Medical software embedding AI: which rules and classifications you need to know

What are we talking about?

Digital medical devices (DMDs) embedding artificial intelligence (AI) are multiplying in the healthcare field, and this is good news for the efficiency of our healthcare system and for patients.

Today, the first barrier between doctors, patients and these tools is regulatory.

What are the regulatory requirements in force for these SaMD (Software as Medical Device) embedding AI?

In this article, we offer you concrete examples of AI-based Medical Device Software (MDSW) and their classification under MDR 2017/745 or IVDR 2017/746, as well as the process to bring them into compliance.

Examples of AI-embedded medical software, regulatory status and classification

Among AI-based medical devices, the most common are:

• Diagnostic support algorithms, notably in radiology (46.9% of AI-based devices approved by the FDA¹), oncology (25%¹), neurology (15.6%¹), dermatology or genomics.

  ➡️ Classification (a priori): IVD MD under IVDR 2017/746, often risk class C or D.

• Patient prioritisation systems in hospital services or in outpatient medicine, using risk scores to recommend care, medication dosages or vaccination reminders.

   ➡️ Classification (a priori): MD under MDR 2017/745, class IIb.

• Insulin dosing applications

  ➡️ Classification (*a priori*): MD under MDR 2017/745, class IIb.

• Prescription support software (PSS) assisting the doctor in writing prescriptions, based on drug databases.

 ➡️ Regulatory status (a priori): this is indeed an MD according to the HAS under MDR 2017/745, class IIa or IIb.

• Gamified patient applications (DTx), offering a rehabilitation programme or personalised and evolving follow-up based on the patient's profile and progress.

  ➡️ Classification (a priori): MD under MDR 2017/745, class I or IIa.

• Self-diagnosis applications, notably for mental, dermatological or other illnesses.

  ➡️ Classification (a priori): MD under MDR 2017/745, class IIa.

Some AI software still remains in a borderline zone and is not always CE marked or approved by the FDA.

Some examples:

• Medical assistant chatbots, allowing healthcare professionals to consult official recommendations from medical databases.
• Patient applications guiding patients on whether or not to consult a doctor based on medical data and patient criteria.

How to obtain CE marking for a medical device embedding AI?

CE marking is mandatory for Medical Devices embedding AI as soon as a clinical benefit is claimed.

The assessment process is based on the following steps:

1️⃣ Implementation of a Quality Management System (QMS) compliant with ISO 13485 and MDR 2017/745 or IVDR 2017/746.

2️⃣ Building the Technical File (TF) demonstrating the safety and performance of the API/Module/Software, in compliance with European regulatory requirements.

3️⃣ Compliance with the AI Act, with progressive harmonisation between the requirements of the AI Act and those of MDR 2017/745-IVDR 2017/746

• Articles 47.3 and 48.5 of the AI Act: a single declaration of conformity and a combined CE marking are possible, allowing manufacturers to use a single QMS and Technical File covering both regulations.
• Similarities between the AI Act and MDR 2017/745-IVDR 2017/746, in particular on change management and post-market surveillance (PMS), but also key differences: management of training data, risk management, transparency obligation, human oversight.

4️⃣ Question your Notified Body (NB) to find out whether it plans to be designated under the AI Act, as none is yet designated to date.

Need regulatory support for your AI software?

At Sparta Care, we support you with:

• The regulatory strategy of your Medical Devices embedding AI (AIaMD).
• Implementation of a Quality Management System compliant with ISO 13485 and MDR 2017/745-IVDR 2017/746.
• Building a Technical File for your API/Module/Software.
• Your procedures with notified bodies and during your audits.

Contact us to learn more!



¹ Source: Benjamens, S., Dhunnoo, P. & Meskó, B. The state of artificial intelligence-based FDA-approved medical devices and algorithms. npj Digit. Med. 3, 118 (2020). https://doi.org/10.1038/s41746-020-00324-0

‍