The registration and marketability of medical devices in Europe is governed by the Regulation (EU) 2017/745 and by guidelines published in terms thereof. This work focuses on rules for risk-based classification of medical devices as well as for the declaration of conformity of each class. We formalized the core rules of the EU regulation for medical devices in Positional-Slotted Object-Applicative (PSOA) RuleML. This resulted in a refinement and subgrouping of the original legal rules. The formalization led to an object-relational PSOA RuleML rulebase, which was supplemented by object-relational facts (data) about medical devices to form a knowledge base (KB). The KB represents knowledge by facts and rules integrating RDF/F-logic-like graphs/frames (`objects') with Prolog-like relationships. We tested this open-source KB, Medical Devices Rules, by querying it in the open-source PSOATransRun system, which provided a feedback loop for refinement and extension. The aim of this formalization is to create a computational guideline to assist regulators, Notified bodies (NBs), manufacturers, importers, distributors, and wholesalers of medical devices in the classification of medical devices. This can support the licensing process for these stakeholders and for the registration of medical devices with a CE conformity mark and a ''notified body number'' affixed. Because the Medical Devices Rules KB already is in a computational rule format, it can become a pluggable component of smart contracts about medical devices.