Developing Artificial Neural Network Models to Predict Functioning One Year After Traumatic Spinal Cord Injury



      To develop mathematical models for predicting level of independence with specific functional outcomes 1 year after discharge from inpatient rehabilitation for spinal cord injury.


      Statistical analyses using artificial neural networks and logistic regression.


      Retrospective analysis of data from the national, multicenter Spinal Cord Injury Model Systems (SCIMS) Database.


      Subjects (N=3142; mean age, 41.5y) with traumatic spinal cord injury who contributed data for the National SCIMS Database longitudinal outcomes studies.


      Not applicable.

      Main Outcome Measures

      Self-reported ambulation ability and FIM-derived indices of level of assistance required for self-care activities (ie, bed-chair transfers, bladder and bowel management, eating, toileting).


      Models for predicting ambulation status were highly accurate (>85% case classification accuracy; areas under the receiver operating characteristic curve between .86 and .90). Models for predicting nonambulation outcomes were moderately accurate (76%–86% case classification accuracy; areas under the receiver operating characteristic curve between .70 and .82). The performance of models generated by artificial neural networks closely paralleled the performance of models analyzed using logistic regression constrained by the same independent variables.


      After further prospective validation, such predictive models may allow clinicians to use data available at the time of admission to inpatient spinal cord injury rehabilitation to accurately predict longer-term ambulation status, and whether individual patients are likely to perform various self-care activities with or without assistance from another person.


      List of abbreviations:

      AIS (American Spinal Injury Association Impairment Scale), ANN (artificial neural network), ASIA (American Spinal Injury Association), AUC (area under the curve), NLR (negative likelihood ratio), PLR (positive likelihood ratio), SCI (spinal cord injury), SCIMS (Spinal Cord Injury Model Systems)
      To read this article in full you will need to make a payment

      Purchase one-time access:

      Academic & Personal: 24 hour online accessCorporate R&D Professionals: 24 hour online access
      One-time access price info
      • For academic or personal research use, select 'Academic and Personal'
      • For corporate R&D use, select 'Corporate R&D Professionals'


      Subscribe to Archives of Physical Medicine and Rehabilitation
      Already a print subscriber? Claim online access
      Already an online subscriber? Sign in
      Institutional Access: Sign in to ScienceDirect


        • Kirshblum S.C.
        • Priebe M.M.
        • Ho C.H.
        • Scelza W.M.
        • Chiodo A.E.
        • Wuermser L.A.
        Spinal cord injury medicine. 3. Rehabilitation phase after acute spinal cord injury.
        Arch Phys Med Rehabil. 2007; 88: S62-S70
        • Ditunno J.F.
        Predicting recovery after spinal cord injury: a rehabilitation imperative.
        Arch Phys Med Rehabil. 1999; 80: 361-364
        • Anderson K.
        Targeting recovery: priorities of the spinal cord injured population.
        J Neurotrauma. 2004; 10: 1371-1383
        • Crozier K.S.
        • Graziani V.
        • Ditunno Jr., J.F.
        • Herbison G.J.
        Spinal cord injury: prognosis for ambulation based on sensory examination in patients who are initially motor complete.
        Arch Phys Med Rehabil. 1991; 72: 119-121
        • Crozier K.S.
        • Cheng L.L.
        • Graziani V.
        • Zorn G.
        • Herbison G.
        • Ditunno J.F.
        Spinal cord injury: prognosis for ambulation based on quadriceps recovery.
        Spinal Cord. 1992; 30: 762-767
        • Waters R.L.
        • Yakura J.S.
        • Adkins R.H.
        • Sie I.
        Recovery following complete paraplegia.
        Arch Phys Med Rehabil. 1992; 73: 784-789
        • Waters R.L.
        • Adkins R.H.
        • Yakura J.S.
        • Sie I.
        Motor and sensory recovery following complete tetraplegia.
        Arch Phys Med Rehabil. 1993; 74: 242-247
        • Waters R.L.
        • Adkins R.
        • Joy Yakura R.P.
        • Vigil D.
        Prediction of ambulatory performance based on motor scores derived from standards of the American Spinal Injury Association.
        Arch Phys Med Rehabil. 1994; 75: 756-760
        • Burns S.P.
        • Golding D.G.
        • Rolle W.A.
        • Graziani V.
        • Ditunno J.F.
        Recovery of ambulation in motor-incomplete tetraplegia.
        Arch Phys Med Rehabil. 1997; 78: 1169-1172
        • Marino R.J.
        • Ditunno J.F.
        • Donovan W.H.
        • Maynard F.
        Neurologic recovery after traumatic spinal cord injury: data from the Model Spinal Cord Injury Systems.
        Arch Phys Med Rehabil. 1999; 80: 1391-1396
        • Scivoletto G.
        • Tamburella F.
        • Laurenza L.
        • Torre M.
        • Molinari M.
        Who is going to walk? A review of the factors influencing walking recovery after spinal cord injury.
        Front Hum Neurosci. 2014; 8: 1-11
        • Kirshblum S.C.
        • Burns S.P.
        • Biering-Sorensen F.
        • et al.
        International standards for neurological classification of spinal cord injury (revised 2011).
        J Spinal Cord Med. 2011; 34: 535-546
        • Daverat P.
        • Sibrac M.C.
        • Dartigues J.F.
        • et al.
        Early prognostic factors for walking in spinal cord injuries.
        Spinal Cord. 1988; 26: 255-261
        • Brown P.J.
        • Marino R.J.
        • Herbison G.J.
        • Ditunno Jr., J.F.
        The 72-hour examination as a predictor of recovery in motor complete quadriplegia.
        Arch Phys Med Rehabil. 1991; 72: 546-548
        • Consortium for Spinal Cord Medicine
        Outcomes following traumatic spinal cord injury: clinical practice guidelines for health-care professionals.
        Paralyzed Veterans of America, Washington (DC)1999
        • Kirshblum S.C.
        • O'Connor K.C.
        Levels of spinal cord injury and predictors of neurologic recovery.
        Phys Med Rehabil Clin N Am. 2000; 11: 1-27
        • Osheroff J.A.
        • Teich J.M.
        • Middleton B.
        • Steen E.B.
        • Wright A.
        • Detmer D.E.
        A roadmap for national action on clinical decision support.
        J Am Med Inform Assoc. 2007; 14: 141-145
        • Schopp L.H.
        • Hales J.W.
        • Brown G.D.
        • Quetsch J.L.
        A rationale and training agenda for rehabilitation informatics: roadmap for an emerging discipline.
        NeuroRehabilitation. 2003; 18: 159-170
        • Zorner B.
        • Blackenhom W.
        • Dietz V.
        • Curt A.
        • EM-SCI Study Group
        Clinical algorithm for improved prediction of ambulation and patient stratification after incomplete spinal cord injury.
        J Neurotrauma. 2010; 27: 241-252
        • van Middendorp J.J.
        • Hosman A.J.
        • Donders A.R.
        • et al.
        A clinical prediction rule for ambulation outcomes after traumatic spinal cord injury: a longitudinal cohort study.
        Lancet. 2011; 377: 1004-1010
        • Wilson J.R.
        • Grossman R.G.
        • Frankowski R.F.
        • et al.
        A clinical prediction model for long-term functional outcome after traumatic spinal cord injury based on acute clinical and imaging factors.
        J Neurotrauma. 2012; 13: 2263-2271
        • Pretz C.R.
        • Kozlowski A.J.
        • Charlifue S.
        • Chen Y.
        • Heinemann A.W.
        Using Rasch motor FIM individual growth curves to inform clinical decisions for persons with paraplegia.
        Spinal Cord. 2014; 52: 671-676
      1. Guide for the Uniform Data Set for Medical Rehabilitation (including the FIMTM instrument), version 5.1. State University of New York at Buffalo, Buffalo1997
      2. Rowland T, Ohno-Machado L, Ohrn A. Comparison of multiple prediction models for ambulation following spinal cord injury. In: Proceedings of the AMIA Symposium; American Medical Informatics Association; 1998 Nov 7-11; Orlando. Hanley & Belfus, Inc. p 528-32.

        • Gant V.
        • Rodway S.
        • Wyatt J.
        Artificial neural networks: practical considerations for clinical applications.
        in: Dybowski R. Gant V. Clinical applications of artificial neural networks. Cambridge Univ Pr, Cambridge2001: 329-356
        • Abdi H.
        • Valentin D.
        • Edelman B.
        Neural networks.
        Sage, Thousand Oaks1999
        • Caudill M.
        • Butler C.
        Naturally intelligent systems.
        Massachusetts Institute of Technology, Cambridge1990
        • Muller B.
        • Reinhardt J.
        • Strickland M.T.
        Neural networks: an introduction.
        2nd ed. Springer-Verlag, Berlin1995
        • Garson G.D.
        Neural networks: an introductory guide for social scientists.
        Sage, Thousand Oaks1998
        • Bradley A.P.
        The use of the area under the ROC curve in the evaluation of machine learning algorithms.
        Pattern Recognit. 1997; 30: 1145-1159
        • Swets J.A.
        Signal detection theory and ROC analysis in psychology and diagnostics: collected papers.
        Psychology Pr, New York2014
        • McGee S.
        Simplifying likelihood ratios.
        J Gen Intern Med. 2002; 17: 647-650
        • Florkowski C.M.
        Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests.
        Clin Biochem Rev. 2008; 29: S83-S87
        • Furlan J.C.
        • Fehlings M.G.
        The impact of age on mortality, impairment, and disability among adults with acute traumatic spinal cord injury.
        J Neurotrauma. 2009; 26: 1707-1717
        • Scivoletto G.
        • Morganti B.
        • Dittuno P.
        • Ditunno J.F.
        • Molinari M.
        Effects on age on spinal cord lesion patients' rehabilitation.
        Spinal Cord. 2003; 41: 457-464
        • Aito S.
        • D'Andrea M.
        • Werhagen L.
        • et al.
        Neurological and functional outcome in traumatic central cord syndrome.
        Spinal Cord. 2007; 45: 292-297
        • Cifu D.X.
        • Seel R.
        • Kreutzer J.
        • Martwitz J.
        • McKinley W.
        • Wisor D.
        A multicenter investigation of age-related differences in lengths of stay, hospitalization charges, and outcomes for a matched tetraplegia sample.
        Arch Phys Med Rehabil. 1999; 80: 733-740
        • Katoh S.
        • el Masry W.S.
        Motor recovery of patients presenting with motor paralysis and sensory sparing following cervical spinal cord injuries.
        Paraplegia. 1995; 33: 506-509
        • Oleson C.V.
        • Burns A.S.
        • Ditunno J.F.
        • Geisler F.H.
        • Coleman W.P.
        Prognostic value of pinprick preservation in motor complete, sensory incomplete spinal cord injury.
        Arch Phys Med Rehabil. 2005; 86: 988-992
      3. Samuel HW, Zaïane, OR, Sobsey D. Towards a definition of health informatics ethics. In: Proceedings of the 1st ACM International Health Informatics Symposium; 2010 Nov 11-12; Arlington. New York: Association for Computing Machinery; 2010. p 257-64.

        • Goodman K.W.
        • Adams S.
        • Berner E.S.
        • et al.
        AMIA's code of professional and ethical conduct.
        J Am Med Inform Assoc. 2013; 20: 141-143
        • Amarasingham R.
        • Audet A.
        • Bates D.
        • et al.
        Consensus statement on electronic health predictive analytics: a guiding framework to address challenges.
        EGEMS (Wash DC). 2016; 4: 1163
        • Cohen I.G.
        • Amarasingham R.
        • Shah A.
        • Xie B.
        • Lo B.
        The legal and ethical concerns that arise from using complex predictive analytics in health care.
        Health Aff (Millwood). 2014; 33: 1139-1147