Machine learning for lymph node metastasis prediction of in patients with gastric cancer: A systematic review and meta-analysis

Abstract

Objective:

To evaluate the diagnostic performance of machine learning (ML) in predicting lymph node metastasis (LNM) in patients with gastric cancer (GC) and to identify predictors applicable to the models.

Methods:

PubMed, EMBASE, Web of Science, and Cochrane Library were searched from inception to March 16, 2022. The pooled c-index and accuracy were used to assess the diagnostic accuracy. Subgroup analysis was performed based on ML types. Meta-analyses were performed using random-effect models. Risk of bias assessment was conducted using PROBAST tool.

Results:

A total of 41 studies (56182 patients) were included, and 33 of the studies divided the participants into a training set and a test set, while the rest of the studies only had a training set. The c-index of ML for LNM prediction in training set and test set was 0.837 [95%CI (0.814, 0.859)] and 0.811 [95%CI (0.785-0.838)], respectively. The pooled accuracy was 0.781 [(95%CI (0.756-0.805)] in training set and 0.753 [95%CI (0.721-0.783)] in test set. Subgroup analysis for different ML algorithms and staging of GC showed no significant difference. In contrast, in the subgroup analysis for predictors, in the training set, the model that included radiomics had better accuracy than the model with only clinical predictors (F = 3.546, p = 0.037). Additionally, cancer size, depth of cancer invasion and histological differentiation were the three most commonly used features in models built for prediction.

Conclusion:

ML has shown to be of excellent diagnostic performance in predicting the LNM of GC. One of the models covering radiomics and its ML algorithms showed good accuracy for the risk of LNM in GC. However, the results revealed some methodological limitations in the development process. Future studies should focus on refining and improving existing models to improve the accuracy of LNM prediction.

Systematic Review Registration:

https://www.crd.york.ac.uk/PROSPERO/, identifier CRD42022320752

Background

Gastric cancer (GC) is the fifth most common malignancy and the third leading cause of cancer-associated death worldwide (1–3). Lymph node metastasis (LNM) is one of the most sensitive prognostic factors for patients with GC (4–6). Patients at different lymph node stages may require different degrees of lymphadenectomy or neoadjuvant therapy (7–10), and typically have different outcomes. Therefore, it is of necessity to accurately predict and evaluate LNM before making treatment decisions (11, 12).

Non-invasive imaging modalities, including computed tomography (CT), functional magnetic resonance imaging (fMRI), and B-ultrasonography, have been widely applied for the evaluation of lymph node status in GC patients. However, the performances of these techniques remain controversial due to their sensitivity, specificity, and accuracy (13–19). Sentinel lymph node (SLN) biopsy is an invasive approach that has also been adopted for LNM detection in GC (20), while is still in debates on its effectiveness. Kitagawa et al. (21) and Miyashiro et al. (22) applied two different SLN biopsy methods, but reached different false negative rates (7% and 46.4%, respectively). Endoscopic ultrasonography combined with fine needle aspiration can be used for local lymph node staging and LNM diagnosing, while the former fails to detect distant metastases (23). Several new molecular biomarkers have been found to be useful for predicting LNM of GC, but the application of these agents is limited due to high cost and complex technological requirements (24, 25). There are indeed multiple methods that have potential to diagnose LNM, whereas their performances are tied down by so many limitations and uncertainties, making it an urgent need to find a more applicable and effective method for the identification of LNM status.

Machine learning (ML) algorithm is a newly emerged technique that is capable of accurate raw data-processing, important data connections-analyzing, and accurate decision-making (26, 27). Compared with conventional statistical methods, ML model has higher prediction accuracy (28, 29). It is of critical application value in assisting disease-diagnosing and prognosis-predicting through processing massive and complex medical data (30, 31). Currently, ML has been increasingly applied for LNM prediction in GC patients (32–72). However, different types of ML prediction models have great differences in both included predictors and calculation methods of the models (73, 74), the results produced by different models are far from unanimous (75). More importantly, there is no systematic review and meta-analysis conducted to assess ML for LNM prediction in GC patients. Therefore, we reviewed and synthesized all the related studies published previously to evaluate the accuracy of ML models for LNM prediction in GC patients.

Methods

This systematic review and meta-analysis was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (76). The study was registered on PROSPERO (Registration No. CRD42022320752).

Literature retrieval strategy

PubMed, EMBASE, Web of Science, and Cochrane Library were systematically searched from inception to March 16, 2022 for all the related published articles. Search items mainly contained: “stomach neoplasms,” “machine learning,” and “lymphatic metastasis.” References of included articles were also searched manually for potential eligible studies. Detailed search procedures and strategy are presented in Supplemental File 1.

Inclusion criteria

The studies were selected according to the following criteria:

• (1) Patients were diagnosed with GC based on histopathological examination;

• (2) Cohort study published in English, with the full-text available;

• (3) Reported assessment of the performance of ML algorithm for LNM prediction;

• (4) Clearly description for ML models and predictor variables used

• (5) Reported the prediction performance indices of ML models and included sufficient data to infer the c-index and/or accuracy.

Studies meeting the following criteria were excluded:

• (1) Limited sample size (less than 100);

• (2) Letter, editorial, animal study, review, conference summary, consensus, case report, and guidelines;

• (3) Research focusing on identifying or analyzing individual predictors, rather than the development and/or verification of models;

• (4) Model performance measurements were not reported;

• (5) Model building process or method was not described.

Data extraction

Data extraction was processed by two reviewers independently (YL and FX). The list of extraction items was designed based on the modified version of Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS) (77). Discrepancies were resolved by a third reviewer (PF). The following data were extracted:

• (1) study characteristics (authors, publication date, study-design, and country or region);

• (2) cohort characteristics (number of participants, number of patients with positive LNM, and cancer stages);

• (3) Feature selection algorithms, number and types of predictor in final model, types of ML prediction model, and model validation and application;

• (4) prediction outcomes, including c-index, accuracy, sensitivity, and specificity.

Risk of bias assessment

Risk of bias assessment and applicability of included studies were performed using the Prediction Model Risk of Bias Assessment Tool (PROBAST) (78), which includes four domains; participants, predictors, outcomes, and analysis. Risk of bias in each study was assessed based on the four domains, while the applicability was evaluated based on the first three domains. Each study was graded as “high risk”, “low risk”, or “unclear risk” (78). This process was conducted and cross-checked by two reviewers (YL and FX) independently. Discrepancies were settled by the third reviewer (PF).

Statistical analysis

Data analysis was performed using R Statistical Software (version 4.1.1) with ‘matrix’, ‘metafor’ and ‘meta’ packages (79, 80). Subgroups were set based on ML algorithms. The c-index and accuracy for LNM prediction in GC patients, which were obtained from each study included, were measured with 95% confidence intervals (95% CIs) in the final analysis. For studies that did not report c-index, we calculated it via plotting receiver operating characteristic (ROC) curves based on reported probability distributions. The results from all included studies were pooled, and an overall estimated effect was evaluated using random-effect model which processed heterogeneity among studies (81). We used one-way ANOVA to discuss the differences in c-index and accuracy between the training group and the test group.

Results

Study selection

There were 2582 articles identified through searching the four databases, in which 1210 were excluded after duplicate-checking, 1126 excluded via browsing titles and abstracts. Full-texts of the remaining 246 articles were read, and 205 articles were excluded for reasons specified in Figure 1. Finally, a total of 41 studies were included (32–72).

Figure 1

Characteristics of included studies

There were 35 studies (85.4%) that were conducted in China (32, 33, 35–43, 45–49, 51–54, 56–63, 65–69, 71, 72), 5 (12.2%) in Korea (34, 44, 50, 55, 64) and 1 (2.4%) in Germany (70), with the publication date ranged from 2004 to 2022. The number of studies using ML for LNM prediction has gradually increased since 2018 (Figure 2). There were 35 retrospective studies (33, 35–53, 56–59, 61–70, 72) and 6 prospective studies (32, 34, 54, 55, 60, 71), with a total of 56182 participants, in which the number of patients with LNM was 12031. Characteristics of included studies are presented in Table 1.

Figure 2

Characteristics of machine learning in included studies

A total of 61 models were retrieved from included studies (ranged from 1 to 7 models in each study), with various modelling methods applied. The most frequently used ML algorithms were logistic regression (LR) (n=30; 49.18%), support vector machine (SVM) (n=5; 8.2%), deep learning (DL) (n=4; 6.56%), and random forest (RF) (n=4; 6.56%) (Table 1). Feature selection is an important step for ML training. The number of features used in the models varied from 2 to 21, and Figure 3 summarizes the 15 most common features. The most commonly used predictors were tumor size (n=35; 14.96%), depth of tumor invasion (n=32; 13.68%), histology differentiation (n=20; 8.55%), imaging techniques (n=17; 7.26%), lymphovascular invasion (n=17; 7.26%), tumor location (n=14; 5.98%), CT-reported LN (n=11; 4.7%), age (n=8; 3.42%), macroscopic features (n=8; 3.42%), and CA199 (n=7; 2.99%).

Figure 3

Risk of bias and applicability assessment

All the included studies were of low risk of bias with respect to the domain of participants and outcome, 19 (46.34%) studies had low risk of bias in predictors (32, 34, 37, 38, 40, 42, 44, 54–56, 60, 62, 63, 66–71), 22 (53.66%) had unclear risk of bias due to that the prediction assessment was performed in the know of outcome data (33, 35, 36, 39, 41, 43, 45–53, 57–59, 61, 64, 65, 72). As for the domain of analysis, the risk of bias in 16 studies was considered high (33, 35, 41, 43, 47, 50, 51, 53, 58, 59, 61, 63, 65, 68, 70, 71), and the reasons were that (1): Insufficient sample size. Eight of the studies did not meet the standard of including at least 100 participants (2). Selection of predictors based on univariable analysis (3). Lack of external validation techniques. Eight studies lacked external validation in model development (35, 50, 51, 53, 59, 63, 65, 70). Concern regarding ‘overall applicability’ was rated as low in 15 studies (36.59%) (32, 34, 37, 38, 40, 42, 44, 54–56, 60, 62, 66, 67, 69), high in 16 studies (39.02%) (33, 35, 41, 43, 47, 50, 53, 58, 59, 61, 63, 65, 68, 70, 71) and unclear in the remaining 10 (24.39%) (36, 39, 45, 46, 48, 49, 52, 57, 64, 72). Risk of bias and applicability assessment were shown in Table 2.

C-index

There were different numbers for training and test models because there were five studies which only reported the training results (35, 50–53). The overall c-index for ML in training group was 0.837 [(95%CI (0.814, 0.859)] (Table 3; eFigure 1). LR, one of the most commonly used ML methods, resulted in an overall pooled c-index of 0.838 [(95%CI (0.812, 0.865)] (eFigure 2), while non-logistic regression (non-LR) model resulted in an overall pooled c-index of 0.83 [(95%CI (0.786, 0.877)] (eFigure 3).

Furthermore, the pooled c-index in test group was 0.811 [(95%CI (0.785, 0.838)] (Table 3; eFigure 4), which was similar with the result in training group. Subgroup analysis showed that 21 models in LR subgroup had a pooled c-index of 0.824 [95%CI (0.791, 0.858)] (eFigure 5), and 13 models that used non-LR model assessment had a pooled c-index of 0.789 [95%CI (0.747, 0.833)] (eFigure 6).

Accuracy

There were different numbers of training and test models because there were ten studies which only reported the results of training group (35, 50–53, 59, 63, 65, 70, 71), whereas two other studies only reported that of test group. The ML models for LNM in training group showed an overall pooled accuracy of 0.781 [95%CI (0.756-0.805)] (Table 4; eFigure 7). Subgroup analysis showed no significant difference in different ML algorithms. LR algorithms had a pooled accuracy of 0.792 [95%CI (0.761-0.82)] (eFigure 8) and non-LR algorithms had that of 0.768 [95%CI (0.725, 0.805)] (eFigure 9).

In test group, the accuracy of the pooled 41 models was 0.753 [95%CI (0.721-0.783)] (Table 4; eFigure 10). Subgroup analysis was conducted based on LR and non-LR algorithms assessment. The pooled accuracy for the 23 models that used LR was 0.787 [95%CI 0.745, 0.824] (eFigure 11). The overall pooled accuracy for non-LR models was 0.707 [95%CI (0.665, 0.746)] (eFigure 12).

Subgroup analysis for early-gastric cancer and advanced gastric cancer

Of the 41 included studies, 21 were early-gastric GC (EGC) (T1) studies (32, 34, 35, 37, 39, 42, 46, 48–53, 55, 56, 60, 64, 65, 69, 71, 72) and 3 were advanced GC (T2-4) studies (43, 45, 67). In EGC, there was a pooled c-index of 0.832 [95%CI (0.804, 0.860)] (Table 5; eFigure 13) and 0.795 [95%CI (0.755, 0.838)] (eFigure 14) for the training and test groups, respectively. As for advanced GC, the pooled c-index for the training and test groups was 0.849 [95%CI (0.801-0.900)](eFigure 15) and 0.804 [95%CI (0.778-0.830)](eFigure 16), respectively.

Thirty-one models evaluated the accuracy of ML for EGC, and their pooled accuracy was 0.765 [95% CI (0.730-0.796)](eFigure 17) for the training group. In terms of test group, the pooled accuracy for EGC was 0.731 [95%CI (0.686-0.773)] (eFigure 18). As for advanced GC, the training group had a pooled accuracy of 0.821 [95%CI (0.737-0.882)] (eFigure 19) while the test group had a pooled accuracy of 0.844 [95%CI (0.794-0.884)] (eFigure 20).

Subgroup analysis for predictors

Furthermore, we reviewed the predictors in the included original studies and we found three cases: Group A included only clinical predictors, Group B included only radiomic predictors, and Group C included both clinical and radiomic predictors.

In the training group, the c-index of groups A, B, and C was 0.822 ± 0.079 (n = 25), 0.852 ± 0.072 (n = 8), and 0.847 ± 0.063 (n = 8), respectively, with no significant difference between them (F = 0.604, p=0.552) (Table 6). In the test group, the c-index of groups A, B, and C was 0.792 ± 0.092 (n = 20), 0.83 ± 0.07 (n = 8), and 0.817 ± 0.043 (n = 6), respectively, and there was also no significant difference between them (F = 0.664, p = 0.522).

In the training group, the accuracy of groups A, B, and C were 0.75 ± 0.087 (n=34), 0.811 ± 0.066 (n = 9), and 0.822 ± 0.073 (n = 7), respectively, and there was a significant difference between them (F = 3.546, p = 0.037) (Table 6), and the model containing radiomics had better accuracy. In the test group, the accuracy of groups A, B, and C were 0.722 ± 0.098 (n = 28), 0.799 ± 0.075 (n = 8), and 0.795 ± 0.04 (n = 5), respectively, although there was no significant difference between them (F = 3.224, p=0.051), the significance probability p-value was close to the critical value of 0.05. The mean value of accuracy was higher for models containing radiomics in the test cohort than for models containing only clinical predictors.

In summary, the model covering radiomics and its machine learning algorithms has better accuracy for the risk of lymph node metastasis in gastric cancer.

Discussion

The number of studies that apply ML to LNM prediction has been gradually increasing since 2018, making it important to systematically review the published studies so as to provide guidance for future research. To our knowledge, this is the first systematic review and meta-analysis that evaluated ML performance in the assessment of LNM in GC patients. ML-related studies can be methodologically categorized into LR and non-LR study. Several included studies were assessed to be of high or unclear risk of bias in the domains of prediction, analysis and overall applicability, which highlighted the current state of technology, as well as the need for methodological quality improvement.

This study demonstrated that ML had an excellent diagnostic performance in predicting LNM with great repeatability, which was in consistence with other studies. The pooled c-index and accuracy were 0.837 [95%CI (0.814,0.859)] and 0.781 [95%CI (0.756–0.805)], respectively. Significant heterogeneity existed between the studies, which could be caused by multiple factors. EGC is defined as a tumor limited to the mucosa and submucosa, regardless of the LNM (82). A subgroup analysis was performed since the difference in the order of magnitude characteristics of LNM between EGC and advanced gastric cancer may have a certain impact on the results of machine learning. It showed no significant difference in c-index or accuracy between EGC and advanced gastric cancer. In addition, since the included studies used LR or non-LR, subgroup analysis based on this variable was conducted to observe the changes in heterogeneity between the two groups. There was also no significant difference in c-index or accuracy among different ML algorithms. The type of ML algorithm had no effect on LNM prediction. Most importantly, this study was not designed to identify one superior algorithm from the other ones.

Feature selection was also critical to the performance and interpretation of the model. The most commonly used variables in the model development were tumor size, depth of tumor invasion, histology differentiation, imaging techniques, lymphovascular invasion, tumor location, CT-reported LN, age, macroscopic features, and CA199. These variables are either anthropometric characteristics serving as markers of disease severity, or important factors contributing to the natural disease progression. These predictive indicators are easy to measure. Another merit of these predictors is the low risk of bias in measurement, resulting in a minimal possibility of exposure misclassification. Previous studies have revealed that the size of tumor is closely related to the incidence of LNM in patients with GC (83–85). Larger tumor typically indicates a higher risk of LNM (86–88), which might be attributed to easiness of invasion for larger tumor to the surrounding tissues. Depth of tumor invasion was found to be a strong predictor in 32 models (32, 34–37, 39, 41, 42, 46, 49, 50, 52, 54–56, 62, 69–71), which is in consistence with substantial evidence supporting its use as a predictor of LNM (36, 63, 89–91). There were 20 models considered histology differentiation as a vital factor for predicting LNM (34–37, 41, 42, 48–55, 61–63). Deeply infiltrated and poorly differentiated tumors might have sufficient nutritional support to facilitate its invasion to tissues, capillaries and lymphatic vessels, and thus to have the potential to grow and metastasize faster (69).

The novel PROBAST was applied for assessment of risk of bias and applicability of included prediction model studies, which allowed more details of the model, such as data source, processing, number of events per variable, feature selection, model development, and model validation, to be checked intensively (92–95). The PROBAST quality assessment revealed some other issues that could be avoided in future studies. First, external validation is rarely performed, which might be a primary limitation in studies of this field. Simple determination of samples for modeling would lead to an overestimation for the model performance (96), and further accuracy verification for these models would be advisable. Guidelines that include external validation should be followed when reporting ML models (97). On the other hand, most of the included studies were retrospective design, leading to confounding and selection bias. More prospective studies are needed to produce evidence of high quality. The included studies also demonstrated another roadblock to the clinical implementation of ML. The data used in most of the included studies were from single institution, which resulted in limited datasets for training and failed to exert the advantage of ML that it is effective in processing large samples on multiple dimensions (98). Also, limited number of studies were less likely to be of broad public health significance. Future studies should take into accounts the expansion of datasets from multiple centers to increase the sample size and to improve classifier performance.

We also note the importance of preoperative assessment of peritoneal metastasis of GC for prognosis. Currently, the assessment of peritoneal metastases is mainly in the form of radiomics, but the data obtained by radiomics is obtained from a variety of sources, usually by CT, which may affect the results obtained (99–106). The heterogeneity of the results can be brought about by the different parameters and bits of CT and the artificial partitioning by different investigators through their own experience, so that the prediction of preoperative peritoneal metastases based on radiomics can be highly heterogeneous. At the same time, the application of radiomics generates a large amount of high-dimensional data, and the screening of these high-dimensional data is a great challenge in clinical practice. Therefore, although the prediction of preoperative peritoneal metastasis based on radiomics has been favored by a large number of researchers in recent years, these studies have not reached a clear consensus, thus resulting in a great variation in C-index (C-index ranged from 0.712 to 0.981) (99–106). We also expect subsequent studies based on radiomics to guide preoperative peritoneal metastases.

There were several limitations in this study. The first limitation was the significant heterogeneity. The sample sizes and distributions varied in different studies, as well as heterogenous variety of feature selection methods and ML algorithms, which compromised the performance and applicability of each model. However, such heterogeneity could be deemed as a key finding that should be addressed by future studies. Second, the estimation for prediction performance was based on limited data due to the incomplete reports of the results in several studies. Third, most of the reviewed studies included GC patients in different cancer stages, which might represent a possible confounding factor that disrupted ML performance in differential diagnosis. Fourth, our findings should be interpreted prudently considering the potential significant publication bias. It is not suggested for investigators to report a test of unsatisfactory prediction values. It is probable that there might be instances in which ML might not have optimal prediction accuracies and so that has not been published yet (107). Last but not least, eight of the included studies that did not report the test set also affected the robustness of this study by causing a false high-performance result. It would be even better if all the studies provided external test results.

Conclusion

ML has shown excellent diagnostic performance for LNM prediction in GC patients, and ML models based on radiomics and clinical features could be a better potential prediction method. However, there were some methodological limitations in their development, and there is still room for improvement in predictive value. Future studies are needed to explore efficient, minimally invasive, and easily collected predictors for LNM so as to build more effective ML models and improve the accuracy of LNM prediction.

Funding

This work support by the National Natural Science Foundation of China. (Grant No. 81673854).

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

• 1

  JoshiSSBadgwellBD. Current treatment and recent progress in gastric cancer. CA: Cancer J Clin (2021) 71(3):264–79. doi: 10.3322/caac.21657

  • CrossRef
  • Google Scholar

• 2

  SmythECNilssonMGrabschHIvan GriekenNCLordickF. Gastric cancer. Lancet (London England) (2020) 396(10251):635–48. doi: 10.1016/S0140-6736(20)31288-5

  • CrossRef
  • Google Scholar

• 3

  BrayFFerlayJSoerjomataramISiegelRLTorreLAJemalA. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: Cancer J Clin (2018) 68(6):394–424. doi: 10.3322/caac.21492

  • CrossRef
  • Google Scholar

• 4

  IsomotoHShikuwaSYamaguchiNFukudaEIkedaKNishiyamaHet al. Endoscopic submucosal dissection for early gastric cancer: a large-scale feasibility study. Gut (2009) 58(3):331–6. doi: 10.1136/gut.2008.165381

  • CrossRef
  • Google Scholar

• 5

  KeBLiangH. Current status of lymph node dissection in gastric cancer. Chin J Cancer Res = Chung-kuo yen cheng yen chiu (2021) 33(2):193–202. doi: 10.21147/j.issn.1000-9604.2021.02.07

  • CrossRef
  • Google Scholar

• 6

  Rawicz-PruszyńskiKCisełBMlakRMielkoJSkórzewskaMKwietniewskaMet al. The role of the lymph node ratio in advanced gastric cancer after neoadjuvant chemotherapy. Cancers (2019) 11(12):1914. doi: 10.3390/cancers11121914

  • CrossRef
  • Google Scholar

• 7

  CoccoliniFMontoriGCeresoliMCimaSValliMCNitaGEet al. Advanced gastric cancer: What we know and what we still have to learn. World J Gastroenterol (2016) 22(3):1139–59. doi: 10.3748/wjg.v22.i3.1139

  • CrossRef
  • Google Scholar

• 8

  DegiuliMDe ManzoniGDi LeoAD'UgoDGalassoEMarrelliDet al. Gastric cancer: Current status of lymph node dissection. World J Gastroenterol (2016) 22(10):2875–93. doi: 10.3748/wjg.v22.i10.2875

  • CrossRef
  • Google Scholar

• 9

  Japanese Gastric cancer treatment guidelines 2018 (5th edition). Gastric Cancer (2021) 24(1):1–21. doi: 10.1007/s10120-020-01042-y

  • CrossRef
  • Google Scholar

• 10

  WangFHZhangXTLiYFTangLQuXJYingJEet al. The Chinese society of clinical oncology (CSCO): Clinical guidelines for the diagnosis and treatment of gastric cancer, 2021. Cancer Commun (London England) (2021) 41(8):747–95. doi: 10.1002/cac2.12193

  • CrossRef
  • Google Scholar

• 11

  AbdelfatahMMBarakatMLeeHKimJJUedoNGrimmIet al. The incidence of lymph node metastasis in early gastric cancer according to the expanded criteria in comparison with the absolute criteria of the Japanese gastric cancer association: a systematic review of the literature and meta-analysis. Gastrointestinal Endoscopy (2018) 87(2):338–47. doi: 10.1016/j.gie.2017.09.025

  • CrossRef
  • Google Scholar

• 12

  YangJWuQXuLWangZSuKLiuRet al. Integrating tumor and nodal radiomics to predict lymph node metastasis in gastric cancer. Radiother Oncol J Eur Soc Ther Radiol Oncol (2020) 150:89–96. doi: 10.1016/j.radonc.2020.06.004

  • CrossRef
  • Google Scholar

• 13

  CardosoRCoburnNSeevaratnamRSutradharRLourencoLGMaharAet al. A systematic review and meta-analysis of the utility of EUS for preoperative staging for gastric cancer. Gastric Cancer (2012) 15 Suppl 1:S19–26. doi: 10.1007/s10120-011-0115-4

  • CrossRef
  • Google Scholar

• 14

  KweeRMKweeTC. Imaging in assessing lymph node status in gastric cancer. Gastric Cancer (2009) 12(1):6–22. doi: 10.1007/s10120-008-0492-5

  • CrossRef
  • Google Scholar

• 15

  ZhongJZhaoWRenFQiSWangXLvTet al. Lymph node metastasis in patients with gastric cancer: a multi-modality, morphologic and functional imaging study. Am J Trans Res (2016) 8(12):5601–9.

  • Google Scholar

• 16

  PangLWangJFanYXuRBaiYBaiL. Correlations of TNM staging and lymph node metastasis of gastric cancer with MRI features and VEGF expression. Cancer Biomarkers Section A Dis Markers (2018) 23(1):53–9. doi: 10.3233/CBM-181287

  • CrossRef
  • Google Scholar

• 17

  KimHJKimAYOhSTKimJSKimKWKimPNet al. Gastric cancer staging at multi-detector row CT gastrography: comparison of transverse and volumetric CT scanning. Radiology (2005) 236(3):879–85. doi: 10.1148/radiol.2363041101

  • CrossRef
  • Google Scholar

• 18

  ChenJCheongJHYunMJKimJLimJSHyungWJet al. Improvement in preoperative staging of gastric adenocarcinoma with positron emission tomography. Cancer (2005) 103(11):2383–90. doi: 10.1002/cncr.21074

  • CrossRef
  • Google Scholar

• 19

  SeevaratnamRCardosoRMcGregorCLourencoLMaharASutradharRet al. How useful is preoperative imaging for tumor, node, metastasis (TNM) staging of gastric cancer? a meta-analysis. Gastric Cancer (2012) 15 Suppl 1:S3–18. doi: 10.1007/s10120-011-0069-6

  • CrossRef
  • Google Scholar

• 20

  HiratsukaMMiyashiroIIshikawaOFurukawaHMotomuraKOhigashiHet al. Application of sentinel node biopsy to gastric cancer surgery. Surgery (2001) 129(3):335–40. doi: 10.1067/msy.2001.111699

  • CrossRef
  • Google Scholar

• 21

  KitagawaYTakeuchiHTakagiYNatsugoeSTerashimaMMurakamiNet al. Sentinel node mapping for gastric cancer: a prospective multicenter trial in Japan. J Clin Oncol (2013) 31(29):3704–10. doi: 10.1200/JCO.2013.50.3789

  • CrossRef
  • Google Scholar

• 22

  MiyashiroIHiratsukaMSasakoMSanoTMizusawaJNakamuraKet al. High false-negative proportion of intraoperative histological examination as a serious problem for clinical application of sentinel node biopsy for early gastric cancer: final results of the Japan clinical oncology group multicenter trial JCOG0302. Gastric Cancer (2014) 17(2):316–23. doi: 10.1007/s10120-013-0285-3

  • CrossRef
  • Google Scholar

• 23

  LiBZhengPZhuQLinJ. Accurate preoperative staging of gastric cancer with combined endoscopic ultrasonography and PET-CT. Tohoku J Exp Med (2012) 228(1):9–16. doi: 10.1620/tjem.228.9

  • CrossRef
  • Google Scholar

• 24

  HuangKHLanYTFangWLChenJHLoSSLiAFet al. The correlation between miRNA and lymph node metastasis in gastric cancer. BioMed Res Int (2015) 2015:543163. doi: 10.1155/2015/543163

  • CrossRef
  • Google Scholar

• 25

  LiWYeFWangDSunXTongWLianGet al. Protein predictive signatures for lymph node metastasis of gastric cancer. Int J Cancer (2013) 132(8):1851–9. doi: 10.1002/ijc.27864

  • CrossRef
  • Google Scholar

• 26

  OliveiraAL. Biotechnology, big data and artificial intelligence. Biotechnol J (2019) 14(8):e1800613. doi: 10.1002/biot.201800613

  • CrossRef
  • Google Scholar

• 27

  MirzaBWangWWangJChoiHChungNCPingP. Machine learning and integrative analysis of biomedical big data. Genes (2019) 10(2):87. doi: 10.3390/genes10020087

  • CrossRef
  • Google Scholar

• 28

  BiQGoodmanKEKaminskyJLesslerJ. What is machine learning? a primer for the epidemiologist. Am J Epidemiol (2019) 188(12):2222–39. doi: 10.1093/aje/kwz189

  • CrossRef
  • Google Scholar

• 29

  WangZLiHCarpenterCGuanY. Challenge-enabled machine learning to drug-response prediction. AAPS J (2020) 22(5):106. doi: 10.1208/s12248-020-00494-5

  • CrossRef
  • Google Scholar

• 30

  HandelmanGSKokHKChandraRVRazaviAHLeeMJAsadiH. eDoctor: machine learning and the future of medicine. J Internal Med (2018) 284(6):603–19. doi: 10.1111/joim.12822

  • CrossRef
  • Google Scholar

• 31

  FleurenLMKlauschTLTZwagerCLSchoonmadeLJGuoTRoggeveenLFet al. Machine learning for the prediction of sepsis: a systematic review and meta-analysis of diagnostic test accuracy. Intensive Care Med (2020) 46(3):383–400. doi: 10.1007/s00134-019-05872-y

  • CrossRef
  • Google Scholar

• 32

  MeiYWangSFengTYanMYuanFZhuZet al. Nomograms involving HER2 for predicting lymph node metastasis in early gastric cancer. Front Cell Dev Biol (2021) 9:781824. doi: 10.3389/fcell.2021.781824

  • CrossRef
  • Google Scholar

• 33

  LiJFangMWangRDongDTianJLiangPet al. Diagnostic accuracy of dual-energy CT-based nomograms to predict lymph node metastasis in gastric cancer. Eur Radiol (2018) 28(12):5241–9. doi: 10.1007/s00330-018-5483-2

  • CrossRef
  • Google Scholar

• 34

  KimSMMinBHAhnJHJungSHAnJYChoiMGet al. Nomogram to predict lymph node metastasis in patients with early gastric cancer: a useful clinical tool to reduce gastrectomy after endoscopic resection. Endoscopy (2020) 52(6):435–43. doi: 10.1055/a-1117-3059

  • CrossRef
  • Google Scholar

• 35

  ZhangMDingCXuLFengSLingYGuoJet al. A nomogram to predict risk of lymph node metastasis in early gastric cancer. Sci Rep (2021) 11(1):22873. doi: 10.1038/s41598-021-02305-z

  • CrossRef
  • Google Scholar

• 36

  JiangYWangWChenCZhangXZhaXLvWet al. Radiomics signature on computed tomography imaging: Association with lymph node metastasis in patients with gastric cancer. Front Oncol (2019) 9:340. doi: 10.3389/fonc.2019.00340

  • CrossRef
  • Google Scholar

• 37

  MuJJiaZYaoWSongJCaoXJiangJet al. Predicting lymph node metastasis in early gastric cancer patients: development and validation of a model. Future Oncol (London England) (2019) 15(31):3609–17. doi: 10.2217/fon-2019-0377

  • CrossRef
  • Google Scholar

• 38

  LiSZhaoZYangHWangDSunWLiSet al. Construction and validation of a nomogram for the preoperative prediction of lymph node metastasis in gastric cancer. Cancer Control J Moffitt Cancer Center (2021) 28:10732748211027160. doi: 10.1177/10732748211027160

  • CrossRef
  • Google Scholar

• 39

  GuoCGChenYJRenHZhouHShiJFYuanXHet al. A nomogram for predicting the likelihood of lymph node metastasis in early gastric signet ring cell carcinoma: A single center retrospective analysis with external validation. Medicine (2016) 95(46):e5393. doi: 10.1097/MD.0000000000005393

  • CrossRef
  • Google Scholar

• 40

  GaoXMaTCuiJZhangYWangLLiHet al. A radiomics-based model for prediction of lymph node metastasis in gastric cancer. Eur J Radiol (2020) 129:109069. doi: 10.1016/j.ejrad.2020.109069

  • CrossRef
  • Google Scholar

• 41

  WangXHeQLiangHLiuJXuXJiangKet al. A novel robust nomogram based on preoperative hemoglobin and albumin levels and lymphocyte and platelet counts (HALP) for predicting lymph node metastasis of gastric cancer. J Gastrointestinal Oncol (2021) 12(6):2706–18. doi: 10.21037/jgo-21-507

  • CrossRef
  • Google Scholar

• 42

  PanSAnWTanYChenQLiuPXuH. Prediction model of lymph node metastasis risk in elderly patients with early gastric cancer before endoscopic resection: A retrospective analysis based on international multicenter data. J Cancer (2021) 12(18):5583–92. doi: 10.7150/jca.56702

  • CrossRef
  • Google Scholar

• 43

  ChenWWangSDongDGaoXZhouKLiJet al. Evaluation of lymph node metastasis in advanced gastric cancer using magnetic resonance imaging-based radiomics. Front Oncol (2019) 9:1265. doi: 10.3389/fonc.2019.01265

  • CrossRef
  • Google Scholar

• 44

  SongBI. Nomogram using f-18 fluorodeoxyglucose positron emission tomography/computed tomography for preoperative prediction of lymph node metastasis in gastric cancer. World J Gastrointestinal Oncol (2020) 12(4):447–56. doi: 10.4251/wjgo.v12.i4.447

  • CrossRef
  • Google Scholar

• 45

  WangLGongJHuangXLinGZhengBChenJet al. CT-based radiomics nomogram for preoperative prediction of No.10 lymph nodes metastasis in advanced proximal gastric cancer. Eur J Surg Oncol J Eur Soc Surg Oncol Br Assoc Surg Oncol (2021) 47(6):1458–65. doi: 10.1016/j.ejso.2020.11.132

  • CrossRef
  • Google Scholar

• 46

  SuiWChenZLiCChenPSongKWeiZet al. Nomograms for predicting the lymph node metastasis in early gastric cancer by gender: A retrospective multicentric study. Front Oncol (2021) 11:616951. doi: 10.3389/fonc.2021.616951

  • CrossRef
  • Google Scholar

• 47

  WangXLiCFangMZhangLZhongLDongDet al. Integrating No.3 lymph nodes and primary tumor radiomics to predict lymph node metastasis in T1-2 gastric cancer. BMC Med Imaging (2021) 21(1):58. doi: 10.1186/s12880-021-00587-3

  • CrossRef
  • Google Scholar

• 48

  YinXYPangTLiuYCuiHTLuoTHLuZMet al. Development and validation of a nomogram for preoperative prediction of lymph node metastasis in early gastric cancer. World J Surg Oncol (2020) 18(1):2. doi: 10.1186/s12957-019-1778-2

  • CrossRef
  • Google Scholar

• 49

  ChenDChenGJiangWFuMLiuWSuiJet al. Association of the collagen signature in the tumor microenvironment with lymph node metastasis in early gastric cancer. JAMA Surgery (2019) 154(3):e185249. doi: 10.1001/jamasurg.2018.5249

  • CrossRef
  • Google Scholar

• 50

  EomBWJooJParkBJoMJChoiSHChoSJet al. Nomogram incorporating CD44v6 and clinicopathological factors to predict lymph node metastasis for early gastric cancer. PloS One (2016) 11(8):e0159424. doi: 10.1371/journal.pone.0159424

  • CrossRef
  • Google Scholar

• 51

  ZhengZZhangYZhangLLiZWuAWuXet al. Nomogram for predicting lymph node metastasis rate of submucosal gastric cancer by analyzing clinicopathological characteristics associated with lymph node metastasis. Chin J Cancer Res = Chung-kuo yen cheng yen chiu (2015) 27(6):572–9. doi: 10.3978/j.issn.1000-9604.2015.12.06

  • CrossRef
  • Google Scholar

• 52

  WangZLiuJLuoYXuYLiuXWeiLet al. Establishment and verification of a nomogram for predicting the risk of lymph node metastasis in early gastric cancer. Rev Espanola Enfermedades Digestivas Organo Oficial la Sociedad Espanola Patologia Digestiva (2021) 113(6):411–7. doi: 10.17235/reed.2020.7102/2020

  • CrossRef
  • Google Scholar

• 53

  ZhengZZhangYZhangLLiZWuXLiuYet al. A nomogram for predicting the likelihood of lymph node metastasis in early gastric patients. BMC Cancer (2016) 16:92. doi: 10.1186/s12885-016-2132-5

  • CrossRef
  • Google Scholar

• 54

  SunZJiangYChenCZhengHHuangWXuBet al. Radiomics signature based on computed tomography images for the preoperative prediction of lymph node metastasis at individual stations in gastric cancer: A multicenter study. Radiother Oncol J Eur Soc Ther Radiol Oncol (2021) 165:179–90. doi: 10.1016/j.radonc.2021.11.003

  • CrossRef
  • Google Scholar

• 55

  NaJELeeYCKimTJLeeHWonHHMinYWet al. Machine learning model to stratify the risk of lymph node metastasis for early gastric cancer: A single-center cohort study. Cancers (2022) 14(5):1121. doi: 10.3390/cancers14051121

  • CrossRef
  • Google Scholar

• 56

  ZhuHWangGZhengJZhuHHuangJLuoEet al. Preoperative prediction for lymph node metastasis in early gastric cancer by interpretable machine learning models: A multicenter study. Surgery (2022) 171(6):1543–51. doi: 10.1016/j.surg.2021.12.015

  • CrossRef
  • Google Scholar

• 57

  FengQXLiuCQiLSunSWSongYYangGet al. An intelligent clinical decision support system for preoperative prediction of lymph node metastasis in gastric cancer. J Am Coll Radiol JACR. (2019) 16(7):952–60. doi: 10.1016/j.jacr.2018.12.017

  • CrossRef
  • Google Scholar

• 58

  WangYLiuWYuYLiuJJXueHDQiYFet al. CT radiomics nomogram for the preoperative prediction of lymph node metastasis in gastric cancer. Eur Radiol (2020) 30(2):976–86. doi: 10.1007/s00330-019-06398-z

  • CrossRef
  • Google Scholar

• 59

  ZhangXPWangZLTangLSunYSCaoKGaoY. Support vector machine model for diagnosis of lymph node metastasis in gastric cancer with multidetector computed tomography: a preliminary study. BMC Cancer (2011) 11:10. doi: 10.1186/1471-2407-11-10

  • CrossRef
  • Google Scholar

• 60

  ZhouCMWangYYeHTYanSJiMLiuPet al. Machine learning predicts lymph node metastasis of poorly differentiated-type intramucosal gastric cancer. Sci Rep (2021) 11(1):1300. doi: 10.1038/s41598-020-80582-w

  • CrossRef
  • Google Scholar

• 61

  LiuSQiaoXXuMJiCLiLZhouZ. Development and validation of multivariate models integrating preoperative clinicopathological parameters and radiographic findings based on late arterial phase CT images for predicting lymph node metastasis in gastric cancer. Acad Radiol (2021) 28 Suppl 1:S167–s78. doi: 10.1016/j.acra.2021.01.011

  • CrossRef
  • Google Scholar

• 62

  JinCJiangYYuHWangWLiBChenCet al. Deep learning analysis of the primary tumour and the prediction of lymph node metastases in gastric cancer. Br J Surgery (2021) 108(5):542–9. doi: 10.1002/bjs.11928

  • CrossRef
  • Google Scholar

• 63

  HuangCHuCZhuJZhangWHuangJZhuZ. Establishment of decision rules and risk assessment model for preoperative prediction of lymph node metastasis in gastric cancer. Front Oncol (2020) 10:1638. doi: 10.3389/fonc.2020.01638

  • CrossRef
  • Google Scholar

• 64

  KimSBaeWJAhnJMHeoJHKimKMChoiKWet al. MicroRNA signatures associated with lymph node metastasis in intramucosal gastric cancer. Modern Pathol (2021) 34(3):672–83. doi: 10.1038/s41379-020-00681-x

  • CrossRef
  • Google Scholar

• 65

  LiuQLiJXinBSunYFengDFulhamMJet al. (18)F-FDG PET/CT radiomics for preoperative prediction of lymph node metastases and nodal staging in gastric cancer. Front Oncol (2021) 11:723345. doi: 10.3389/fonc.2021.723345

  • CrossRef
  • Google Scholar

• 66

  MengLDongDChenXFangMWangRLiJet al. 2D and 3D CT radiomic features performance comparison in characterization of gastric cancer: A multi-center study. IEEE J Biomed Health Informatics (2021) 25(3):755–63. doi: 10.1109/JBHI.2020.3002805

  • CrossRef
  • Google Scholar

• 67

  DongDFangMJTangLShanXHGaoJBGigantiFet al. Deep learning radiomic nomogram can predict the number of lymph node metastasis in locally advanced gastric cancer: an international multicenter study. Ann Oncol (2020) 31(7):912–20. doi: 10.1016/j.annonc.2020.04.003

  • CrossRef
  • Google Scholar

• 68

  LiJDongDFangMWangRTianJLiHet al. Dual-energy CT-based deep learning radiomics can improve lymph node metastasis risk prediction for gastric cancer. Eur Radiol (2020) 30(4):2324–33. doi: 10.1007/s00330-019-06621-x

  • CrossRef
  • Google Scholar

• 69

  TianHNingZZongZLiuJHuCYingHet al. Application of machine learning algorithms to predict lymph node metastasis in early gastric cancer. Front Med (2021) 8:759013. doi: 10.3389/fmed.2021.759013

  • CrossRef
  • Google Scholar

• 70

  BollschweilerEHMönigSPHenslerKBaldusSEMaruyamaKHölscherAH. Artificial neural network for prediction of lymph node metastases in gastric cancer: a phase II diagnostic study. Ann Surg Oncol (2004) 11(5):506–11. doi: 10.1245/ASO.2004.04.018

  • CrossRef
  • Google Scholar

• 71

  ZhangYLiuYZhangJWuXJiXFuTet al. Construction and external validation of a nomogram that predicts lymph node metastasis in early gastric cancer patients using preoperative parameters. Chin J Cancer Res = Chung-kuo yen cheng yen chiu (2018) 30(6):623–32. doi: 10.21147/j.issn.1000-9604.2018.06.07

  • CrossRef
  • Google Scholar

• 72

  GaoXMaTCuiJZhangYWangLLiHet al. A CT-based radiomics model for prediction of lymph node metastasis in early stage gastric cancer. Acad Radiol (2021) 28(6):e155–e64. doi: 10.1016/j.acra.2020.03.045

  • CrossRef
  • Google Scholar

• 73

  WangZKLinJXLiPXieJWWangJBLuJet al. Higher risk of lymph node metastasis in young patients with early gastric cancer. J Cancer (2019) 10(18):4389–96. doi: 10.7150/jca.30260

  • CrossRef
  • Google Scholar

• 74

  ChuYNYuYNJingXMaoTChenYQZhouXBet al. Feasibility of endoscopic treatment and predictors of lymph node metastasis in early gastric cancer. World J Gastroenterol (2019) 25(35):5344–55. doi: 10.3748/wjg.v25.i35.5344

  • CrossRef
  • Google Scholar

• 75

  ChenCQinYChenHZhuDGaoFZhouX. A meta-analysis of the diagnostic performance of machine learning-based MRI in the prediction of axillary lymph node metastasis in breast cancer patients. Insights into Imaging (2021) 12(1):156. doi: 10.1186/s13244-021-01034-1

  • CrossRef
  • Google Scholar

• 76

  LiberatiAAltmanDGTetzlaffJMulrowCGøtzschePCIoannidisJPet al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ (Clinical Res ed). (2009) 339:b2700. doi: 10.1136/bmj.b2700

  • CrossRef
  • Google Scholar

• 77

  MoonsKGde GrootJABouwmeesterWVergouweYMallettSAltmanDGet al. Critical appraisal and data extraction for systematic reviews of prediction modelling studies: the CHARMS checklist. PloS Med (2014) 11(10):e1001744. doi: 10.1371/journal.pmed.1001744

  • CrossRef
  • Google Scholar

• 78

  WolffRFMoonsKGMRileyRDWhitingPFWestwoodMCollinsGSet al. PROBAST: A tool to assess the risk of bias and applicability of prediction model studies. Ann Internal Med (2019) 170(1):51–8. doi: 10.7326/M18-1376

  • CrossRef
  • Google Scholar

• 79

  BalduzziSRückerGSchwarzerG. How to perform a meta-analysis with r: a practical tutorial. Evidence-Based Ment Health (2019) 22(4):153–60. doi: 10.1136/ebmental-2019-300117

  • CrossRef
  • Google Scholar

• 80

  ViechtbauerW. Conducting meta-analyses in R with the metafor package. J Stat Softw (2010) 36(3):1–48. doi: 10.18637/jss.v036.i03

  • CrossRef
  • Google Scholar

• 81

  DersimonianRNanL. Meta-analysis in clinical trials. Controlled Clinical Trials (1986) 7(3):177. doi: 10.1016/0197-2456(86)90046-2

  • CrossRef
  • Google Scholar

• 82

  OnoHYaoKFujishiroMOdaIUedoNNimuraSet al. Guidelines for endoscopic submucosal dissection and endoscopic mucosal resection for early gastric cancer (second edition). Digestive Endoscopy (2021) 33(1):4–20. doi: 10.1111/den.13883

  • CrossRef
  • Google Scholar

• 83

  WangXWanFPanJYuGZChenYWangJJ. Tumor size: a non-neglectable independent prognostic factor for gastric cancer. J Surg Oncol (2008) 97(3):236–40. doi: 10.1002/jso.20951

  • CrossRef
  • Google Scholar

• 84

  AdachiYOshiroTMoriMMaeharaYSugimachiK. Tumor size as a simple prognostic indicator for gastric carcinoma. Ann Surg Oncol (1997) 4(2):137–40. doi: 10.1007/BF02303796

  • CrossRef
  • Google Scholar

• 85

  LiuXXuYLongZZhuHWangY. Prognostic significance of tumor size in T3 gastric cancer. Ann Surg Oncol (2009) 16(7):1875–82. doi: 10.1245/s10434-009-0449-x

  • CrossRef
  • Google Scholar

• 86

  NovotnyARSchuhmacherC. Predicting lymph node metastases in early gastric cancer: radical resection or organ-sparing therapy? Gastric Cancer (2008) 11(3):131–3. doi: 10.1007/s10120-008-0479-2

  • CrossRef
  • Google Scholar

• 87

  OhashiSOkamuraSUranoFMaedaM. Clinicopathological variables associated with lymph node metastasis in submucosal invasive gastric cancer. Gastric Cancer (2007) 10(4):241–50. doi: 10.1007/s10120-007-0442-7

  • CrossRef
  • Google Scholar

• 88

  MaXZhangQZhuSZhangSSunX. Risk factors and prediction model for non-curative resection of early gastric cancer with endoscopic resection and the evaluation. Front Med (2021) 8:637875. doi: 10.3389/fmed.2021.637875

  • CrossRef
  • Google Scholar

• 89

  AbeNWatanabeTSuzukiKMachidaHTodaHNakayaYet al. Risk factors predictive of lymph node metastasis in depressed early gastric cancer. Am J Surgery (2002) 183(2):168–72. doi: 10.1016/S0002-9610(01)00860-1

  • CrossRef
  • Google Scholar

• 90

  ChenSNieRCOuYangLYLiYFXiangJZhouZWet al. Nomogram analysis and external validation to predict the risk of lymph node metastasis in gastric cancer. Oncotarget (2017) 8(7):11380–8. doi: 10.18632/oncotarget.14535

  • CrossRef
  • Google Scholar

• 91

  NakagawaMChoiYYAnJYChungHSeoSHShinHBet al. Difficulty of predicting the presence of lymph node metastases in patients with clinical early stage gastric cancer: a case control study. BMC Cancer (2015) 15:943. doi: 10.1186/s12885-015-1940-3

  • CrossRef
  • Google Scholar

• 92

  Palazón-BruAMares-GarcíaELópez-BruDMares-ArambulEFolgado-de la RosaDMCarbonell-TorregrosaMet al. A critical appraisal of the clinical applicability and risk of bias of the predictive models for mortality and recurrence in patients with oropharyngeal cancer: Systematic review. Head neck (2020) 42(4):763–73. doi: 10.1002/hed.26025

  • CrossRef
  • Google Scholar

• 93

  Di TannaGLWirtzHBurrowsKLGlobeG. Evaluating risk prediction models for adults with heart failure: A systematic literature review. PloS One (2020) 15(1):e0224135. doi: 10.1371/journal.pone.0224135

  • CrossRef
  • Google Scholar

• 94

  NagendranMChenYLovejoyCAGordonACKomorowskiMHarveyHet al. Artificial intelligence versus clinicians: systematic review of design, reporting standards, and claims of deep learning studies. BMJ (Clinical Res ed). (2020) 368:m689. doi: 10.1136/bmj.m689

  • CrossRef
  • Google Scholar

• 95

  DretzkeJChuchuNAgarwalRHerdCChuaWFabritzLet al. Predicting recurrent atrial fibrillation after catheter ablation: a systematic review of prognostic models. Europace Eur Pacing Arrhythmias Cardiac Electrophysiology J Working Groups Cardiac Pacing Arrhythmias Cardiac Cell Electrophysiology Eur Soc Cardiol (2020) 22(5):748–60. doi: 10.1093/europace/euaa041

  • CrossRef
  • Google Scholar

• 96

  BellouVBelbasisLKonstantinidisAKTzoulakiIEvangelouE. Prognostic models for outcome prediction in patients with chronic obstructive pulmonary disease: systematic review and critical appraisal. BMJ (Clinical Res ed). (2019) 367:l5358. doi: 10.1136/bmj.l5358

  • CrossRef
  • Google Scholar

• 97

  BedrikovetskiSDudi-VenkataNNKroonHMSeowWVatherRCarneiroGet al. Artificial intelligence for pre-operative lymph node staging in colorectal cancer: a systematic review and meta-analysis. BMC Cancer (2021) 21(1):1058. doi: 10.1186/s12885-021-08773-w

  • CrossRef
  • Google Scholar

• 98

  Al-JarrahOYYooPDMuhaidatSKaragiannidisGKTahaK. Efficient machine learning for big data: A review. Big Data Res (2015) 2(3):87–93. doi: 10.1016/j.bdr.2015.04.001

  • CrossRef
  • Google Scholar

• 99

  DongDTangLLiZYFangMJGaoJBShanXHet al. Development and validation of an individualized nomogram to identify occult peritoneal metastasis in patients with advanced gastric cancer. Ann Oncol (2019) 30(3):431–8. doi: 10.1093/annonc/mdz001

  • CrossRef
  • Google Scholar

• 100

  XueBJiangJChenLWuSZhengXZhengXet al. Development and validation of a radiomics model based on (18)F-FDG PET of primary gastric cancer for predicting peritoneal metastasis. Front Oncol (2021) 11:740111. doi: 10.3389/fonc.2021.740111

  • CrossRef
  • Google Scholar

• 101

  HuangWZhouKJiangYChenCYuanQHanZet al. Radiomics nomogram for prediction of peritoneal metastasis in patients with gastric cancer. Front Oncol (2020) 10:1416. doi: 10.3389/fonc.2020.01416

  • CrossRef
  • Google Scholar

• 102

  ChenYXiWYaoWWangLXuZWelsMet al. Dual-energy computed tomography-based radiomics to predict peritoneal metastasis in gastric cancer. Front Oncol (2021) 11:659981. doi: 10.3389/fonc.2021.659981

  • CrossRef
  • Google Scholar

• 103

  LiuDZhangWHuFYuPZhangXYinHet al. A bounding box-based radiomics model for detecting occult peritoneal metastasis in advanced gastric cancer: A multicenter study. Front Oncol (2021) 11:777760. doi: 10.3389/fonc.2021.777760

  • CrossRef
  • Google Scholar

• 104

  LiuSHeJLiuSJiCGuanWChenLet al. Radiomics analysis using contrast-enhanced CT for preoperative prediction of occult peritoneal metastasis in advanced gastric cancer. Eur Radiol (2020) 30(1):239–46. doi: 10.1007/s00330-019-06368-5

  • CrossRef
  • Google Scholar

• 105

  MirniaharikandeheiSHeidariMDanalaGLakshmivarahanSZhengB. Applying a random projection algorithm to optimize machine learning model for predicting peritoneal metastasis in gastric cancer patients using CT images. Comput Methods Programs Biomed (2021) 200:105937. doi: 10.1016/j.cmpb.2021.105937

  • CrossRef
  • Google Scholar

• 106

  WangLLvPXueZChenLZhengBLinGet al. Novel CT based clinical nomogram comparable to radiomics model for identification of occult peritoneal metastasis in advanced gastric cancer. Eur J Surg Oncol J Eur Soc Surg Oncol Br Assoc Surg Oncol (2022) 30:S0748-7983(22)00539-X. doi: 10.1016/j.ejso.2022.06.034

  • CrossRef
  • Google Scholar

• 107

  IannattonePAZhaoXVanHoutenJGargAHuynhT. Artificial intelligence for diagnosis of acute coronary syndromes: A meta-analysis of machine learning approaches. Can J Cardiol (2020) 36(4):577–83. doi: 10.1016/j.cjca.2019.09.013

  • CrossRef
  • Google Scholar