Computer Diagnosis and Discriminate Analysis Decision Schemes
Arch W. Templeton, Kim Bryan, Rex Adney Waid, James R. Townes, Md Sirajul Huque, Samuel J. Dwyer · Radiology · 1970
Linear decision, clustering, and polynomial discriminate analysis decision schemes were used to distinguish normal from abnormal hearts in 907 cases. Heart disease was correctly diagnosed in 74 per cent. More significantly, when all cases of heart disease were grouped together and compared with normal hearts, the decision schemes diagnosed abnormalities correctly in 98 per cent. If adequate data can be collected, a computerized scheme can identify important measurements with diagnostic results similar to those achieved by a radiologist.