Device-assisted therapies (DAT) benefit people with Parkinsons Disease (PwP) but many referrals for DAT are unsuitable or too late, and a screening tool to aid in identifying candidates would be helpful. This study aimed to produce such a screening tool by building a classifier that models specialist identification of suitable DAT candidates. To our knowledge, this is the first objective decision tool for managing DAT referral. Subjects were randomly assigned to either a construction set (n = 112, to train, develop, cross validate, and then evaluate the classifier's performance) or to a test set (n = 60 to test the fully specified classifier), resulting in a sensitivity and specificity of 89% and 86.6%, respectively. The classifier's performance was then assessed in PwP who underwent deep brain stimulation (n = 31), were managed in a non-specialist clinic (n = 81) or in PwP in the first five years from diagnosis (n = 22). The classifier identified 87%, 92%, and 100% of the candidates referred for DAT in each of the above clinical settings, respectively. Furthermore, the classifier score changed appropriately when therapeutic intervention resolved troublesome fluctuations or dyskinesia that would otherwise have required DAT. This study suggests that information from objective measurement could improve timely referral for DAT.