The model will leverage AI and machine learning for prior authorization of services vulnerable to fraud or inappropriate use, CMS says.