Artificial intelligence (AI) has transcended its status as a futuristic concept and is now actively transforming healthcare practices, clinical decisions, and patient interactions on a global scale. The recent conference titled "AI in Healthcare: Innovation, Risk and Responsibility" showcased the immense potential of AI to enhance patient outcomes, streamline operations, and assist medical professionals in navigating the complexities of modern healthcare. The discourse at the conference also focused on the inherent risks, regulatory challenges, and legal considerations associated with AI from an international standpoint.
Here are the main insights shared during the conference, highlighting focal points for public health entities, private healthcare practitioners, and insurance providers.
### The Effect of AI on Care Providers and Patients
Recent studies indicate that AI systems operating with “humans in the loop” are yielding more favorable patient results. Nonetheless, it is crucial for clinicians to articulate their reasoning if they disagree with an AI-generated recommendation, and established procedures such as peer reviews should be in place to oversee clinical judgments in these cases. Patients must be informed whenever differing opinions arise that may influence their treatment options.
Healthcare providers are encouraged to maintain their personal connections with patients and to apply their clinical judgment alongside AI tools. Clinicians should be educated about AI outputs to effectively guide patients regarding decisions informed by AI. Understanding the workings of the AI “black box” is ideal.
It is recommended that clinicians utilize only those AI tools they are trained on and that have received the appropriate approvals from their healthcare institutions.
Implementing AI solutions in healthcare demands a strategic approach. To yield tangible benefits and improve patient care, healthcare organizations must first assess the specific needs or issues before deploying AI technology.
Providers should continuously assess the clinical impact of AI technologies to verify their effectiveness in real time prior to widespread implementation, thereby mitigating risk to patients.
Integrating AI to automate mundane tasks can free up healthcare professionals for more targeted patient engagement. Additional training is essential to capture the learning potential from these routine activities, such as compiling discharge summaries.
The employment of AI scribes in healthcare settings should be restricted to those that have been vetted and authorized for use. Caution is advised post-training, as clinicians remain accountable for the documentation produced and must ensure its accuracy. Patients should also be informed about the use of AI tools in their medical care and be required to give written consent.
### Safety and Risk Considerations for Healthcare Providers
The establishment of the National Commission into the Regulation of AI in Healthcare by the Medicines and Healthcare products Regulatory Authority represents a significant step toward developing a comprehensive regulatory framework for health-related AI. However, as the pace of technological advancement accelerates, aligning regulations with these changes will require time.
Regulatory measures exist for medical devices, and to minimize the potential for harm and liability, healthcare providers must ensure that the devices in their facilities possess the highest levels of classification approval. Conducting thorough due diligence on suppliers and using the devices strictly for their intended purposes is essential.
Healthcare staff must be appropriately trained in using AI tools, including medical devices, before their integration into practice.
Healthcare organizations should conduct risk mapping regarding the AI and medical device tools employed within their facilities. This process, ideally initiated from the board level, is crucial for developing robust policies, ensuring protective contracts with suppliers, and maintaining adequate indemnity coverage.
Implementing solid governance and risk mapping across the AI tool chain in each specialty setting will facilitate quick identification of incidents, trace their origins, and enable prompt action, thus reducing potential harm and accelerating liability assessment.
### Data Usage and Security Issues
To fully harness the capabilities of AI in healthcare, facilitating access to sensitive patient data for machine learning is essential. However, establishing access rights for data usage poses significant challenges for public health organizations and other providers. Consent from patients must be obtained for each specific instance of data usage, and clinicians must be aware of how AI utilizes this data to accurately inform patients.
Healthcare providers must also navigate hurdles related to data storage and cross-border data use. In the UK, strict regulations govern the processing and usage of health data. Thus, healthcare institutions should ensure that their operational AI tools comply with data protection standards, maintaining patient data securely, preferably on-site or in private cloud environments, as public clouds may be susceptible to security breaches.
Should patient data be shared with a third party, it is vital for healthcare providers to establish contractual restrictions governing data disclosure and usage, alongside implementing practical safeguards such as enhancing interface security, limiting the queries AI tools can address, and fortifying protection against hacking and unauthorized access.
In conclusion, while AI offers tremendous potential in revolutionizing healthcare delivery, it is accompanied by its own unique set of challenges.
For more insights and guidance on managing emerging legal risks associated with AI, please reach out to the author.
