TY - JOUR
T1 - Machine learning approach in predicting clinically significant improvements after surgery in patients with cervical ossification of the posterior longitudinal ligament
AU - Maki, Satoshi
AU - Furuya, Takeo
AU - Yoshii, Toshitaka
AU - Egawa, Satoru
AU - Sakai, Kenichiro
AU - Kusano, Kazuo
AU - Nakagawa, Yukihiro
AU - Hirai, Takashi
AU - Wada, Kanichiro
AU - Katsumi, Keiichi
AU - Fujii, Kengo
AU - Kimura, Atsushi
AU - Nagoshi, Narihito
AU - Kanchiku, Tsukasa
AU - Nagamoto, Yukitaka
AU - Oshima, Yasushi
AU - Ando, Kei
AU - Takahata, Masahiko
AU - Mori, Kanji
AU - Nakajima, Hideaki
AU - Murata, Kazuma
AU - Matsunaga, Shunji
AU - Kaito, Takashi
AU - Yamada, Kei
AU - Kobayashi, Sho
AU - Kato, Satoshi
AU - Ohba, Tetsuro
AU - Inami, Satoshi
AU - Fujibayashi, Shunsuke
AU - Katoh, Hiroyuki
AU - Kanno, Haruo
AU - Imagama, Shiro
AU - Koda, Masao
AU - Kawaguchi, Yoshiharu
AU - Takeshita, Katsushi
AU - Matsumoto, Morio
AU - Ohtori, Seiji
AU - Yamazaki, Masashi
AU - Okawa, Atsushi
N1 - Publisher Copyright:
© 2021 Wolters Kluwer Health, Inc. All rights reserved.
PY - 2021/12/15
Y1 - 2021/12/15
N2 - Study Design. A retrospective analysis of prospectively collected data. Objective. This study aimed to create a prognostic model for surgical outcomes in patients with cervical ossification of the posterior longitudinal ligament (OPLL) using machine learning (ML). Summary of Background Data. Determining surgical outcomes helps surgeons provide prognostic information to patients and manage their expectations. ML is a mathematical model that finds patterns from a large sample of data and makes predictions outperforming traditional statistical methods. Methods. Of 478 patients, 397 and 370 patients had complete follow-up information at 1 and 2years, respectively, and were included in the analysis. A minimal clinically important difference (MCID) was defined as an acquired Japanese Orthopedic Association (JOA) score of ≥2.5 points, after which a ML model that predicts whether MCID can be achieved 1 and 2years after surgery was created. Patient background, clinical symptoms, and imaging findings were used as variables for analysis. The ML model was created using LightGBM, XGBoost, random forest, and logistic regression, after which the accuracy and area under the receiver-operating characteristic curve (AUC) were calculated. Results. The mean JOA score was 10.3 preoperatively, 13.4 at 1year after surgery, and 13.5 at 2years after surgery. XGBoost showed the highest AUC (0.72) and high accuracy (67.8) for predicting MCID at 1year, whereas random forest had the highest AUC (0.75) and accuracy (69.6) for predicting MCID at 2years. Among the included features, total preoperative JOA score, duration of symptoms, body weight, sensory function of the lower extremity sub-score of the JOA, and age were identified as having the most significance in most of ML models. Conclusion. Constructing a prognostic ML model for surgical outcomes in patients with OPLL is feasible, suggesting the potential application of ML for predictive models of spinal surgery.
AB - Study Design. A retrospective analysis of prospectively collected data. Objective. This study aimed to create a prognostic model for surgical outcomes in patients with cervical ossification of the posterior longitudinal ligament (OPLL) using machine learning (ML). Summary of Background Data. Determining surgical outcomes helps surgeons provide prognostic information to patients and manage their expectations. ML is a mathematical model that finds patterns from a large sample of data and makes predictions outperforming traditional statistical methods. Methods. Of 478 patients, 397 and 370 patients had complete follow-up information at 1 and 2years, respectively, and were included in the analysis. A minimal clinically important difference (MCID) was defined as an acquired Japanese Orthopedic Association (JOA) score of ≥2.5 points, after which a ML model that predicts whether MCID can be achieved 1 and 2years after surgery was created. Patient background, clinical symptoms, and imaging findings were used as variables for analysis. The ML model was created using LightGBM, XGBoost, random forest, and logistic regression, after which the accuracy and area under the receiver-operating characteristic curve (AUC) were calculated. Results. The mean JOA score was 10.3 preoperatively, 13.4 at 1year after surgery, and 13.5 at 2years after surgery. XGBoost showed the highest AUC (0.72) and high accuracy (67.8) for predicting MCID at 1year, whereas random forest had the highest AUC (0.75) and accuracy (69.6) for predicting MCID at 2years. Among the included features, total preoperative JOA score, duration of symptoms, body weight, sensory function of the lower extremity sub-score of the JOA, and age were identified as having the most significance in most of ML models. Conclusion. Constructing a prognostic ML model for surgical outcomes in patients with OPLL is feasible, suggesting the potential application of ML for predictive models of spinal surgery.
KW - Artificial intelligence
KW - Cervical spine
KW - Machine learning
KW - Myelopathy
KW - Ossification of the posterior longitudinal ligament
KW - Prognosis
KW - Spinal cord
KW - Surgical outcomes
UR - http://www.scopus.com/inward/record.url?scp=85120708680&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85120708680&partnerID=8YFLogxK
U2 - 10.1097/BRS.0000000000004125
DO - 10.1097/BRS.0000000000004125
M3 - Article
C2 - 34027925
AN - SCOPUS:85120708680
SN - 0362-2436
VL - 46
SP - 1683
EP - 1689
JO - Spine
JF - Spine
IS - 24
ER -