Forecasting the performance status of head and neck cancer patient treatment by an interval arithmetic pruned perceptron.

Abstract

The integration of chemotherapy and radiotherapy for the treatment of advanced head and neck cancer is still a matter of clinical investigation. An important limitation is that the concomitant administration of chemotherapy and radiotherapy still induces severe toxicity. In this paper, a simple artificial neural network is used to predict, on the basis of biological and clinical data, if the cumulative toxicity of the combined chemo-radiation treatment itself would be tolerated. The resulting method, tested on clinical data from a phase II trial, proved to be able to forecast which patients will tolerate a combined chemo-radiotherapeutic approach. This result should open a new perspective in the clinical approach, by supplying a potential predictive indicator for toxicity.