{
  "abstract": "Artificial Intelligence (AI) documentation systems are being rapidly adopted throughout the healthcare industry as solutions to help alleviate the clinician documentation burden. The use of Ambient AI documentation systems enables passive capture of clinical conversations and the creation of medical documentation. Unlike routine clinical encounters, trauma resuscitation is characterized by fragmented speech, overlapping conversations, temporal complexities, and immediacy of decision-making. All of these aspects could present challenges to the many AI models based on commonly held assumptions about communication. Without trauma-specific safeguards, AI documentation systems may introduce automation bias, speaker identity errors, time relationship errors, and convert uncertainty into definitive statements. This article provides an examination of the risks associated with the use of AI documentation systems in trauma care and proposes several safety guardrails, including mandatory human review of high-risk elements, preserving uncertainty in documentation, structuring documentation of safety-critical data, providing multidisciplinary oversight of implementation, conducting trauma-specific validation testing, and ongoing auditing of performance.",
  "authors": [
    {
      "affiliations": [
        "Stony Brook University Hospital, Stony Brook, New York, USA"
      ],
      "name": "Samita M Heslin"
    }
  ],
  "title": "Artificial Intelligence documentation in trauma resuscitation: efficiency requires guardrails",
  "uid": "af2f3a1f-7951-56c2-9a68-1622ce2543d6"
}