Background and Aim: AI-supported diagnostics promise earlier detection and more efficient care but expose healthcare institutions to legal, organisational, ethical and educational challenges. As the EU Artificial Intelligence Act intersects with the Medical Devices Regulation and the In Vitro Diagnostic Medical Devices Regulation, institutions need clear operational guidance. This study aimed to translate EU requirements into a practical implementation framework and identify competencies required for safe clinical use.

Materials and Methods: An interdisciplinary doctrinal analysis of the EU AI Act, MDR, IVDR, European guidance and medical-legal literature was conducted. Obligations applying to healthcare institutions as deployers were mapped across the AI lifecycle and converted into operational steps. A detailed guideline and concise checklist were developed within a student research project funded by the Research Council of Lithuania. A structured expert survey was developed to assess relevance, completeness, clarity and practical usability.

Results: The study produced a six-stage framework: (1) determining whether a system falls within the AI Act and high-risk categories; (2) verifying system and provider compliance; (3) establishing governance, responsibilities and data-protection arrangements; (4) conducting local performance evaluation, training staff and integrating the system into clinical workflows; (5) ensuring intended-purpose use, meaningful human oversight, data quality, logging and incident management; and (6) periodic monitoring, reassessment and safe discontinuation. The framework identifies AI literacy as a patient-safety requirement. Clinicians must understand system purpose and limitations, interpret outputs critically, recognise automation bias, maintain decision-making authority and report anomalies.

Conclusions: Legal compliance for medical AI cannot be achieved through technical validation alone. It requires lifecycle governance linking regulation, the fulfilment of AI literacy obligations, clinical oversight and patient safety. The framework offers a transferable basis for institutional training and implementation programmes.

Keywords: EU AI Act; medical diagnostics; AI literacy; human oversight; healthcare governance.