A Singapore-based team of clinician-scientists has developed a machine-learning tool that accurately predicts liver cancer recurrence after surgery. By combining clinical data with genetic biomarkers, including a 15-gene signature, the tool achieved an 86% predictive performance score, significantly outperforming the traditional TNM staging system's 56% to 68% range. The researchers also identified two distinct biological pathways of recurrence—polyclonal and monoclonal seeding—enabling highly personalized post-operative surveillance and targeted clinical trials.
Machine learning recurrence prediction
- ▪The Singapore-developed machine-learning tool predicts liver cancer recurrence by combining genetic and clinical data
- ▪A Singapore-based team of clinician-scientists and researchers developed a machine-learning tool that predicts the likelihood of liver cancer recurrence following surgical intervention
- ▪The Singapore-developed recurrence prediction tool was validated across three independent cohorts, including the publicly available Cancer Genome Atlas Liver Hepatocellular Carcinoma dataset
Genetic biomarker integration
- ▪Researchers identified two distinct biological patterns of hepatocellular carcinoma recurrence: polyclonal seeding, associated with intrahepatic recurrence, and monoclonal seeding, associated with distant metastasis
- ▪The Singapore-developed recurrence prediction tool integrates a specific group of 15 genes that are strongly linked to cancer recurrence
Clinical data utilization
- ▪The research team analyzed clinical data and conducted genetic analyses of tumor samples from 106 patients in the PLANet cohort, of whom 68 experienced recurrence
- ▪The Singapore-developed recurrence prediction tool combines clinical indicators including tumor size, blood marker levels, and cancer stage
TNM staging system comparison
- ▪The Singapore-developed machine-learning tool outperformed the conventional TNM staging system in predicting liver cancer recurrence across independent patient groups
- ▪The Singapore-developed recurrence prediction tool achieved a performance score of 86%, compared to 56% to 68% for the conventional TNM staging system
Personalized oncology surveillance
- ▪The Singapore-developed recurrence prediction tool enables clinicians to implement tailored surveillance strategies and targeted adjuvant interventions for high-risk patients
- ▪The research team is working to incorporate CT imaging and machine learning to further improve the accuracy of the multi-omics recurrence prediction tool
Story comments
Loading comments…