Machine learning and treatment outcome prediction for oral cancer

Abstract

Background The natural history of oral squamous cell carcinoma (OSCC) is complicated by progressive disease including loco‐regional tumour recurrence and development of distant metastases. Accurate prediction of tumour behaviour is crucial in delivering individualized treatment plans and developing optimal patient follow‐up and surveillance strategies. Machine learning algorithms may be employed in oncology research to improve clinical outcome prediction. Methods Retrospective review of 467 OSCC patients treated over a 19‐year period facilitated construction of a detailed clinicopathological database. 34 prognostic features from the database were used to populate 4 machine learning algorithms, linear regression (LR), decision tree (DT), support vector machine (SVM) and k‐nearest neighbours (KNN) models, to attempt progressive disease outcome prediction. Principal component analysis (PCA) and bivariate analysis were used to reduce data dimensionality and highlight correlated variables. Models were validated for accuracy, sensitivity and specificity, with predictive ability assessed by receiver operating characteristic (ROC) and area under the curve (AUC) calculation. Results Out of 408 fully characterized OSCC patients, 151 (37%) had died and 131 (32%) exhibited progressive disease at the time of data retrieval. The DT model with 34 prognostic features was most successful in identifying “true positive” progressive disease, achieving 70.59% accuracy (AUC 0.67), 41.98% sensitivity and a high specificity of 84.12%. Conclusion Machine learning models assist clinicians in accessing digitized health information and appear promising in predicting progressive disease outcomes. The future will see increasing emphasis on the use of artificial intelligence to enhance understanding of aggressive tumour behaviour, recurrence and disease progression.

Comparison of machine learning algorithms for the prediction of five‐year survival in oral squamous cell carcinoma

“…It would seem reasonable to consider these metrics as a benchmark given the robust nature of the study. The best performing model, Decision tree classifier from the present study, displayed comparable accuracy (76%) and ROC‐AUC score (0.77) to Karadaghy et al Other recent studies have found Decision tree algorithms to perform strongly when applied to predicting OSCC outcomes 20,21 …”

Predicting Treatment Nonresponse in Hispanic/Latino Children Receiving Silver Diamine Fluoride for Caries Arrest: A Pilot Study Using Machine Learning

“…In this study, our approach considered two methods to yield a predictive model for treatment non-response using a sample of patients who had previously received silver diamine fluoride for dental caries. While research using machine learning in oral epidemiology is limited, previous studies have successfully applied ML-approaches to predict malodor using salivary microbiota [ 23 ], hypertension from gingival inflammation [ 24 ], oral cancer [ 25 ], oral poliovirus vaccine immunogenicity [ 26 ], temporomandibular joint osteoarthritis, and radiographic detection of tooth and dental restorations [ 27 ]. Notably, a recent review concluded that artificial intelligence, particularly machine learning and deep learning, can use data from the oral microbiome to predict systemic disease [ 28 ].…”

Prediction model of laparoendoscopic single-site surgery in gynecology using machine learning algorithm.

“…Machine learning models have achieved rapid development and are widely used in learning modeling and medical case prediction [ 19 – 21 ]. The systems are made up of three major parts, which are: (1) model: the system that makes predictions or identifications; (2) parameters: the signals or factors used by the model to form its decisions; (3) learner: the system that adjusts the parameters – and in turn the model – by looking at differences in predictions versus actual outcome.…”