How to Write Your Clinical Notes Faster
On average, doctors spend two minutes on documentation in EMRs for every minute spent with patients. Recently startups have been focused on automating clinical notes primarily using natural language processing (NLP). This new technology can be used to build templates, chatbots, voice to text technologies and medical summaries that help physicians improve their workflow.
As a doctor, the ability to quickly and accurately document in the electronic medical record (EMR) is essential to not become overburdened by their workloads. The average doctor already spends two minutes in the EMR for every minute with a patient. Fortunately, there are some technologies that are assisting doctors with clinical notes by automating portions of the workflow through machine learning and AI.
Over the last few years, the advancement of the field of natural language processing (NLP) has been the primary technology behind automating clinical notes. This type of technology uses machine learning algorithms to understand and interpret human language by training from very large datasets. In particular, the NLP model known as the transformer has been instrumental for the most recent advancements. Using NLP transformers, companies and innovative startups are building assistive tools for doctors such as through pre-populated templates, dictation, and summarization.
Pre-populated templates assist with documentation in the EMR as doctors can quickly and easily fill in the relevant information in the clinical notes such as diagnoses, allergies, family and social history, and medications. NLP technology can use extraction to pull the information forward automatically from the historical unstructured notes in the patient chart. This can save a significant amount of time, as the doctor does not need to manually enter data into the notes that have already been entered elsewhere in the EMR. Additionally, medical centers and physician practices can provide chatbots to patients before their visits to complete necessary forms and questionnaires; these answers would then seamlessly flow into the medical notes. Chatbots have traditionally been built with branching logic and knowledge graphs, but NLP transformers have begun to revolutionize the space with incredible fluency and accuracy. Patients will soon have the impression that the chatbot emulates a real human.

NLP can also assist with dictation by providing real-time transcriptions of what the doctor is saying that flows in-line into their clinical notes. Dictation has been present in healthcare for more than a decade through services such as Dragon Dictation. Though, doctors have traditionally needed to speak their notes exactly as they would need to be structured and appear in the EMR. Younger doctors have generally found that writing their notes manually and using pre-populated templates is more convenient than speaking notes word-for-word into a microphone. Recently, companies have been advancing the space with ambient voice to text technologies; where a virtual assistant like Alexa or Siri listens in the exam room in the background and attempts to structure the clinical notes automatically for the doctor into the EMR. In order to seamlessly re-organize, classify, and document automatically for the doctors, the most innovative startups are using Transformers. As the technology is still in its infancy and restructuring note content is a really challenging problem, a large portion of the companies are doing a bit of wizard-of-oz and have medical scribe assistants in the background cleaning up the notes before a doctor reviews them. Furthermore, portions of the clinical note may not be captured during the doctor-patient conversation such as the physician’s assessment of the disease or the after-visit plan for the patient, so the technology would need to limit predictions of the content that should be entered manually. While still with faults and in its infancy, we should see a reduction in the reliance on medical scribes through this technology.
In addition to assisting with dictation and extraction, NLP transformers can also help doctors by summarizing their previous clinical notes and medical history. Doctors only have a few minutes to review a patient’s chart, which is not enough time to fully understand the record; a medical synopsis would help physicians review the most important information. The medical history is what prepopulates a number of the notes doctors write such as the SOAP note, History & Physician, and Discharge Summary. And medical summaries have been shown to improve patient outcomes when present in the EMR for downstream physicians. For this reason, clinicians are tasked today with writing these summaries manually during a transition of care that can take anywhere from 15 minutes to 2 hours, so automating a medical summary through NLP would eliminate this workflow. Our company, Abstractive Health, is focused on this solution to improve the note-writing workflow for doctors.
So, NLP offers several ways for doctors to write their clinical notes faster and more efficiently. Over the next few years, we are going to see startups and companies flourish that are focused on getting doctors back to their roots of patient care and away from the EHR systems. Innovative physician practices and healthcare systems should explore these novel technologies through partnerships and pave the way for improving the day-to-day jobs of doctors. With the epidemic of physician burnout across the world, medical centers should begin investing in these technologies today.
