5. General considerations
5. General considerations For BtX, as for all devices, clinical evidence 8 must demonstrate compliance with the requirements of the MDR or IVDR, including that the benefit-risk ratio is acceptable. To that end, the manufacturer is required to conduct a clinical evaluation (for MDs) or a performance evaluation (for IVDs). 9 As described in section 4.2.2, a device with a high degree of novelty may be associated with uncertainty regarding the device’s expected safety or performance, due to limited prior scientific knowledge or data from similar devices (if any). These devices may be associated with new risks or unexpected side effects; the necessary clinical evidence has to appropriately address this uncertainty. For BtX, a balanced approach is needed to ensure that the level of uncertainty associated with the device doesn’t unnecessarily prevent timely patient access to devices that provide significant positive clinical impact compared with the state of the art and alternatives (if any). When comparing the life-cycle clinical evidence requirements of BtX with other devices, the overall clinical evidence necessary does not differ. However, for BtX, given the potential for significant positive clinical impact, consideration should be given to whether the necessary longer term data 10 can be collected in a way that would allow for more timely availability of these devices to patients without compromising the clinical evidence needed to confirm the acceptability of the benefit risk ratio. In particular in cases where the BtX is fulfilling unmet medical needs it may be acceptable for a BtX to be placed on the market with a higher level of uncertainty and more limited pre-market clinical evidence provided that a well-defined plan is in place to collect confirmatory data through comprehensive PMS and PMCF/PMPF. In summary, provided that the available pre-market data give an appropriate level of assurance of safety and clinical performance of the device and support the manufacturer’s claim of significant positive clinical impact compared with the state of the art and available alternatives, it may be appropriate for certain confirmatory data to be collected through well- defined, scientifically valid, and milestone-based PMS, PMCF/PMPF plans (e.g. studies, investigations, other activities). With respect to clinical evidence requirements, it may, in general, be acceptable to place a BtX on the market provided that all of the following can be justified for the device in question: 8 For BtMD, clinical evidence must be provided by clinical data, whereas for BtIVD, clinical evidence must be provided by scientific validity, analytical performance, and clinical performance data, per MDR Article 61(1) and IVDR Article 56(1), respectively. 9 In accordance with MDR Chapter VI and Annex XIV and IVDR Chapter VI and Annex XIII. 10 Clinical data for MDs; scientific validity, analytic performance, and clinical performance data for IVDs. Medical Device Medical Device Coordination Group Document MDCG 2025-9 14 - the available data 11 sufficiently demonstrates: o that the benefit-risk ratio is acceptable; and o that it is reasonably expected the device will provide a clinical benefit 12 , taking into account the clinical condition, the state of the art, and the safety of patients, - the manufacturer has an appropriate PMS, PMCF/PMPF plan (see section 9), with a clear justification as to why the clinical data to be collected through PMCF/PMPF are not being collected in the pre-market phase, - users and patients will be adequately informed (e.g. by provision of information in the labelling, IFU, SSCP/SSP (if applicable), and/or other accompanying documentation) of the breakthrough status of the device and instructions to users on how to report incidents, complaints, and other clinical experience to the manufacturer, - appropriate measures are in place to meet any conditions or provisions of certification that have been imposed by a notified body. Risk management It is good practice in the planning phase of the device development to document in the risk management file how evidence will be provided to demonstrate that risks, including in particular increased risks or new risks arising from the innovative nature of the design or technology, have been identified and effectively mitigated. Risk management is a lifecycle process, which must be linked also in an early stage of the development with the clinical evaluation process to ensure that the residual risks have been weighed against the benefits offered by the BtX device. This ensures a comprehensive evaluation of safety and performance, while supporting the availability of the innovation's benefit for the patients, as well as during the clinical investigations / performance studies in the treatment group, and during the post-market phase. In the early stages of developing BtX devices, safety and performance data from patient use are in principle unavailable. This means that the occurrence rates of side effects, incidents, and harms from residual risks may not be fully quantifiable or quantified under certain assumptions. As such, the manufacturer’s criteria for accepting risks, when the probability of harms cannot be fully estimated or estimated accurately, become particularly significant. Input from the state of the art assessment will need to provide the risks levels of the currently accepted clinical practice and specific focus on the new design features and/or resulting clinical procedure needs ensure that acceptable risk levels are determined from a safety perspective. Appropriate technical and clinical inputs should be considered to verify and validate (as far as possible), the assumptions made in relation to the estimated occurrence levels, the predicted harm(s) associated with the hazards, the severity of those harms and the acceptance criteria, all in comparison with current state of the art. An estimation of the occurrence rate is necessary before the clinical investigation(s) / clinical performance study, in order to establish the acceptance criterion which will enable identification of unacceptable risks while the studies are conducted according to the plan and clinical data are gradually collected. These data, including PMCF/PMPF, enable the continuous verification of the initial estimation, and a data- 11 Including non-clinical/pre-clinical data, clinical data for MDs, and scientific validity, analytical performance, and clinical performance data for IVDs. 12 As defined in MDR Article 2(53) for MDs, and IVDR Article 2(37) for IVDs. Medical Device Medical Device Coordination Group Document MDCG 2025-9 15 based quantification of all known clinical risks and is expected to be incorporated into the risk management process. Considerations relevant for IVDs The principles outlined above remain relevant for IVDs, however consideration of IVD specific aspects is warranted. When it comes to IVD BtX, the performance evaluation should include the evaluation of the analytical performance, scientific validity, and clinical performance. In line with the guidance in the rest of this document, manufacturers should utilise avenues at their disposal to generate data where required and possible. Similar to BtXMDs, there may be insufficient existing evidence to draw on for certain BtIVDs. In such circumstances it may be challenging to demonstrate scientific validity based on available information. There may be a need to conduct proof of concept studies or other activities to generate evidence to support the scientific validity. Consideration may be given to additional methodologies, e.g. in-silico, to support the evidence body for scientific validity. Given the potential for limitations with respect to both scientific validity and clinical performance, the robustness of the analytical performance data is key for BtIVDs. The use of novel markers or markers without certified reference materials can lead to challenges in assessing BtIVDs. Where there are no comparative methods, comparisons to other well- documented methods or the composite reference standard can be used where appropriate and justified (IVDR Annex XIII, 1.2.2). In the absence of comparator assays, consider other means to verify performance e.g. robust orthogonal testing, creation of internal reference material etc. Considerations relevant for MDAI In the case of BtX that are or contain one or more high-risk AI systems (MDAI), additional considerations are required due to the simultaneous and complementary application of both the MDR/IVDR and the AI Act. 13 These considerations apply across the entire device life cycle, including design, pre-clinical, clinical, and performance evaluation, post-market surveillance and change management activities. Particular attention should be given to the quality and representativeness of data sets used for training, validation and testing of MDAI. Manufacturers should implement robust data governance and management practices, ensuring that data are relevant, accurate, complete and representative to the best extent possible, and that appropriate measures are in place to identify and mitigate bias in the data sets used for training, validation and testing, in accordance with the requirements of the AI Act. These practices form a core part of both the risk management system and the clinical evaluation or performance evaluation process, given the significant influence that data integrity has on device safety and performance. In this context, manufacturers are expected to ensure that the clinical evidence adequately demonstrates the performance, reliability, and generalisability of the MDAI within its intended purpose and intended use environment. The evaluation should take into account the dynamic characteristics of MDAI, including the impact of training, updates, and data drift, as well as the controls established through Pre-Determined Change Control Plans (PCCPs) to manage any modifications in a predictable and transparent manner. 13 MDCG 2025-6 - FAQ on Interplay between the Medical Devices Regulation (MDR) & In vitro Diagnostic Medical Devices Regulation (IVDR) and the Artificial Intelligence Act (AIA). Medical Device Medical Device Coordination Group Document MDCG 2025-9 16 When applying the principles outlined in the IMDRF guidance “Good Machine Learning Practice for Medical Device Development: Guiding Principles”, manufacturers should ensure that the AI system’s design and validation processes are consistent with the MDR/IVDR requirements on risk management, clinical evidence, and post-market follow-up. This integrated approach supports a continuous demonstration of conformity, ensuring that the safety and performance of the MDAI remain consistent throughout its lifecycle while preserving the benefits of innovation for patients and healthcare systems.