Skip to main content

Active & Software Devices

Overview​

Active medical devices — including software — have their own classification rules in Schedule 2, Part 2 of the MD Regulations. From 1 November 2024, the grace period for software as a medical device (SaMD) ended and all software-based medical devices must comply with the updated classification rules and be included in the ARTG (or covered by a valid exemption). This was one of the most significant regulatory changes for the digital health sector in Australia in recent years.

What is an Active Medical Device?​

An active medical device relies on a source of energy other than energy directly generated by the human body or gravity. This includes devices powered by electricity, batteries, or other external energy sources. Software that meets the medical device definition is classified as an active medical device, even if it runs on a general-purpose device such as a smartphone or computer.

Active Device Classification Rules (Schedule 2, Part 2)​

The classification of active devices is determined by their intended purpose and the nature of the energy or information they deliver or process. Key rules include:

Active therapeutic devices (designed to administer energy to or exchange energy with the patient):

  • Those where exchange of energy is potentially hazardous: Class IIb
  • Those intended to control or monitor Class IIb therapeutic active devices: Class IIb
  • All other active therapeutic devices: Class IIa

Active devices for diagnosis:

  • Those providing diagnostic information where the disease may lead to death or severe deterioration without urgent treatment or poses a high public health risk: Class III
  • Those for serious conditions or moderate public health risk: Class IIb
  • All other active diagnostic devices: Class IIa
  • Devices intended only to illuminate the patient's body using visible light: Class I

Software classification rules (from 1 November 2024): Software is classified based on the potential severity of harm from an incorrect output:

  • Disease/condition may lead to death/severe deterioration or high public health risk: Class III
  • Disease/condition is serious or poses moderate public health risk: Class IIb
  • Disease/condition is not serious or no public health risk: Class IIa
  • Where software only provides a recommendation to a health professional (not replacing clinical judgement): one class lower than would otherwise apply

Software as a Medical Device (SaMD)​

Software is a medical device in Australia if it meets the definition in Section 41BD of the Therapeutic Goods Act 1989 — that is, if it is intended by the manufacturer to be used for a medical purpose. This includes mobile apps, clinical decision support software, AI-based diagnostic tools, and software integrated into medical hardware.

Software is excluded from regulation (not a medical device) if it meets specific TGA exclusion criteria — primarily that it does not take data directly from a medical device, is intended only to provide a recommendation to a health professional (not replacing clinical judgement), and is based on established guidelines that can be independently verified. The TGA has published separate guidance for each category of potentially excluded software.

👉 Software decision tool — TGA

From 1 November 2024:

  • All software that meets the medical device definition must be included in the ARTG or covered by an exemption
  • Software that was being supplied without ARTG inclusion during the grace period must now be included
  • New classification rules for software apply based on the risk of the intended output

Artificial Intelligence (AI) in Medical Devices​

AI and machine learning (ML) based software that meets the medical device definition is subject to the same regulatory framework as other SaMD. The TGA is aligned with IMDRF guidance on AI/ML-based software as a medical device. Key considerations for AI-based devices include:

  • Classification based on intended purpose and potential harm from incorrect output
  • Clinical evidence requirements proportionate to classification and novelty
  • Post-market surveillance obligations to monitor AI performance over time
  • Cybersecurity requirements under Essential Principle 12.1

👉 AI and software as medical devices — TGA

Official Sources​