An app that stores medical reports, an algorithm that flags possible clinical abnormalities, and a platform for booking appointments all operate in healthcare. Yet they are not necessarily subject to the same rules. The distinction depends primarily on what the software is intended to do and how its manufacturer presents it.
Determining whether a product qualifies as medical device software is a crucial step for medtech startups, healthcare organizations, and research centers. Incorrect classification can affect product design, required testing, and the route to placing the solution on the European market.
Let’s look at how to distinguish administrative applications from software with a medical purpose, what the Medical Device Regulation (MDR) requires, and which questions to ask before developing a new e-health solution.
Table of contents
What is medical device software?
Regulation (EU) 2017/745, known as the Medical Device Regulation (MDR), includes software in its definition of a medical device when the manufacturer intends it for specific medical purposes, such as diagnosing, preventing, monitoring, predicting, prognosing, or treating disease. Therefore, a product does not qualify simply because it is used in a hospital or contains health data: its stated and documented intended purpose is what matters.
The key legal reference is Regulation (EU) 2017/745, available on EUR-Lex. It establishes safety and performance requirements, conformity assessment procedures, and obligations applicable to medical devices.
Consider two applications: one stores appointments and patient contact details; the other processes clinical parameters to support a physician’s diagnostic decision. The first may have a purely administrative function, while the second may fall under medical device legislation. In either case, qualification requires a careful assessment of actual functionality and the manufacturer’s claims.
Healthcare apps and medical devices: what is the difference?
Health-related content, a patient-facing interface, or the use of artificial intelligence does not automatically make an app a medical device. Actual functions and stated purpose must be considered together. It is particularly important to distinguish among:
- Administrative software. Manages bookings, staff schedules, or billing without having an inherently medical purpose.
- Document management systems. Store or transfer records without interpreting them for diagnosis or treatment; their specific features must still be assessed.
- Wellness apps. Encourage general habits, such as physical activity or meditation, without claiming a specific medical purpose.
- Clinical decision support software. Analyzes patient data and provides information intended to inform diagnostic or therapeutic decisions.
Italy’s Ministry of Health highlights the importance of assigning products the correct legal qualification, especially in borderline cases where the boundaries between regulatory categories are not immediately clear.
For further details, see the Ministry of Health’s guidance on borderline products.
How does MDR software classification work?
Once software has been identified as a medical device, its risk class must be determined using the rules in Annex VIII of the MDR. Rule 11 specifically addresses software that provides information used to make diagnostic or therapeutic decisions, or to monitor physiological processes.
Classification depends on the intended purpose and the potential consequences of decisions based on the software’s output. A risk class cannot be assigned automatically just because the product uses AI, is a mobile app, or runs in the cloud. In certain cases, other rules may apply, including Regulation (EU) 2017/746 on in vitro diagnostic medical devices.
Where required, CE marking is the outcome of the relevant conformity assessment process, not simply a symbol to add to an interface or website. Whether a notified body must be involved also depends on the device classification and applicable rules.
Seven checks before developing medical software
Involving clinical, technical, and regulatory expertise early on reduces the risk of having to redesign a product after initial testing. A useful checklist consists of seven questions:
- What clinical problem does it address? Define users, patients, indications, and expected outcomes without vague claims.
- What is its intended purpose? Clearly state what the manufacturer intends the software to do.
- What data does it process? Distinguish between simple storage, display, processing, and clinical interpretation.
- Which decisions could it influence? Assess how incorrect, incomplete, or delayed information could affect patients.
- How will safety and performance be demonstrated? Plan risk management, verification, validation, and clinical evaluation, where applicable.
- How will data and systems be protected? Build cybersecurity, access controls, and personal data protection into the design from the outset.
- How will it be managed after release? Plan updates, change management, post-market surveillance, and responsibilities.
These checks do not replace a formal regulatory assessment, but they help turn a technological idea into a clearer project for developers, clinicians, and potential partners.
Artificial intelligence and medical software: what deserves attention?
Machine learning algorithms can support image analysis, pattern recognition, and the assessment of clinical parameters. However, model quality does not automatically translate into clinical safety. Data representativeness, performance across patient groups, errors, human oversight, and changes to the software over time all need to be considered.
The user interface matters too: even an accurate result can be misunderstood if it is displayed ambiguously. Documentation, usability, change traceability, and user training should therefore be designed alongside the code.
For an application area where clinical validation deserves particular attention, read our article on digital therapeutics (DTx).
Frequently asked questions about the MDR
Do all medical apps need CE marking?
No. Whether the MDR applies depends on whether the software qualifies as a medical device and on its intended purpose. Products that fall within the regulation must undergo the relevant conformity assessment process before being placed on the market.
Can cloud-based software be a medical device?
Yes. Where software runs is not decisive by itself. Cloud-based solutions and mobile apps can also qualify if they have a medical purpose under applicable legislation.
Is GDPR compliance enough to meet MDR requirements?
No. Personal data protection and medical device compliance are separate regulatory matters, and both must be addressed when applicable.
IPPOCRATE AS: designing healthcare technology around real needs
An effective e-health solution starts with a clearly defined healthcare need and a structured software development process. IPPOCRATE AS, part of JO Group, develops e-health and m-health solutions for companies and research centers, covering analysis, design, development, testing, and validation.
Identifying a potential medical intended purpose early in the healthcare innovation process helps teams work with clinicians, technical partners, and regulatory advisers without confusing prototyping and research with market placement.
Do you have a healthcare software idea or a research project you want to turn into a digital solution? The first step is to clarify goals, users, functions, and requirements before choosing technologies and architectures.
