The recent decision T-0273/26, issued by a Board of Appeal of the European Patent Office (EPO), provides an important reference point for understanding how inventive step is currently assessed for artificial intelligence-based solutions in the healthcare sector. Beyond the specific outcome of the case, the decision offers valuable insight into the requirements that software and artificial intelligence (AI) innovations must meet to overcome patentability hurdles in Europe.

European patent application EP 3729460, entitled A medical intervention control system, concerned a system for controlling medical interventions capable of collecting clinical patient information, processing physiological sensor data through a detector, analysing medical waveforms and generating recommendations on potential therapeutic interventions. The parameters used included metrics widely employed in interventional cardiology, such as FFR, iFR, CFR and Pd/Pa.

However, despite the sophistication of the system and its use of substantial volumes of clinical data, the EPO found that the invention lacked an inventive step, and the Board of Appeal fully upheld the Examining Division’s decision.

What makes this decision particularly relevant is that it reflects an established trend in European practice: the mere application of advanced algorithms, predictive models or artificial intelligence techniques to medical information is not sufficient to establish an inventive step.

AI alone is not enough

Recent years have seen exponential growth in the use of artificial intelligence technologies in healthcare, including clinical decision-support systems, personalised medicine platforms, predictive diagnostic models and treatment-planning tools.

However, decision T 0273/26 serves as a reminder that, from the EPO’s perspective, the clinical or commercial value of a solution does not necessarily correspond to its relevance for patentability purposes.

The Board considered that many of the claimed features were directed towards the analysis of medical information and the formulation of clinical recommendations. Although such activities may be extremely valuable to healthcare professionals, they were primarily regarded as intellectual processes and, consequently, did not in themselves provide a technical contribution capable of supporting an inventive step.

The decision therefore confirms that the patentability assessment of such inventions must focus on the existence of an objective and demonstrable technical effect, rather than merely on the application of artificial intelligence techniques to a particular purpose.

The role of the technical effect

One of the most significant aspects of the decision is its emphasis on the need to establish a clear connection between AI-based features and a specific technical improvement.

The applicant argued that incorporating additional physiological parameters and waveform extraction techniques broadened the range and type of information used to train the analyser and, as a result, improved the system’s performance.

However, the Board observed that the claim did not specify the type of analysis performed by the analyser or how the additional data effectively contributed to a technical improvement in the system.

In essence, the EPO rejected the proposition that simply having more clinical information available necessarily implies a technical improvement. According to the decision, for such features to be taken into account when assessing inventive step, it is necessary to demonstrate specifically how they produce a verifiable technical effect.

The claims did not establish a functional link between the input data and the system’s output, as they failed to explain how the analyser or other modules used those data to achieve the claimed improvement.

This reasoning is particularly relevant to many current patent applications involving machine learning. Claims frequently describe models that process broader or more complex datasets than those previously used. As this decision demonstrates, however, this alone is insufficient to establish an inventive step. Applicants must also explain adequately what specific technical improvement is achieved as a result of that processing.

Implications for the healthtech sector

The practical significance of this decision is particularly clear for healthtech and medtech companies.

A significant proportion of healthcare innovation today is based on the combination of medical sensors, remote monitoring, advanced data analytics and artificial intelligence. In many cases, the differentiating value of these solutions lies precisely in their enhanced ability to transform large volumes of clinical data into useful recommendations for healthcare professionals or patients.

Decision T 0273/26 shows that, for patentability purposes in Europe, it will be essential to demonstrate that the technology goes beyond merely generating information or recommendations using AI.

Moreover, although the case arises in the healthcare context, its implications extend beyond the medical sector. The decision provides another example of the approach taken by the EPO towards computer-implemented inventions and, in particular, technologies based on artificial intelligence.

It reinforces the principle that algorithms, predictive models and machine learning techniques are not automatically considered technical simply because they use large volumes of data or generate complex predictions.

For these features to contribute to inventive step, applicants must demonstrate that they produce an identifiable technical improvement.

From a practical perspective, this means that patent applications involving AI should pay particular attention to describing aspects such as the system’s technical architecture, the specific processing mechanisms used, and the relationship between the processing steps and the resulting technical outcome.

Crucially, it is not sufficient simply to list these elements separately. The patent application should explain how they interact to produce the claimed technical effect.

A decision aligned with an evolving market

The significance of the decision should also be considered against the backdrop of rapid growth in the use of artificial intelligence across highly regulated sectors.

Companies are increasingly investing in AI-assisted diagnostic platforms, personalised medicine, intelligent medical devices and predictive analytics solutions capable of processing vast volumes of clinical information. In these environments, patent protection continues to play a strategic role in securing sustainable competitive advantage.

Decision T 0273/26 nevertheless illustrates the demanding standard applied by the EPO: it must be demonstrated that AI produces a specific technical effect, rather than simply providing a greater capacity to process data.

Decision T 0273/26 provides a clear practical lesson for AI-based inventions involving medical data: the complexity of an algorithm and the usefulness of the clinical recommendations it generates are no substitute for demonstrating a technical contribution when assessing inventive step.

For an AI-based invention to successfully meet the EPO’s patentability requirements, it is essential to establish a specific, objective and verifiable technical contribution. The key lies not only in the conclusions produced by the system, but in how those conclusions result in a genuine technical improvement within a system, device or technical process.

The decision therefore provides valuable guidance for innovators, healthtech companies and intellectual property professionals, offering a clear indication of the standard currently applied by the EPO to artificial intelligence technologies in healthcare and, more broadly, to inventions based on advanced software.