Factors influencing the estimation of phacoemulsification procedure time in cataract surgery: Analysis using neural networks

Abstract

Background and Objective

Realistic and accurate estimation of the surgery duration is one of the key factors influencing the optimization of hospital work and, consequently, the planning and management of the budget. In the present study, the authors proposed a method for predicting the phacoemulsification cataract surgery based on ophthalmic and systemic factors.

Methods

The study group included 1192 patients aged 70.4 ± 10 years who underwent phacoemulsification cataract surgery. The surgical procedures were performed by both experienced surgeons and trainees (15 % of procedures). 25 parameters were extracted, on the basis of which, using neural networks with backpropagation, an algorithm was proposed that predicted the surgery duration based on a set of input features.

Results

For the proposed method, the mean absolute error between the actual and predicted operation time was 5.09 min, whereas the accuracy of the obtained results was 69.74 % (for the best set of 7 input features).

Conclusions

The obtained results indicate that machine learning algorithms can be successfully used to predict the time of cataract surgery, and factors such as: surgeon's experience, patient's visual acuity (UCVA), intraocular pressure (IOP), corneal curvature and sphere value (SF) significantly influence the phacoemulsification cataract surgery duration.