Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Multicenter Study
. 2025 Feb 21;23(1):98.
doi: 10.1186/s12916-025-03912-7.

Development and external validation of a machine learning-based model to predict postoperative recurrence in patients with duodenal adenocarcinoma: a multicenter, retrospective cohort study

Xu Liu #  1 , Qifeng Xiao #  2 , Zongting Gu  3 , Xin Wu  4 , Chunhui Yuan  5 , Xiaolong Tang  6 , Fanbin Meng  7 , Dong Wang  8 , Ren Lang  9 , Kaiqing Guo  10 , Xiaodong Tian  11 , Yu Zhang  12 , Enhong Zhao  13 , Zhenzhou Wu  14 , Jingyong Xu  15 , Ying Xing  16 , Feng Cao  17 , Chengfeng Wang  18 , Jianwei Zhang  19
Affiliations
Multicenter Study

Development and external validation of a machine learning-based model to predict postoperative recurrence in patients with duodenal adenocarcinoma: a multicenter, retrospective cohort study

Xu Liu et al. BMC Med. .

Abstract

Background: Duodenal adenocarcinoma (DA) has a high recurrence rate, making the prediction of recurrence after surgery critically important.

Methods: Our objective is to develop a machine learning-based model to predict the postoperative recurrence of DA. We conducted a multicenter, retrospective cohort study in China. 1830 patients with DA who underwent radical surgery between 2012 and 2023 were included. Wrapper methods were used to select optimal predictors by ten machine learning learners. Subsequently, these ten learners were utilized for model development. The model's performance was validated using three separate cohorts, and assessed by the concordance index (C-index), time-dependent calibration curve, time-dependent receiver operating characteristic curves, and decision curve analysis.

Results: After selecting predictors, ten feature subsets were identified. And ten feature subsets were combined with the ten machine learning learners in a permutation, resulting in the development of 100 predictive models, and the Penalized Regression + Accelerated Oblique Random Survival Forest model (PAM) exhibited the best predictive performance. The C-index for PAM was 0.882 (95% CI 0.860-0.886) in the training cohort, 0.747 (95% CI 0.683-0.798) in the validation cohort 1, 0.736 (95% CI 0.649-0.792) in the validation cohort 2, and 0.734 (95% CI 0.674-0.791) in the validation cohort 3. A publicly accessible web tool was developed for the PAM.

Conclusions: The PAM has the potential to identify postoperative recurrence in DA patients. This can assist clinicians in assessing the severity of the disease, facilitating patient follow-up, and aiding in the formulation of adjuvant treatment strategies.

Keywords: Cancer recurrence; Duodenal adenocarcinoma; Machine learning-based model; Surgery.

PubMed Disclaimer

Conflict of interest statement

Declarations. Ethics approval and consent to participate: This study adheres to ethical standards and was ensured by conducting the study in compliance with the Declaration of Helsinki (revised in 2013), and ethical approval was secured from the Hospital Ethics Committee of the National Cancer Center (No. 22/213–3415). Written informed consent to participate in the study was obtained from all participants prior to their inclusion. No organs or tissues were obtained from vulnerable groups, including prisoners, in this study. Consent for publication: Not applicable. Competing interests: The authors declare no competing interests.

Figures

Fig. 1
Fig. 1
Flow diagram of the study
Fig. 2
Fig. 2
Concordance index of top 50 machine learning models. The C-index for the top 50 out of 100 machine learning models was calculated for the training cohort and three validation cohorts. Ranking of the models was based on the average C-index of three validation cohorts. AKE, Akritas estimator; GB, Gradient Boosting, GAMB, Boosted Generalized Additive Model; GLMB, Boosted Generalized Linear Model; ST, Survival Tree; CIT, Conditional Inference Tree; RSF, Random Survival Forest; CRF, Conditional Random Forest; AORSF, Accelerated Oblique Random Survival Forest; PR, Penalized Regression; C-index, concordance index
Fig. 3
Fig. 3
Evaluating the calibration of PAM by time-dependent calibration curves. A Training cohort, B validation cohort 1, C validation cohort 2, D Validation cohort 3
Fig. 4
Fig. 4
Evaluating the predictive accuracy of PAM by time-dependent ROC curves. A Training cohort, B validation cohort 1, C validation cohort 2, D validation cohort 3
Fig. 5
Fig. 5
Interpretation the PAM by time-dependent feature importance curves
See this image and copyright information in PMC

References

    1. Severson RK, Schenk M, Gurney JG, Weiss LK, Demers RY. Increasing incidence of adenocarcinomas and carcinoid tumors of the small intestine in adults. Cancer Epidemiol Biomarkers Prev. 1996;5:81–4. - PubMed
    1. Haselkorn T, Whittemore AS, Lilienfeld DE. Incidence of small bowel cancer in the United States and worldwide: geographic, temporal, and racial differences. Cancer Causes Control. 2005;16:781–7. - PubMed
    1. Schottenfeld D, Beebe-Dimmer JL, Vigneau FD. The epidemiology and pathogenesis of neoplasia in the small intestine. Ann Epidemiol. 2009;19:58–69. - PMC - PubMed
    1. Benson AB, Venook AP, Al-Hawary MM, Arain MA, Chen YJ, Ciombor KK, et al. Small bowel adenocarcinoma, version 1.2020, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2019;17:1109–33. - PMC - PubMed
    1. Jensen KK, Storkholm JH, Chen I, Burgdorf SK, Hansen CP. Long-term results after resection of primary duodenal adenocarcinoma: a retrospective cohort study. Int J Surg. 2022;100:106599. - PubMed

LinkOut - more resources