Identifying relevant information in medical conversations to summarize a clinician-patient encounter

Abstract

To inform the development of automated summarization of clinical conversations, this study sought to estimate the proportion of doctor-patient communication in general practice (GP) consultations used for generating a consultation summary. Two researchers with a medical degree read the transcripts of 44 GP consultations and highlighted the phrases to be used for generating a summary of the consultation. For all consultations, less than 20% of all words in the transcripts were needed for inclusion in the summary. On average, 9.1% of all words in the transcripts, 26.6% of all medical terms, and 27.3% of all speaker turns were highlighted. The results indicate that communication content used for generating a consultation summary makes up a small portion of GP consultations, and automated summarization solutions—such as digital scribes—must focus on identifying the 20% relevant information for automatically generating consultation summaries.

Evidence synthesis, digital scribes, and translational challenges for artificial intelligence in healthcare

“… 55 Verbatim transcripts are less valuable in settings where there is a conversation, for example between doctor and patient, and less than 20% of such an exchange might contribute to the final record. 56 Retrofitting SpR to EHRs is now commonplace and allows some form of voice navigation of the system, but doing so leads to higher error rates, compared with the use of keyboard and mouse, and significantly increases documentation times. 57 …”

Digital scribe utility and barriers to implementation in clinical practice: a scoping review

“…Two studies evaluated the efficacy of voice recognition software and determined that summarization technology is feasible in non-linear settings, and that computer-based systems can be successfully used in clinical practice [ 29 , 30 ]. Five studies detailed practicalities of digital scribe integration and its associated challenges; [ 16 , 19 , 26 – 28 ] two of these detailed how the current way of practicing medicine must change in order to work with digital scribe technology [ 16 , 27 ], one detailed how implementation can be done in discreet steps allowing for physicians to train the digital scribe to ultimately work together [ 19 ], and two outlined details of implementation and potential barriers [ 26 , 28 ]. Finally, one study evaluated the cost effectiveness of implementing digital scribe technology into clinical practice [ 31 ].…”

“…The software and algorithms underlying digital scribe technologies must also be trained to understand medical terminology and must formulate a comprehensive clinical encounter from a transcript in a way that accurately uses these words and phrases. This may prove to be difficult given the non-linear format of natural conversations and redundancy of conversations during a visit, as studies have shown that 80% of conversation capture is unnecessary for meaningful note generation [ 26 ]. Further, this task may require additional time-intensive editing by the clinician if the speech-to-text conversion is inaccurate due to deficiencies in the lexicon of the digital scribe [ 24 , 30 , 42 ].…”

The digital scribe in clinical practice: a scoping review and research agenda

“…The suitability phase falls outside the scope of this review but is nevertheless vital for developing and implementing the digital scribe. One research group has published several studies investigating which parts of a clinical conversation are relevant for creating a summary and how physicians see the potential role of a digital scribe 53 , 54 . These studies should be the starting point for researchers and developers working on a digital scribe.…”