Skip to main content
In the article “Developing an Artificial Intelligence (AI) Tool to Derive Social Determinants of Health (SDOH) for Primary Care Patients: Qualitative Findings From a Codesign Workshop,” the authors describe numerous workshops hosted to brainstorm ways for AI to document and display SDOH information to providers. While the impact of many SDOH have previously been demonstrated, the authors did not adequately discuss all potential negative outcomes of an automated system documenting and displaying this information. There is patient risk associated with implementing this process.
Consent is a vital component of providing care, but the authors did not describe a clear path for obtaining patient consent in this AI process. Privacy protection is a fundamental component of providing medical care, but the authors offered little discussion on how this information would be shared only with appropriate health care members. The authors did discuss participants concerns that inherent bias may be amplified by AI in this process, but the true extent of this bias cannot be measured, risking its further propagation. The authors also did not offer a way for health care providers to address this information in a timely manner, creating a possibility of significant risk factors being documented without appropriate care being provided. While the authors discussed concerns that implementing this AI tool may damage the therapeutic relationship, this article failed to include a patient's perspective in sensitive information being obtained and documented.
We as providers cannot resist the gradual implementation of AI into our personal and professional lives; but intentionally using this programming to auto-populate sensitive SDOH information risks patient harm. Patient autonomy and consent must continue to be prioritized in the medical record with the integration of AI.