Prediction of Patient Height and Weight With a 3-Dimensional Camera

Abstract

Objective: The aim of this study was to determine accuracy of height and weight prediction by a 3-dimensional (3D) camera.

Methods: A total of 453 patients whose computed tomography imaging used a 3D camera from December 19, 2018 to March 19, 2019 were retrospectively identified. An image of each patient was taken before the computed tomography by a 3D camera mounted to the ceiling. Using infrared imaging and machine learning algorithms, patient height and weight were estimated from this 3D camera image. A total of 363 images were used for training. The test set consisted of 90 images. The height and weight estimates were compared with true height and weight to determine absolute and percent error. A value of P < 0.05 indicated statistical significance.

Results: There was 2.0% (SD, 1.4) error in height estimation by the 3D camera, corresponding to 3.35 cm (SD, 2.39) absolute deviation (P = 1, n = 86). Weight estimation error was 5.1% (SD, 4.3), corresponding to 3.99 kg (SD, 3.11) absolute error (P = 0.74, n = 90).

Conclusion: Pictures obtained from a 3D camera can accurately predict patient height and weight.