The use of artificial intelligence to identify people at risk of oral cancer and precancer

摘要: Artificial intelligence is being used increasingly as an aid to diagnosis in medicine. The purpose of this study was evaluate the ability a neural network predict likelihood individual having malignant or potentially oral lesion based on knowledge their risk habits. Performance compared with group dental screeners screening programme involving 2027 adults. performance measured terms sensitivity, specificity and ratios. All subjects were examined independently by screener specialist, who provided definitive diagnosis, 'gold standard', for each individual. also completed interview questionnaire regarding personal details, attendance smoking drinking trained 1662 screened population using ten input variables derived from along outcome specialist's diagnosis. Following training, asked classify remaining unseen proportion (365 individuals) positive negative presence cancer precancer. overall sensitivity dentists 0.74 [95% confidence interval (CI), 0.62-0.86] 0.99 (95% CI, 0.985-0.994) respectively 0.80 (99% 0.55-1.00) 0.77 0.73-0.81) network. In view potential costs involved implementing programme, may be value identification individuals high precancer further clinical examination health education.