Singapore publishes Artificial Intelligence in Healthcare Guidelines (AIHGIe)
The Ministry of Health (MOH) together with the Health Sciences Authority (HSA) and Integrated Health Information Systems (IHiS) has co-developed a guideline on Artificial Intelligence in Healthcare which was published in October 2021. The guideline aims to promote good practices and encourage safe development and implementation of artificial intelligence in the healthcare industry. The scope of this document affects:
- Artificial Intelligence Medical Devices (AI-MDs)
- AI application in healthcare settings e.g. training, wellness, administration
- Developers of AI-MDs who are defined as organisations or individuals who plan, fund, develop and/or maintain AI-MD, including standalone software medical devices that can interact with patients directly, or AI-MD intended to be used as part of healthcare service provision by organisations or individual healthcare professionals.
- Implementers of AI-MDs who are defined as organisations or individuals who use AI-MD to deliver healthcare services.
Artificial Intelligence Medical Devices (AI-MDs) refer to AI solutions that are intended to be used for investigation, detection, diagnosis, monitoring, treatment or management of any medical condition, disease, anatomy or physiological process. Therefore, AI-MDs are subjected to regulatory controls by the HSA and are to comply with the following guidelines:
- Regulatory Guidelines for Telehealth products (published April 2019)
- Regulatory Guidelines for Software Medical Devices – A Lifecycle Approach (published Dec 2019)
Apart from regulatory controls imposed by the HSA, key legislations affecting AI-MDs are also stipulated in the document. Recommendations for development and implementation for AI-MDs are outlined below:
Clinical and end-user inputs, representativeness of datasets used, mitigations to foreseeable cybersecurity threats or data breaches and clarity in explanation of the effectiveness of the device to the intended end-users are determined.
Appropriate development standards such as ISO 14971, ISO 13485 and IEC 62304, versioning controls for iterative builds and incorporating self-validation mechanisms could be adopted.
Validation of the built model with retrospective data, actual data and peer-review of results could be performed. Additionally, a decision on how the AI-MD could be incorporated into clinical workflow is established.
Clinical governance over safe use with organisational leadership’s approval. Performance of the device should also be tracked and monitored in this phase and end-user should understand that AI is in use.
Continued monitoring on device’s safety and performance is necessary as well as investigation and response to any device issues or related adverse events.
Both unscheduled and scheduled review on safety and efficacy of the device as well as periodic maintenance is encouraged.
The AIH guideline also includes a case study of a Class B AIMD which shows how the different recommendations are applied in development and implementation.
The AIHGle is meant to be a “living” document that will be constantly updated along with the developments of this field. So you can send in your feedback on the AIHGle, by using the feedback form: MOH’s Artificial Intelligence in Healthcare Guidelines (AIHGle) (form.gov.sg)
To learn more about registering medical devices in Singapore click here.
To learn more about the medical device market in Singapore click here.
To register your medical device in Singapore, please click the button below.