AI Doctors' Notes Miss Key Cues: Are We Losing Patient Stories?

Artificial intelligence tools designed to help doctors take clinical notes are raising serious concerns, with a new study suggesting they are failing to capture "potentially vital information." This crucial data often comes from non-verbal cues like facial expressions, gestures, and even the tone of a patient's voice, which AI systems are currently missing.

While many medical professionals are enthusiastic about these so-called AI scribes, hoping they will drastically reduce administrative paperwork, experts warn that the patient's experience is being "poorly considered." These "ambient voice technology" systems combine speech recognition with generative AI to transcribe spoken words into structured medical records.

The goal is to save time and improve interactions, but a significant flaw has been identified.

A review of 27 published papers by researchers at the University of Edinburgh highlighted that these AI tools listen only to audio. This means they cannot pick up on the subtle, yet often critical, physical gestures, facial expressions, and vocal nuances used by both doctors and patients during consultations.

The researchers describe this as a "written language bias," where spoken interactions are converted to text and then processed as mere textual records.

Experts also suggest that the presence of AI might cause patients to "adapt their behaviour," potentially leading to a "lack of trust" and subsequently, "less disclosure" of important information. Dr. Lucas Seuren, a research fellow at the University of Edinburgh's Centre for Biomedicine, Self and Society, emphasized this point.

Many clinicians are excited about ambient AI scribes, because they promise to cut down on paperwork,

Dr.

Seuren stated. "But the experiences of patients are poorly considered, and there are real risks that the patients' stories are lost."

This oversight, he warns, could further disadvantage individuals who already face marginalisation within health and social care services. The findings, which underscore the need for a more comprehensive approach to AI integration in healthcare, were published in BMJ Digital Health and AI. The study urges a deeper look into how these technologies impact the delicate doctor-patient relationship and the overall quality of care.

#AIinHealthcare #MedicalTech #PatientCare