Artificial intelligence-powered digital scribes have been heralded as an effective solution to the issue of physician burnout, which is greatly exacerbated by the amount of time required for clinical documentation.
These digital scribes, however, come with their own set of challenges "due to the complex nature of clinical environments and clinical conversations," according to a new study published in npj Digital Medicine.
In the report, researchers from the Australian Institute of Health Innovation at Macquarie University in Sydney outlined five major obstacles to the development of automated speech-based documentation in clinical settings:
1. Recording high-quality audio and converting audio to transcripts using speech recognition
2. Inducing topic structure from conversation data
3. Extracting medical concepts
4. Generating clinically meaningful summaries of conversations
5. Obtaining clinical data for AI and machine learning algorithms
According to the researchers, these issues will be better solved individually, rather than with a single catch-all solution. Special attention must be paid to the final challenge, they wrote: "Collective efforts must be made to make clinical data available for AI researchers to advance automated clinical documentation, while also protecting the data from misuse with ethical considerations in place."
More articles on health IT:
How HIPAA has grown 'less and less adequate' in protecting health data
Uber Health encourages partners to leverage PHI to benefit patients
Human-AI collaboration more effective than either working alone: Stanford study