An objective of the Veterinary Services Regulatory Board of South Australia (the Board) is to regulate the provision of veterinary services for the purposes of maintaining high standards of competence and conduct by veterinarians supporting animal health, safety and welfare and the public interest. This information sheets aims to assist veterinarians to adhere to their professional obligations when using artificial intelligence in clinical practice.
What is AI
AI is the ability of a computer system to perform tasks that would normally require human intelligence, such as learning, visual perception, language translation, reasoning, and making decisions.
Potential benefits
The application of artificial intelligence (AI) in clinical practice can have the potential to achieve quicker diagnosis and interventions, streamline communication and reduce workload. AI can also be helpful for creating client handouts, reviewing the latest literature on conditions, interpreting results, assisting diagnostics and monitoring, disease prevention and prediction, non-clinical administrative use and recording clinical notes.
The Board has a responsibility to anticipate and understand the future of the delivery of veterinary services and ensure that if veterinarians are experimenting with emerging technologies (such as AI) in clinical practice, that veterinary services are still provided in a way that is accountable and in accordance with a veterinarian’s professional obligations.
AI may be able to provide some commercial efficiencies for veterinary clinics. However, if a veterinarian is utilising AI in clinical practice, consideration should be given to professional obligations, requirements under other legislation and data management and security. Key principles for veterinarians to consider to ensure that they are meeting their professional obligations may include any of the following.
Informed consent
Informed consent from clients ought to be sought before using AI in clinical practice to ensure that they understand the benefits, risks and limitations of these technologies.
Veterinary practices should be mindful of the requirements under the Surveillance and Devices Act 2016 (SA) (SD Act) in relation to the use of AI-enabled audio or visual recordings. The purpose of the SD Act is to protect people’s privacy by regulating how surveillance devices are used, and how information obtained from them is handled when attempting to distribute visual or audio recordings.
Before using any AI tool that listens to or records a consultation, you should ask the client for their permission, and they should be able to say “no” without any impact on their animal’s care. When obtaining a client’s permission, they should also be alerted to the fact that any transcribed notes may include conversations between anyone in the room which may then be shared with other veterinarians taking over care of their animal, or the Board in the event of a complaint.
It would be prudent to have a multifactor approach to consent including prominent practice wide signage, individual signed consent forms and a policy on verbally addressing the use of AI as part of a consultation. The same standards of review, accountability and disclosure apply, regardless of the physical setting in which the AI tool is used (e.g. telemedicine vs an in clinic consultation).
Accountability
Veterinarians and support staff should educate themselves about the limitations of the specific tool they are using to be able to use it safely and in accordance with their professional obligations. At a minimum, the product information of a particular tool should be reviewed, including how it’s trained and tested on populations, its intended use, the true meaning of words within that software, limitations and clinical contexts where it should not be used.
In relation to the use of AI for clinical diagnostics, it is a veterinarian’s responsibility to ensure that AI is only being used in a supportive capacity and it must not displace clinical judgment.
Record keeping
Under the Act, complaints can be received by the Board about veterinarians for up to 7 years after an event. Veterinarians should be mindful of how they are going to store records created by AI for the purposes of providing these records to the Board in the event of a complaint.
If a veterinarian is using AI for clinical note taking, it is their responsibility to check the accuracy of each clinical record to ensure that it reflects the veterinarians own clinical thinking. The Board will not accept inaccurate notes at the fault of AI as justification for clinical records that do not reflect the actual veterinary services that have been provided.
Each clinical note that is drafted, transcribed or materially assisted by an AI tool should include a notation indicating that AI assistance was used and a verification that the clinical record has been checked by the veterinarian for accuracy.
Transparency
It is important for veterinarians to be open with clients about how and why AI is being used in clinical practice. For example, information advising that the clinic is using AI may want to include the extent of its use and whether AI is being used for administrative efficiency, clinical note taking, diagnostics or a combination of uses.
Care should be taken to avoid conflicts of interest in circumstances where AI may prioritise commercial outcomes over clinical decisions. Veterinarians should be mindful to always retain their own clinical judgment and to not rely on AI, which may not be aware of or detect subtle or atypical clinical signs or species-specific nuances, or simply may not consider the full patient history.
Confidentiality
It is a veterinarian’s responsibility when using AI to ensure that confidentiality and privacy is being upheld by checking that data is collected, stored, used and disclosed appropriately and that privacy is not inadvertently breached. Veterinarians need to be aware of whether the client data being used or recorded is also used to train the AI model for future use, and whether identifiable client data will then find its way into that learning database.
Examples of where confidentiality may be at risk when using AI include uploading clinical records or x-rays of clients, commercially sensitive data and client’s personal information. The Board recommends veterinarians treat any data input into AI as potentially publicly accessible.
Insurance
You may need to consult your professional indemnity insurer if you are using AI in clinical practice to ensure that the use of this technology does not impact on you holding appropriate insurance pursuant to section 52 of the Veterinary Services Act 2023.
Use of AI in complaints
The Board has received an increasing number of complaints from members of the public which have been compiled using AI. The Board has also received queries from veterinarians relating to the provision of AI generated transcripts and recordings of consultations in response to complaints.
The Board will not receive audio or visual recordings created as a byproduct of the use of AI in clinical practice. If you wish to submit your clinical records produced with the assistance of an AI transcription service in relation to a consult the subject of a complaint, those records should disclose that they were generated by AI and be appropriately signed by the treating veterinarian attesting to their accuracy.
The Board continues to assess complaints objectively and in accordance with the Act and the complaints administrative processes and procedures and does not rely on any inference or suggested conclusion that may be offered by an AI complaint.
Take home notes
The Board recommends that members of the profession undertake a thorough risk assessment about the use of AI in clinical and administrative practice as it relates to your individual circumstances.
If you are using AI, it is vital that the decisions you make, and the supporting documentation that provides evidence of your clinical judgment as a veterinarian, do not detract from the standard of veterinary services that you provide, which remain your responsibility, whether or not AI was used.
Circumstances where the Board strongly recommends that AI is not used include:
- As the only source of diagnosis
- Triage without human oversight
- Generating clinical records without human verification (the use of AI does not displace record keeping guidelines)
- Uploading confidential data into unsecure tools
Every clinical practice should consider having a formal policy in place that is regularly reviewed to establish the internal policies and procedures around the use of AI.
This document is the first version and is intended as a reference tool to assist veterinarians to make informed decisions about the use of AI in providing veterinary services and does not constitute formal policy or guideline under the Veterinary Services Act 2023. This document will be reviewed every 6 months to ensure currency.