Studies have shown the time spent in monotonous administrative work to entre patient health report into electronic health record (HER) systems have increases the working hours for the doctors. The introduction of artificial intelligence tools will help the doctors to save their time and manage their shifts efficiently.
A researcher from Google Brain has lately developed a new language-modelling task that can estimate content of freshly written notes by examining patient medical records. The records will include data related to laboratory measurements, demographics, medications, and past notes. According to the study, already published on arXiv, Brain trained generative models by using Medical Information Mart for Intensive Care (MIMIC-III) EHR dataset and after that he compared the notes derived by the models with real notes from the dataset.
Dictation services and employment of assistants to write up notes for clinicians are the most commonly used methods used by clinicians. These methods helps them to reduce their time spent on administrative work. However, the use of artificial intelligence tools will help in better management of such issues. AI tools will also reduce cost spent on employing staff and other resources.
The language model also benefits in auto correction and error checking while adding notes, says Peter Liu, researcher of the study. The focus of this research is to build effective language models for clinical notes.
Liu also adds up by saying, maximum context provided by the EHR is not sufficient to predict the note completely. This happens mostly in the case of imaging data provided in radiology reports by MIMIC_III. Therefore, further work is required for its efficient use.