New diagnostic model can determine if patients will respond to treatment with 80 percent accuracy