Summary: Using a patient's individual risk of progressing (I'm guessing through some combination of Risser/Cobb angle, but the abstract doesn't say) and then recording after a month how well the brace fit and how many hours a day it was worn allowed the research to predict within +/- 3.5 degrees what the final curve size would be.

***

http://www.ncbi.nlm.nih.gov/pubmed/1...,f1000m,isrctn
Preliminary validation of curve progression model for brace treatment.

Brace treatment is the most commonly used non-surgical treatment for adolescent idiopathic scoliosis (AIS). A brace compliance monitoring system consisting of a microcomputer and a force transducer was used to monitor how brace candidates used their braces during daily activates. A prediction model of the brace treatment outcome was developed based on 20 AIS subjects. Six subjects (1M, 5F) with AIS who had worn their braces for six weeks participated into this study. One month data was recorded during the study period. Knowing the risk progression at the beginning of brace treatment plus how brace subjects used their braces in terms of brace tightness and wear time during brace treatment yielded a predicted outcome which was compared to the final treatment outcomes with 2 years followed-up. This preliminary result demonstrated that the prediction model was able to predict the treatment outcome within +/-3.5 degrees.