Subsite-specific prediction of lymphatic spread in head-and-neck cancer using a mixture of hidden Markov models
The study develops a statistical model that estimates the risk of occult lymph node metastases in head-and-neck squamous cell carcinoma patients based on clinical lymph node involvement, T-stage, and tumor subsite. Using a mixture of hidden Markov models trained on 2437 patients from 13 subsites, the model identifies distinct spread patterns for anterior oral cavity, oropharynx, hypopharynx, and glottic larynx, and predicts low risk of metastases in specific lymph node levels for certain subsites. The framework offers interpretable, individualized predictions that could inform personalized el…
The study develops a statistical model that estimates the risk of occult lymph node metastases in head-and-neck squamous cell carcinoma patients based on clinical lymph node involvement, T-stage, and tumor subsite. Using a mixture of hidden Markov models trained on 2437 patients from 13 subsites, the model identifies distinct spread patterns for anterior oral cavity, oropharynx, hypopharynx, and glottic larynx, and predicts low risk of metastases in specific lymph node levels for certain subsites. The framework offers interpretable, individualized predictions that could inform personalized el…
This research demonstrates how advanced probabilistic modeling can personalize cancer treatment, aligning with Swiss AI initiatives that aim to integrate data-driven decision support into clinical workflows.
Original source record. Open the original record to verify the underlying announcement.