Estimating apparent age using artificial intelligence: Quantifying the effect of blepharoplasty
Kendall Goodyear, Persiana S. Saffari, Mahtash Esfandiari, Samuel Baugh, Daniel B. Rootman, Justin N. Karlin
Abstract
ObjectivesQuantify the rejuvenation effect of blepharoplasty.MethodsA photographic dataset of human faces was assembled and randomly split in to 90% training and 10% validation sets. A deep learning model was trained to input a facial photograph and to infer age. Retrospective chart review of patients who underwent blepharoplasty was used to assemble a test set—preoperative and postoperative photographs were subsequently analyzed by the model.Results47,394 images of patients aged 26 to 89 years old were used for model training and validation. On the validation set, the model achieved 75% accuracy with a mean absolute error of 1.38 years and Pearson’s r of 0.92. A total of 103 patients (29 males and 74 females) met test set inclusion criteria (upper blepharoplasty n = 28, lower blepharoplasty n = 33 and quadrilaterarl blepharoplasty n = 42). Test set age ranged from 30.3 to 83.8 years old (mean 60.8, standard deviation 11.4). Overall, the model predicted test set patients to be 0.74 years younger preoperatively versus 2.52 years younger postoperatively (p < 0.01). Significant underestimation of age was observed in women who underwent lower blepharoplasty (n = 23, 1.28 y older pre vs. 2.32 y younger post, p = 3.8 ×10-4) and men who underwent quadrilateral blepharoplasty (n = 10, 0.71 y younger pre vs. 5.34 y younger post, p = 0.02).ConclusionsThe deep learning algorithm developed in this study demonstrates that, on average, blepharoplasty provides a rejuvenating effect of approximately two years.