In the June 12 webinar, several co-authors of the National Academy of Medicine (NAM) new special publication explained the AI code framework and guided the responsible, effective, and human-centered use of AI in medicine.
The AI Code Behavioral Framework is intended as a touchstone for organizations and groups that develop approaches to use in a particular context. This publication presents six commitments and ten principles to coordinate areas regarding the responsible development and application of AI. The commitments to provide the framework anchor elements are: Ensuring advanced humanity, encouraging equity, influencing individuals, improving workforce benefits, monitoring performance, innovating and learning.
Patient advocate and founder of the Enlightened Results, Dr. Grace Cordovano said he praised the National Academy of Medicine and the AI Code of Conduct team for being one of the first major health AI efforts to genuinely represent and systematically embed the perspectives of patients and care partners.
She said that trust conversations in AI have gained priorities across the healthcare ecosystem, and the code of conduct will be a conduit for building trust, especially for patients, families, care partners and the patient community. “I would like to emphasize that the Code discusses patients as key stakeholders, end users, and co-creators of health AI, and recognizes that patient-driven governance and rights are essential,” Cordovano said.
Some of the advocacy priorities include patient voice at every stage of the AI lifecycle. “This code establishes the transparency requirements for developers and systems, explains which data is being used, how AI tools are being used, how they are being used, how they are being used, how they are being used, how they are being used. “We also look forward to it through ecosystems where AI is not only in the background, but also supporting trust in patient education and transparency and building, rather than what we hear in the media.”
Impact on researchers
Dr. Philip Payne, PhD, Associate Dean for Health Information and Data Science at Washington University School of Medicine, and Chief Data Scientist at the School of Medicine, shared the perspectives of the team who worked on the impact of AI research and research using AI. He said that given the rapid pace of innovation, it involves a way to ensure that there is a proper mechanism to share failure as well as success.
According to Payne, researchers are aware of the fundamental nature of AI methods and the increasing use of even more vast data they have used to train, evaluate and deploy these technologies. “This means that, not only ethics in the context and research scenarios for the time being, but how we must consider the ethical, legal and social implications of the AI scale, as we think for future applications, if those projects are successful,” he added.
Kaiser Permanente's real world example
Dr. Andrew Bindman, executive vice president and chief medical officer at Kaiser Permanente, shared the perspective of the healthcare system. He points out that health systems can play a leadership role in specifying and promoting business and clinical needs and requirements for using health AI, and when used in care delivery, AI can ensure that it is used in a way that benefits patients equitably and builds trust in the health system. “This includes carefully considering issues such as patient privacy, consent, agency, accountability, addressing legal and financial liability, and overcoming workforce-related challenges.”
Bindman gave a real-world example of how Kaiser Permanente has used a responsible AI framework. He said this reflects the commitment outlined in the NAM report.
“At Kaiser Permanente, AI tools need to promote our core mission to provide high quality, affordable care to our members. This means that AI technology must demonstrate health benefits, such as improving patient outcomes and experiences, and prioritize safety, equity, and health outcomes in the development and deployment of AI-based technologies. “We evaluate potential AI technologies, continuously monitor performance, and adhere to established clinical standards and guidelines. We use AI to ensure health by improving care, strengthening patient clinician relationships, optimizing clinician time, and addressing equity and experience of care and health outcomes.”
Bindman highlighted how Kaiser Permanente operated and applied its responsible AI principles in the implementation, evaluation and monitoring of assistant clinical documentation tools. This tool supports physicians and clinicians to safely capture clinical notes during inpatient visits with patients, helping them focus on conversations with patients rather than documenting or administrative tasks. “We adopted QA. [quality assurance] Deployment-related processes. This, in reflection, is very well in line with the health system commitment outlined in the NAM report. “One of the things identified in the NAM report is to advance humanity at a high level. We found this technology to help physicians and other clinicians develop an environment where they can provide effective communication and transparency while meeting the individual needs of each patient who comes to them,” he said.
Kaiser Permanente assessed the performance of patients with limited English proficiency based on clinician feedback explaining inconsistent performance among non-English encounters. “We conducted specialized QA tests to assess the performance of non-English speakers, but they paused when they used it in that setting, so we found out there were some issues identified,” Bindman said. “We spoke with vendors and fixed those issues. In this case, we engaged with the affected individuals who were patients. We always required patient consent to use the tool. We evaluated the patient's experience and observed significant positive effects related to clinicians using this tool.
Kaiser Permanente gained clinical experience and found that the tool helps reduce burnout feelings by reducing the amount of time spent on administrative tasks.
Bindman said one of the questions Health Systems must address is representing best practices for ongoing AI monitoring. “As an organization committed to being a learning healthcare system, we accept the responsibility to learn from our latest contributions to the evidence and accept responsibility for our knowledge of this new technology,” he said. “We and other health systems have made important advances in launching responsible AI governance systems and developing new tools, but as an industry there is a lot to learn collectively. We are collaborating on best practices on what early and continuous monitoring of AI tools can help. Between governments or various regulatory or certification bodies, this can be particularly costly for healthcare institutions with limited resources.
