Abstract—The prediction of the occurrence of short-term adverse events in phototherapy treatments is important for dermatologists who administrate phototherapy to adjust the treatment and standardize the clinical outcomes. Recently, a modeling technique that can detect the potential occurrence of short-term adverse events in phototherapy treatments is required for clinicians. Based on data mining, this study tends to explore the significant features and the class distribution of training data for predicting the occurrence of short-term adverse events in NB-UVB phototherapy treatments. The experimental results highlight that an acceptable prediction accuracy can be achieved using the significant features and the performance of the classifiers can be significantly improved by sampling 40% of the negative class samples in the training data, hyper-parameter tuning of the classifiers and use of stacked classifiers in creating the prediction models.
Index Terms—Adverse events, classification, data mining, dermatology, phototherapy, prediction.
S. Mohamed and M-T. Kechadi are with University College Dublin, Belfield, Dublin 4 (email@example.com, firstname.lastname@example.org, email@example.com).
Anne-Marie Tobin is with Dermatology Department, Tallagh Hospital.
Alan D. Irvine is with School of Medicine, Trinity College Dublin.
Dmitri R Wall is with Irish Skin Foundation. Neil J. O’Hare is with School of Public Health, Physiotherapy & Sports, University College Dublin.
Cite: S. Mohamed, A-M. Tobin, A. D. Irvine, D. R. Wall, N. J. O’Hare, and M-T. Kechadi, "The Application of Data Mining to Predict the Occurrence of Short-Term Adverse Events in NB-UVB Phototherapy Treatments," International Journal of Machine Learning and Computing vol. 8, no. 2, pp. 104-111, 2018.