AI and Clinical Investigations of Medical Devices - Where Do We Stand at the Beginning of 2026?

Clinical investigations of medical devices will be increasingly shaped by AI by mid-2026 – both as standalone AI medical devices to be tested and as integral background assistance for tasks in clinical investigations and studies.

Automation is used wherever possible and sensible, but not all of it is "Artificial Intelligence" and not all of it is worthwhile at the current time. Below is a nuanced look at the current situation:

Role of AI in Clinical Investigations

Artificial intelligence plays two fundamentally different roles in clinical investigations. Firstly, AI itself is the subject of clinical evaluation for individualized or personalized medicine, for example, when it is an AI-based diagnostic or therapy support system that is to be placed on the market as a medical device or for which therapeutic claims are to be substantiated after market placement. Secondly, AI methods can be used as tools in almost all phases of the study process, wherever people have to spend a long time searching, comparing, checking, or summarizing.

AI as the Subject of Investigation (AI Medical Device)

For AI as a medical device, compliance with MDR/IVDR (including Software Rule 11 and MDCG 2019-11), the EU AI Act as high-risk AI, and the relevant standards for software lifecycle, quality management, risk management, and IT/cybersecurity (e.g., EN ISO 13485, EN ISO 14971, IEC 62304, IEC 82304-1, ISO/IEC 42001, ISO/IEC 27001) must be observed from the very beginning of development. Specific challenges arise from learning algorithms, version changes, and the strong dependence on training and validation data, which necessitate thorough planning of endpoints, population, and study design. At the same time, questions regarding generalizability, bias, and transferability to different healthcare realities are becoming more prominent. High-ranking scientific publications already exist that demonstrate the added benefit in various indication areas. This type of medicine is rapidly and widely entering clinical practice and fundamentally changing our understanding of treatment.

AI as a Tool in the Study Process

Regardless of the actual investigational product, AI can make clinical studies more efficient and targeted. Application areas range from analysing historical data to optimise study designs, intelligent selection of study sites, and AI-supported identification of suitable study participants, to monitoring data quality and protocol adherence. Additionally, generative AI approaches can pre-structure content such as reports, lay summaries, or publication drafts, thereby significantly reducing the workload for the project team – always provided that careful (currently still human) review is conducted – from literature research, writing a precise study synopsis, coordinating all documents of a complex investigation plan (especially during classic iterative amendment rounds), planning, recruitment, data management, evaluation, and reporting.

‍

The guidelines from the FDA and EMA make it very clear: AI in the development of therapeutics requires governance, documentation, performance evaluation, and monitoring throughout the entire lifecycle – not just of the models' performance.

‍

By mid-2026, AI in clinical investigations of medical devices will be a lived reality with further growth potential. At the same time, many questions remain open – for example, regarding the interpretation of regulatory requirements, the handling of continuously learning systems in clinical study evaluation, and the practical integration of AI solutions into existing quality and study processes. Therefore, it is becoming increasingly important for manufacturers, study sites, and service providers to bring together technical, regulatory, and ethical perspectives early on.

‍

Are you interested in learning whether and how MEDIACC, as a CRO, uses AI systems in digital clinical trials for medical devices? Stay tuned or feel free to contact us for an initial consultation!