ASO Author Reflections: Development and Validation of a Novel Risk Score Using Machine-Learning Methodology to Predict Recurrence After Hepatectomy for Colorectal Liver Metastases

Excerpt

Despite significant improvement in the management of colorectal liver metastases (CRLM), a need still remains to identify patients likely to obtain maximum therapeutic benefit from a hepatectomy. Traditional prognostication tools have weighed factors such as carcinoembryonic antigen level, size of largest hepatic tumor, number of hepatic tumors, disease-free interval from resection of primary tumor to development of metastases, presence of metastatic nodes, and KRAS status equally, which may inaccurately estimate the prognostic impact of these variables on patient outcomes, specifically recurrence.1,2 As such, the current study aimed to use bootstrap resampling methodology in tandem with multivariable mixed-effects logistic regression analysis to construct a CRLM recurrence prediction model. The model was subsequently validated and compared with previously proposed scores.1,2 …