Artificial Intelligence Predicts Health-Related Quality of Life for Adolescent Idiopathic Scoliosis

PurposeAdolescent idiopathic scoliosis (AIS) has a large impact on health-related quality of life (HRQoL) including poor psychosocial functioning, body image, and psychological distress. Surgical management for AIS is common; however, there is limited consensus on preoperative and intraoperative strategies to optimize HRQoL outcomes. Accurate prediction of postoperative outcomes can help guide operative planning and lead to improved HRQoL. This study aimed to generate machine learning models (MLMs) using preoperative and intraoperative variables to predict the difference in HRQoL outcomes from preoperative assessment to two years following AIS surgery. MethodsA prospective, longitudinal, multicenter database was queried for AIS patients of Lenke 1 or 5 classification with two-year follow-up. MLMs were generated using preoperative and intraoperative variables to predict the difference in Scoliosis Research Society-22 ({Delta}SRS-22) questionnaire scores from preoperative assessment to two-year follow-up. MLMs were compared to a model that estimates the mean score by evaluating the mean squared error (MSE) and the fraction of times the prediction was within a predesignated value of the actual score (i.e., buffer accuracy). ResultsA total of 1,477 patients (84.6% female, 75.0% White) were included. The lowest MSE for each {Delta}SRS-22 outcome ranged from 0.18-0.48, while the highest 0.25-buffer, 0.5-buffer, 0.75-buffer, and 1-buffer accuracies for each {Delta}SRS-22 outcome ranged from 34.8%-53.4%, 56.8%-83.1%, 75.0%-94.6%, and 87.2%-97.3%, respectively. These MSEs and buffer accuracies outperformed mean estimates. ConclusionMLMs built using preoperative and intraoperative variables enabled prediction of the difference in HRQoL outcomes from preoperative assessment to two years following AIS surgery. Findings provide key insights into the feasibility of implementing MLMs to guide operative planning and counsel patients on expected outcomes of surgical management. Future work should implement additional surgeon, institution, and patient factors as model predictors to increase predictive accuracy of HRQoL outcomes and ultimately improve individualized patient care through data-driven surgical planning.