141. Comparative analysis of utilization of artificial intelligence in minimally invasive adult spinal deformity surgery

• Published in: The spine journal Vol. 22; no. 9; pp. S74 - S75
• Main Authors: Passias, Peter G., Tretiakov, Peter, Williamson, Tyler, Krol, Oscar, Imbo, Bailey, Joujon-Roche, Rachel, McFarland, Kimberly, Passfall, Lara, Diebo, Bassel G., Vira, Shaleen N., Smith, Justin S.
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.09.2022
• ISSN: 1529-9430; 1878-1632
• DOI: 10.1016/j.spinee.2022.06.159

Advancements in artificial intelligence (AI), machine learning, and minimally-invasive (MIS) technique may offer enhanced preoperative planning, intraoperative robotic or navigational guidance, and prediction of postoperative complications for adult spinal deformity patients. Despite relatively widespread utilization, few studies in the literature assess the clinical and radiographic impact of AI in MIS surgery. To assess the impact of artificial intelligence on peri- and postoperative course in minimally-invasive adult spinal deformity corrective surgery. Retrospective cohort review. This study included 524 MIS patients. Intra- and postoperative complication rates; reoperation rate; HRQLs Operative cervical deformity patients 18 years old with complete pre-(BL) and up to 2-year (2Y) postop radiographic/HRQL data were stratified by primary utilization AI-based patient-specific rod customization and robotic or navigational assistance in pre- and perioperative course (AI+) or not (AI-). Differences in demographics, clinical outcomes, radiographic alignment targets, perioperative factors and complication rates were assessed via means comparison analysis. Analysis of covariance (ANCOVA) assessed postoperative complications while controlling for BL age and gender. A total of 133 MIS patients were included (51.74±11.59 years, 41% female, 30.85±6.93 kg/m2). Of these patients, 44 (33.1%) were classified as AI+. At baseline, patient groups were comparable in BL age, BMI and CCI (all p>.05), though AI+ patients were more likely to be male (p=.040). Patient groups were comparable in terms of both regional and global radiographic alignment, as well as HRQLs at BL (all p>.05). Surgically, AI+ patients had significantly shorter operative times overall (p=.022) and decreased EBL (p=.001), as well as decreased likelihood of undergoing corpectomy (p=.001). Furthermore, AI+ patients reported significantly lower hospital LOS vs AI- patients (p=.012). At 2 years postoperatively, AI+ patients with custom rods were noted to have significantly improved segmental alignment in terms of decreased pelvic tilt (S1PT) and pelvic incidence (S1PI) (both p <.001). Adjusted complications analysis revealed that AI+ patients were significantly less likely to experience any postoperative complication (p=.003), neurological complications (p=.021) or complication requiring reoperation (p=.003). Artificial intelligence and machine learning technologies may provide a substantial benefit to patients undergoing minimally-invasive adult spinal deformity surgery. The findings in this study demonstrate that patients operated on using AI-based robotic or navigational guidance, as well as the utilization of customized instrumentation, may reduce intraoperative invasiveness, shorten hospital length of stay, and decrease complication rates. As such, surgeons should consider utilization of AI-based technology in practice. This abstract does not discuss or include any applicable devices or drugs.