TY - JOUR
T1 - Artificial Intelligence Based Hierarchical Clustering of Patient Types and Intervention Categories in Adult Spinal Deformity Surgery
T2 - Towards a New Classification Scheme that Predicts Quality and Value
AU - Ames, Christopher P.
AU - Smith, Justin S.
AU - Pellisé, Ferran
AU - Kelly, Michael
AU - Alanay, Ahmet
AU - Acaroǧlu, Emre
AU - Pérez-Grueso, Francisco Javier Sánchez
AU - Kleinstück, Frank
AU - Obeid, Ibrahim
AU - Vila-Casademunt, Alba
AU - Burton, Douglas
AU - Lafage, Virginie
AU - Schwab, Frank
AU - Shaffrey, Christopher I.
AU - Bess, Shay
AU - Serra-Burriel, Miquel
N1 - Funding Information:
From the *Department of Neurosurgery, University of California San Francisco, San Francisco, California; †Department of Neurosurgery, University of Virginia Medical Center, Charlottesville, Virginia; zSpine Surgery Unit, Hospital Vall d’Hebron, Barcelona, Spain; §Department of Orthopaedic Surgery, Washington University, St Louis, Missouri; {Department of Orthopedics and Traumatology, Acibadem University, Istanbul, Turkey; ||Ankara Spine Center, Ankara, Turkey;**Spine Surgery Unit, Hospital Universitario La Paz, Madrid, Spain; ††Spine Center Division, Department of Orthopedics and Neurosurgery, Schulthess Klinik, Zurich, Switzerland; zzSpine Surgery Unit, Bordeaux University Hospital, Bordeaux, France; §§Vall d’Hebron Institute of Research (VHIR) Barcelona, Spain; {{Department of Orthopaedic Surgery, University of Kansas Medical Center, Kansas City, Kansas; ||||Department of Orthopaedic Surgery, Hospital for Special Surgery, New York, New York; ***Denver International Spine Center, Presbyterian St. Luke’s/Rocky Mountain Hospital for Children, Denver, Colorado; and †††Center for Research in Health and Economics, Universitat Pompeu Fabra, Barcelona, Spain. Acknowledgment date: July 16, 2018. First revision date: November 12, 2018. Acceptance date: December 13, 2018. The manuscript submitted does not contain information about medical device(s)/drug(s). The International Spine Study Group Foundation receives funding support from DePuy Synthes, K2M, Nuvasive, Orthofix, and Zimmer Biomet. The European Spine Study Group receives funding support from DePuy Synthes and Medtronic. Additional support was provided through Project PI16/ 01283, funded by Instituto de Salud Carlos III and co-funded by European Union (ERDF/ESF). Relevant financial activities outside the submitted work: board membership, consultancy, grants, stocks, royalties, payment for lecture. Address correspondence and reprint requests to Miquel Serra-Burriel, PhD, Center for Research in Health Economics, Department of Economics and Business, Universitat Pompeu Fabra, Office 23.111 Merce Rodoreda Building (Ciutadella Campus), Ramon Trias Fargas, 25-27, Barcelona 08005, Spain; E-mail: miquel.serrab@upf.edu
Publisher Copyright:
© 2019 Lippincott Williams and Wilkins. All rights reserved.
PY - 2019/7
Y1 - 2019/7
N2 - Study Design.Retrospective review of prospectively-collected, multicenter adult spinal deformity (ASD) databases.Objective.To apply artificial intelligence (AI)-based hierarchical clustering as a step toward a classification scheme that optimizes overall quality, value, and safety for ASD surgery.Summary of Background Data.Prior ASD classifications have focused on radiographic parameters associated with patient reported outcomes. Recent work suggests there are many other impactful preoperative data points. However, the ability to segregate patient patterns manually based on hundreds of data points is beyond practical application for surgeons. Unsupervised machine-based clustering of patient types alongside surgical options may simplify analysis of ASD patient types, procedures, and outcomes.Methods.Two prospective cohorts were queried for surgical ASD patients with baseline, 1-year, and 2-year SRS-22/Oswestry Disability Index/SF-36v2 data. Two dendrograms were fitted, one with surgical features and one with patient characteristics. Both were built with Ward distances and optimized with the gap method. For each possible n patient cluster by m surgery, normalized 2-year improvement and major complication rates were computed.Results.Five hundred-seventy patients were included. Three optimal patient types were identified: Young with coronal plane deformity (YC, n=195), older with prior spine surgeries (ORev, n=157), and older without prior spine surgeries (OPrim, n=218). Osteotomy type, instrumentation and interbody fusion were combined to define four surgical clusters. The intersection of patient-based and surgery-based clusters yielded 12 subgroups, with major complication rates ranging from 0% to 51.8% and 2-year normalized improvement ranging from-0.1% for SF36v2 MCS in cluster [1,3] to 100.2% for SRS self-image score in cluster [2,1].Conclusion.Unsupervised hierarchical clustering can identify data patterns that may augment preoperative decision-making through construction of a 2-year risk-benefit grid. In addition to creating a novel AI-based ASD classification, pattern identification may facilitate treatment optimization by educating surgeons on which treatment patterns yield optimal improvement with lowest risk.Level of Evidence: 4.
AB - Study Design.Retrospective review of prospectively-collected, multicenter adult spinal deformity (ASD) databases.Objective.To apply artificial intelligence (AI)-based hierarchical clustering as a step toward a classification scheme that optimizes overall quality, value, and safety for ASD surgery.Summary of Background Data.Prior ASD classifications have focused on radiographic parameters associated with patient reported outcomes. Recent work suggests there are many other impactful preoperative data points. However, the ability to segregate patient patterns manually based on hundreds of data points is beyond practical application for surgeons. Unsupervised machine-based clustering of patient types alongside surgical options may simplify analysis of ASD patient types, procedures, and outcomes.Methods.Two prospective cohorts were queried for surgical ASD patients with baseline, 1-year, and 2-year SRS-22/Oswestry Disability Index/SF-36v2 data. Two dendrograms were fitted, one with surgical features and one with patient characteristics. Both were built with Ward distances and optimized with the gap method. For each possible n patient cluster by m surgery, normalized 2-year improvement and major complication rates were computed.Results.Five hundred-seventy patients were included. Three optimal patient types were identified: Young with coronal plane deformity (YC, n=195), older with prior spine surgeries (ORev, n=157), and older without prior spine surgeries (OPrim, n=218). Osteotomy type, instrumentation and interbody fusion were combined to define four surgical clusters. The intersection of patient-based and surgery-based clusters yielded 12 subgroups, with major complication rates ranging from 0% to 51.8% and 2-year normalized improvement ranging from-0.1% for SF36v2 MCS in cluster [1,3] to 100.2% for SRS self-image score in cluster [2,1].Conclusion.Unsupervised hierarchical clustering can identify data patterns that may augment preoperative decision-making through construction of a 2-year risk-benefit grid. In addition to creating a novel AI-based ASD classification, pattern identification may facilitate treatment optimization by educating surgeons on which treatment patterns yield optimal improvement with lowest risk.Level of Evidence: 4.
KW - adult spinal deformity
KW - artificial intelligence
KW - classification
KW - complications
KW - hierarchical clustering
KW - outcomes
KW - predictive analytics
KW - quality
KW - scoliosis
KW - surgery
UR - http://www.scopus.com/inward/record.url?scp=85065059857&partnerID=8YFLogxK
U2 - 10.1097/BRS.0000000000002974
DO - 10.1097/BRS.0000000000002974
M3 - Article
C2 - 31205167
AN - SCOPUS:85065059857
SN - 0362-2436
VL - 44
SP - 915
EP - 926
JO - Spine
JF - Spine
IS - 13
ER -