(2) : 102-10.
Olsson E, Svartling N, Asko-Seljavaara S, et al. Activation of coagulation and fibrinolysis in microsurgical reconstructions in the lower extremities [J]. British journal of plastic surgery, 2001, 54(7) : 597-603.
Liu Chang, Zhang Kai, Li Jian-cheng, et al. Correlation of biochemical factors of coagulation and vascular crisis after oral and maxillofacial free tissue transplantation [J]. China J Oral Maxillofac Surg, 2020, 18(02) : 148-50.
Yang Liang -hui, Mai Hua -ming, Wu Jia -xiao, et al. Correlation of serum biochemical factors and vascular crisis of free flap for reconstruction of oral and maxillofacial defect [J]. China J Oral Maxillofac Surg, 2017, 15(05) : 431-4.
Al Hariri M, Zibara K, Farhat W, et al. Cigarette Smoking-Induced Cardiac Hypertrophy, Vascular Inflammation and Injury Are Attenuated by Antioxidant Supplementation in an Animal Model [J]. Frontiers in pharmacology, 2016, 7: 397.
El-Mahdy M A, Abdelghany T M, Hemann C, et al. Chronic cigarette smoke exposure triggers a vicious cycle of leukocyte and endothelial-mediated oxidant stress that results in vascular dysfunction [J]. American journal of physiology Heart and circulatory physiology, 2020, 319(1) : H51-h65.
Crippen M M, Patel N, Filimonov A, et al. Association of Smoking Tobacco With Complications in Head and Neck Microvascular Reconstructive Surgery [J]. JAMA facial plastic surgery, 2019, 21(1) : 20-6.
Sanati-Mehrizy P, Massenburg B B, Rozehnal J M, et al. Risk Factors Leading to Free Flap Failure: Analysis From the National Surgical Quality Improvement Program Database [J]. J Craniofac Surg, 2016, 27(8) : 1956-64.
Ishimaru M, Ono S, Suzuki S, et al. Risk Factors for Free Flap Failure in 2,846 Patients With Head and Neck Cancer: A National Database Study in Japan [J]. Journal of oral and maxillofacial surgery : official journal of the American Association of Oral and Maxillofacial Surgeons, 2016, 74(6) : 1265-70.
Panpan C, Jacob K, Jessica D, et al. Asian Race Negatively Impacts Peripheral Vascular Disease Severity and Postoperative Outcomes %J Journal of Vascular Surgery [J]. 2022, 76(3).
Rosado P, Cheng H T, Wu C M, et al. Influence of diabetes mellitus on postoperative complications and failure in head and neck free flap reconstruction: a systematic review and meta-analysis [J]. Head Neck, 2015, 37(4) : 615-8.
Lee Y C, Chen W C, Chen S H, et al. One versus Two Venous Anastomoses in Anterolateral Thigh Flap Reconstruction after Oral Cancer Ablation [J]. Plastic and reconstructive surgery, 2016, 138(2) : 481-9.
Zhigang， Y, Zhuang， M, Jiang H. Analysis of the risk factors of free flaps necrosis used for reconstruction of head-neck defects [J]. J Clin Stomatol, 2016, 32(11) : 681-4.
Pavlou M, Ambler G, Seaman S R, et al. How to develop a more accurate risk prediction model when there are few events [J]. BMJ (Clinical research ed), 2015, 351: h3868.
Chawla N V, Bowyer K W, Hall L O, et al. SMOTE: Synthetic Minority Over-sampling Technique [J]. 2011.
Sinha S, Ruskin O, D'Angelo A, et al. Are overweight and obese patients who receive autologous free-flap breast reconstruction satisfied with their postoperative outcome? A single-centre study [J]. J Plast Reconstr Aesthet Surg, 2016, 69(1) : 30-6.
Kalmar C L, Drolet B C, Kassis S, et al. Breast Reconstruction Free Flap Failure: Does Platelet Count Matter? [J]. Annals of plastic surgery, 2022, 89(5) : 523-8.
Stevens M N, Freeman M H, Shinn J R, et al. Preoperative Predictors of Free Flap Failure [J]. Otolaryngology--head and neck surgery : official journal of American Academy of Otolaryngology-Head and Neck Surgery, 2022, : 1945998221091908.
Ying Z, Xiaoqin B. Paired logistic regression analysis of influencing factors for vascular crisis after repair of oral and maxillofacial tumors with free tissue flaps [J]. J Prev Treat Stomatol Dis, 2022, 30(09) : 651-7.
CHEN Haoliang, SUN Guowen, Xin. C, et al. Risk factors of vascular crisis of free tissue flap after the reparation of oral and maxillofacial tissue defect [J]. Chin J Mieraaurg, 2020, 43(04) : 347-52.
Shi Y C, Li J, Li S J, et al. Flap failure prediction in microvascular tissue reconstruction using machine learning algorithms [J]. World journal of clinical cases, 2022, 10(12) : 3729-38.
DING Chun-xia, QU Ruo-wei, Ying1 Y. Automatic recognition of thalamic section in fetal two-dimensional ultrasound images based on convolutional neural network [J]. Chin Clin Med Imaging, 2020, 31(05) : 358-62.
O'Neill A C, Yang D, Roy M, et al. Development and Evaluation of a Machine Learning Prediction Model for Flap Failure in Microvascular Breast Reconstruction [J]. Annals of surgical oncology, 2020, 27(9) : 3466-75.
Qian Y, Li G, Zang H, et al. A Systematic Review and Meta-analysis of Free-style Flaps: Risk Analysis of Complications [J]. Plast Reconstr Surg Glob Open, 2018, 6(2) : e1651.

Table 1 Univariate analysis of continuous variables
Variable (unit)	Total N = 570	Normal Group N =524	Crisis Group N = 46	t/Z	P
`x ± s OR Median (P₂₅ , P₇₅ )
Age (years)	49.5 (42.0, 57.0)	49 (42, 57.0)	50.5 (39.8, 58.0)	-0.221	0.825
BMI (Kg/m² )	22.8 (20.8, 25.1)	22.6 (20.8, 25.0)	24.4 (21.3, 26.3)	-2.849	0.004^**
Prothrombin time(S)	10.7 (9.7, 12.2)	10.7 (9.6, 12.2)	10.9 (9.7, 12.3)	-0.289	0.773
Haemoglobin (g/L)	115(99,130)	116 (100.3, 131)	106 (92.0,119.0)	-3.552	<0.001^***
Albumin (g/L)	34.6±4.7	34.7±4.7	34.3±4.7	0.513	0.608
Prealbumin (mg/L)	205.8±3.5	206.5±54.6	203.5±43.7	0.353	0.724
Blood glucose value (mmol/L)	5.55 (4.88, 6.58)	5.54 (4.88, 6.58)	5.79 (4.92, 6.78)	-0.329	0.742
Platelet count (10^9/L)	233.5 (185.0, 299.2)	232 (185, 297.8)	257.5 (214.8, 330.2)	-2.435	0.015^**
Fibrinogen (g/dl)	3.27 (2.56, 4.26)	3.22 (2.52, 4.10)	4.83 (3.41, 5.61)	-5.343	<0.001^***
Operating time (min)	380 (305, 485)	380 (300, 480)	500 (427.5, 600)	-5.476	<0.001^***
Trauma area (cm² )	50 (30, 96)	50 (30, 96)	63 (31, 109)	-1.146	0.246
Intraoperative blood loss (ml)	200 (100, 300)	200 (100, 300)	300 (137.5, 325)	-1.597	0.11
Postoperative infusion volume (ml)	2150 (1750, 2462.5)	2150 (1800, 2500)	1900 (1700, 2300)	-1.947	0.052^*
BMI: Body mass index : P<0.1; : P<0.05; **: P<0.01

Table 2 Univariate analysis of categorical variables
variable	Total N = 570 [N (%)]	Normal Group N =524 [N (%)]	Crisis Group N = 46 [N (%)]	χ² /Fisher's	P
Gender				0.995	0.318
Female	162 (28.4%)	146 (27.9%)	16 (34.8%)
Male	408 (71.6%)	378 (72.1%)	30 (65.2%)
Age				0.039	0.843
<60 years old	452 (79.3%)	415 (79.2%)	37 (76.1%)
≥60 years old	118 (20.7%)	109 (20.8%)	19 (23.9%)
Smoking				9.767	0.002^**
Yes	284 (50.4%)	250 (47.7%)	34 (73.9%)
No	286 (49.6%)	274 (52.3%)	12 (26.1%)
Drinking				0.801	0.371
Yes	349 (61.2%)	206 (39.3%)	15 (32.6%)
No	221 (38.8%)	318 (60.7%)	31 (67.4%)
Hypertensive				0.085	0.77
Yes	55 (9.6%)	50 (9.5%)	5 (10.9%)
No	515 (90.4%)	474 (90.5%)	41 (89.1%)
Diabetes				7.284	0.007^**
Yes	50 (8.8%)	41 (7.8%)	9 (19.6%)
No	520 (91.2%)	483 (92.2%)	37 (80.4%)
Peripheral vascular disease				6.35	0.014^**
Yes	7 (1.2%)	4 (0.8%)	3 (6.5%)
No	563 (98.8%)	520 (99.2%)	43 (93.5%)
radiotherapy/chemotherapy				3.16	0.078^*
Yes	6 (1.1%)	4 (0.8%)	2 (4.3%)
No	564 (98.9%)	520 (99.2%)	44 (95.7%)
ASA score ≥ 3				0.357	0.55
Yes	261 (45.8%)	238 (45.4%)	23 (50%)
No	309 (54.2%)	286 (54.6%)	23 (50%)
Preoperative use of Doppler flowmetry				0.101	0.751
Yes	448 (78.6%)	411 (78.4%)	37 (80.4%)
No	122 (21.2%)	113 (21.6%)	9 (19.6%)
Intraoperative blood transfusion or not				2.845	0.092^*
Yes	129 (22.6%)	114 (21.8%)	15 (32.6%)
No	441 (77.4%)	410 (78.2%)	31 (67.4%)
Number of anastomotic veins				5.543	0.019^**
1	195 (34.2%)	172 (32.8%)	23 (50%)
Two or more	375 (68.8%)	348 (67.2%)	23 (50%)
Arterial anastomosis method				7.729	0.024^**
E-E	513 (90%)	475 (90.6%)	38 (82.6%)
E-S	56 (8.8%)	49 (9.4%)	7 (15.2%)
Vascular graft	1 (0.2%)	0	1 (2.2%)
Venous anastomosis method				3.23	0.206
E-E	554 (97.2%)	509 (97.1%)	45 (97.8%)
E-S	13 (2.3%)	13 (2.5%)	0 (0%)
Vascular graft	3 (0.5%)	2 (0.4%)	1 (2.2%)
Flap type				9.005	0.04^**
Skin flap	469 (82.3%)	434 (82.8%)	35 (76.1%)
Fascial flap	15 (2.6%)	15 (2.9%)	0
Myocutaneous flap	47 (8.2%)	39 (7.5%)	8 (17.4%)
Skeletal muscle flap	31 (5.4%)	30 (5.7%)	1 (2.2%)
Lymph node flap	8 (1.4%)	6 (1.1%)	2 (4.3%)
Donor site				19.61	0.002^***
Forearms	62 (10.9%)	61 (11.6%)	1 (2.2%)
Femoral side	328 (57.5%)	298 (56.9%)	30 (65.2%)
Gastrocnemius	63 (11.1%)	62 (11.8%)	1 (2.2%)
Back	21 (3.7%)	16 (3.1%)	5 (10.9%)
Iliac crest	43 (7.5%)	41 (7.8%)	2 (4.3%)
Foot	43 (7.5%)	39 (7.4%)	4 (8.7%)
Other	10 (1.8%)	7 (1.3%)	3 (6.5%)
Recipient site				9.753	0.018^**
Oral and maxillofacial	123 (21.6%)	115 (21.9%)	8 (17.4%)
Upper limb	213 (37.4%)	203 (38.7%)	10 (21.7%)
Lower limbs	230 (40.3%)	203 (38.7%)	27 (58.7%)
Trunk	4 (0.7%)	3 (0.6%)	1 (2.2%)
Postoperative Microcirculatory drugs				7.566	0.233
none	3 (0.5%)	3 (0.6%)	0
1 kind	82 (14.4%)	78 (14.9%)	4 (8.7%)
2 kinds	135 (23.7%)	127 (24.2%)	8 (17.4%)
3 kinds	159 (27.9%)	143 (27.3%)	16 (34.8%)
4 kinds	126 (22.1%)	110 (21%)	16 (34.8%)
5 kinds	57 (10%)	55 (10.5%)	2 (4.3%)
6 kinds	8 (1.4%)	8 (1.5%)	0
Surgeon				0.95	0.924
Group I	106 (18.6%)	99 (18.9%)	7 (15.2%)
Group II	98 (17.2%)	89 (17.0%)	9 (19.6%)
Group III	115 (20.2%)	107 (20.4%)	8 (17.4%)
Group IV	130 (22.8%)	118 (22.5%)	12 (26.1%)
Group V	121 (21.2%)	111 (21.2%)	10 (21.7%)
: P<0.1; : P<0.05; **: P<0.01

Table 3 Results of multi-factor logistic regression analysis
variable	Odds ratio (95% CI)	P
BMI	1.126 (0.998, 1.271)	0.054
Fibrinogen	1.706 (1.283, 2.27)	<0.001^*
Haemoglobin	0.987 (0.959, 0.998)	0.03^*
Operating time	1.004 (1.001,1.006)	0.005^*
Platelet count	1.003 (0.998,1.007)	0.212
Smoking	4.430 (1.842, 10.65)	0.001^*
Diabetes	2.335 (0.812,6.712)	0.116
Peripheral vascular disease	19.45 (2.16, 175.07)	0.008^*
Number of anastomotic veins	0.144 (0.056, 0.371)	<0.001^*
Arterial anastomosis method		0.918
Flap Type		0.812
Donor site		0.119
Recipient site		0.098
*P< 0.05

Table 4 Model performance evaluation

No competing interests reported.

Supplementaryfile.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

A predictive model for vascular complications of free flap transplantation based on machine learning

Status:

Version 1

Abstract

Figures

Introduction

Patients And Methods

Results

Discussion

Conclusion

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1