Artificial intelligence (AI) is technology that uses computers and machines to perform tasks that normally use human intelligence, such as learning, reasoning, and problem-solving. It's important to remember that AI tools are just that—tools. They are not meant to replace your clinical decision-making, judgment, or responsibility. Ensure your practice is evidenced-informed and aligns with BCCNM standards.
Some areas where AI is being used in health care include:
- Diagnosing and treating conditions such as tumors, fractures, or other abnormalities.
- Documentation software programs to record and summarize clinician-client encounters.
- Clinical decision support systems built into electronic health records.
- Client monitoring systems, such as alerts and alarms.
Considerations
The use of AI is rapidly evolving. While BCCNM does not have a specific standard related to AI, many BCCNM standards are relevant when considering using AI in your practice.
Here are some things to consider before using AI to maximize benefits and minimize risks, and ensure safe, ethical, and competent care.
Data privacy and protection
AI uses a combination of algorithms, data, and computing power to perform tasks. Sources of data include electronic health records (EHRs), medical imaging (such as X-rays, MRIs), clinical trials, wearable devices, and public health datasets. If you use AI tools in your practice, it is your responsibility to make sure this sensitive data is protected. Clients are not always given control over the use of their information or not fully informed about privacy impacts such as hacking or selling of personal data.
Before using AI, ask yourself:
Do the AI tools I plan to use comply with applicable provincial and federal regulations to protect client data? Most should, but it is good to check.
Do the tools comply with my employers' policies about data protection, encryption, access, and secure storage (for example, the Confidentiality, Collection, Use, and Disclosure of Personal Information and Relevant Consent Policy; Access to Client Records, Storage, Retention, Transportation, and Disposal of Client Records Policy)? What is my plan if my employer doesn't have a policy (for example, contact the privacy officer)?
Do I understand where and how the AI tool uses and stores client information and data? For example, is the data stored within the organization or is it stored externally, or even outside the country?
Can I maintain client privacy and confidentiality when using AI tools?
Can I store, access, and transmit personal client data securely?
Review
Fairness and biases
AI algorithms are designed by humans. As a result, there is a risk that racist, biased, or harmful ideas could influence AI's decision-making, reflecting human and societal biases. For example, if the data used to train an AI tool only comes from one group of people, the AI may not work well for others, which could lead to unfair treatment or outcomes. It is important to be aware of and report these issues when using AI tools.
Before using AI, ask yourself:
Am I aware that there is potential bias with AI tools? How do I report suspected biases? How do I make my clients aware of this?
How can I consider Indigenous-specific anti-racism, cultural safety, and cultural humility, and an equity, diversity, and inclusion (EDI) values when applying AI recommendations?
Am I aware of my own biases regarding AI in practice? For example, if it is available and I don't use it, why is that?
How can I build and maintain client trust by ensuring AI use enhances, rather than replaces, human interaction and empathy in care?
Review
Informed consent
Clients have the right to make informed decisions about their care, including understanding AI options. Clients may not know who will use their data or how it will be used. You must inform them about how AI impacts their care and treatment options.
Before using AI, ask yourself:
Can I explain to the client the benefits, limitations, and risks of AI in their care in a way they understand?
Have I explained to the client how their personal data will be used and stored, including its use in the training of AI tools?
If appropriate, have I given the client the choice to consent or opt out if they do not want AI used in their care?
Do I have client consent to transfer their information from one clinical environment to another?
Review
Safety and transparency
AI should support, not replace, human decision-making. As a health-care professional, you are ultimately responsible for client care decisions. Remember, there is a limit to what AI should do without some form of human oversight or input.
Before using AI, ask yourself:
Have I critically evaluated AI recommendations? For example, did I analyze the clinical appropriateness of a result and use my clinical judgement regarding its findings?
Am I aware that I am responsible and accountable for decisions made with the help of AI?
Do I know if the AI tools I am using align with ethical standards, so clients are not harmed or exploited?
Why am I using an AI tool? Is it being used to improve client care and health outcomes or for another reason such as AI owner incentives?
How will I ensure the validity of the AI tool, its appropriate application, and ongoing evaluation for reliability?
Will I know if I am over-relying on AI tools in my practice?
How will I educate myself about AI, i.e., how it works, its limitations, and how to interpret its outputs?
Review