Natural Language Processing of the “One in a Million” Dataset
Research type
Research Study
Full title
Natural Language Processing of the “One in a Million” Dataset of Primary Care Consultations
IRAS ID
291476
Contact name
Yvette Pyne
Contact email
Sponsor organisation
Research and Enterprise
Duration of Study in the UK
1 years, 11 months, 23 days
Research summary
Technology will continue to become ever more pervasive in the world of primary care. Whether it is the new routes through which the patient contacts their general practitioner (GP), or the doctor’s use of computers to add, store and retrieve patient clinical information and contact other professionals. A significant proportion of a doctor’s clinical time is spent interacting with the patient electronic health record (EHR) including entering clinical information regarding a patient consultation. While these powerful tools can improve patient care, they can also create a barrier in the human connection needed in medical care and can contribute to physician burnout, increased cognitive load, and documentation errors.
Some countries employ scribes (more typically in the hospital setting) and there has been some progress in the creation of ‘Digital Scribes’ using computers to record a consultation enabling a clinician to fully engage with a patient, maintain eye contact, and eliminate the need to split attention by turning to a computer to manually document the encounter. This translation of the multiple cues in a consultation to a concise and complete medical record is incredibly complex.
My project plans to use the ‘One in a Million’ dataset of over 300 video recorded primary care consultations and look at the relationship between the direct audio recording transcriptions and the consultation notes written by the GP using "Natural Language Processing" - a branch of Artificial Intelligence that uses machine learning to extract information from speech or text. Through this research, I hope to begin to understand how a fully automated “Digital Scribe” might be developed - this would go beyond simply transcribing raw conversational speech to create concise and complete consultation notes.
REC name
South West - Central Bristol Research Ethics Committee
REC reference
21/SW/0069
Date of REC Opinion
3 Jun 2021
REC opinion
Favourable Opinion