Statistical diagnosis of lumbar spine disorders using computerized patient pain drawings

N. Horace Mann, Mark D. Brown, Isadore Enger

Research output: Contribution to journalArticle

18 Scopus citations

Abstract

Discriminant analysis is applied to 250 quantified low back patient pain drawings to study the ability of a computerized statistical method for classifying novel cases into one of five clinically-significant lumbar spine disorders. Tests on independent data were 46.2 percent (%) correct overall. Benign disorder (55.6%), herniated disc (51.7%), and psychogenic (56.3%) pain drawings were more accurately discriminated than the spinal stenosis (32.2%) and underlying disorder cases (35.2%). It is concluded that computerized patient pain drawings provide valid "initial impressions" of lumbar spine disorders. Further research is suggested to better distinguish between herniated disc and spinal stenosis pain descriptions, and for better recognition of serious underlying disorder pain drawings.

Original languageEnglish (US)
Pages (from-to)383-397
Number of pages15
JournalComputers in Biology and Medicine
Volume21
Issue number6
DOIs
StatePublished - 1991

    Fingerprint

Keywords

  • Benign disorder
  • Discriminant analysis
  • Herniated disc
  • Low back pain
  • Patient pain drawings
  • Psychogenic disorder
  • Spinal stenosis
  • Underlying disorder

ASJC Scopus subject areas

  • Computer Science Applications

Cite this