Artificial intelligence in rectal cancer: What is the future?

Table 1 Published studies on artificial intelligence and machine learning in rectal carcinoma

S.No.	Ref.	No. of patients in AI-based study	Methods	Result	Conclusion
1	Pham et al[29], 2023	N = 53, rectal cancer biopsy	CNN based extraction of IHC images	SVMs extraction; total accuracy = 85%, Prediction of survival rate of more than 5 yr = 90%, and less than 5 yr = 75%	Use of AI can be informative for clinical decision making- whether required preoperative therapy or not
2	Kim et al[67], 2023	N = 39, mid to lower rectal cancer patients who underwent chemoradiotherapy	Deep learning-based imaging reconstruction (DLR) effect on MRI quality	Compared to conventional MRI DLR-MRI showed significantly higher specificity values (P < 0.036)	Compared to conventional MRI, DLR significantly increased the specificity of MRI for identifying pathological complete response (pCR)
3	Wang et al[68], 2023	N = 1651, machine learning model used for predicting major LARS following laparoscopic surgery of rectal cancer and their quality of life	The trained random forest (RF) model performed, and clinical utility of the model was tested by decision curve analysis	Compared to the conventional preoperative LARS score model, current machine learning model exhibited superior predictive performance in predicting major LARS	This model could potentially be used in the clinic to identify patients with a high risk of developing major LARS and then improve the quality of life
4	Qiu et al[69], 2023	N = 27180, used eight machine learning Model for predicting chances lung metastasis in rectal cancer patients	They used DCA and calibration analysis to test all the models to predict risk of lung metastasis in patients with rectal cancer	XGB model had better clinical decision making and prediction ability than other models	XGB model based on clinicopathological information to predict the risk of lung metastasis in patients with rectal cancer, which may help physicians make clinical decisions
5	Shao et al[39], 2023	N = 2469, consecutive patients with stage I-III rectal adenocarcinoma who received anterior resection and did not receive neoadjuvant therapy	Five AI algorithms, (SVM), logistic regression (LR), Naive Bayes (NB), stochastic gradient descent (SGD) and random forest (RF), were employed to generate five models	In summary, the present study developed a high-performance AI model based on clinical preoperative and intraoperative les, which may be supportive for the guidance of the intraoperative decision-making by calculating the risk of AL	The application of this app can predict the risk of AL in patients with rectal cancer who have undergone anterior resection
6	Xia et al[70], 2023	172 rectal cancer patients were used for model training, and 18 patients were used for model validation. Another 40 rectal cancer patients were used for an end-to-end evaluation for both auto-segmentation and treatment planning	The PTV and OAR segmentation was compared with manual segmentation	The PTV DICE similarity coefficient was greater than 0.85 for all 40 patients in the evaluation dataset while the DICE indices of the OARs also indicated good performance	Deep learning-based automatic solution for rectal cancer treatment that can improve the efficiency of treatment planning

AI: Artificial intelligence; AL: Anastomotic leak; CNN: Artificial Neural Network; DLR: Deep learning; SVM: Linear support vector machine; LARS: Low anterior resection syndrome; XGB: Extreme gradient boosting; DICE: Diverse counterfactual explanation; OAR: Object attribute relation; PTV: Particle tracking velocimetry; IHC: Immunohistochemistry.