Artificial Intelligence EHR System May Speed Up Physician Processes

MIT researchers have developed an artificial intelligence EHR system that combines machine learning and human-computer interaction to help physicians speed up the process of looking up medical records and ease clinician burden.

Divided interfaces and data entry procedures may cause physicians to spend more time searching for health records and less time interacting with patients. The new EHR system, MedKnowts, uses artificial intelligence to create one interactive interface that brings together the process of looking up medical records and documenting patient information.

“I think a lot of clinicians feel they have had this burden of EHRs put on them for the benefit of bureaucracies and scientists and accountants,” David Karger, professor of computer science in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and senior author of the MedKnowts paper, stated in the press release.

“We came into this project asking how EHRs might actually benefit clinicians.”

The EHR optimization includes autocomplete features for clinical terms and auto-populates fields with patient information to increase efficiency among physicians.

The researchers also included a note-taking editor that displays information from the patient’s medical history on a side panel. If a clinician enters a certain condition while recording a patient’s medical information, the EHR tool will automatically pull up information from past records, such as medications and lab values, that are relevant to that condition.

The information is color-coded depending on the category; for example, a medical condition is highlighted in red and medication is highlighted in green, according to the researchers.

EHRs typically house past medical records on separate pages and list medications and lab values alphabetically or chronologically, the press release noted. The researchers’ system displays information that is relevant to whatever the physician is taking notes about, eliminating the need for her to search through data to find what they need.

“[Doctors] will look through a medications page and only focus on the medications that are relevant to the current conditions,” Luke Murray, a graduate student at CSAIL and lead author of the paper, stated in the press release.

“We are helping to do that process automatically and hopefully move some things out of the doctor’s head so they have more time to think about the complex part, which is determining what is wrong with the patients and coming up with a treatment plan.”

Once the EHR system was complete, the researchers implemented the software in the emergency department at Beth Israel Deaconess Medical Center in Boston, Massachusetts. They worked with an emergency physician and four hospitals scribes who enter notes into EHRs.

They found that it was challenging to get the staff to change their previous EHR use methods. The COVID-19 pandemic also threw a minor wrench in the researchers’ plans. The researchers could not continue their visits to the emergency department to observe workflows after the pandemic hit.

However, after one month, the scribes gave the system an average rating of 83.75 out of 100 for usability. They noted that the autocomplete function helped speed up their work and the color-coded categories made it easier to scan notes to find relevant information.

Going forward, the researchers hope to improve the system without taking away from the importance of doctors thinking twice before making a decision.

The researchers want to further the machine learning algorithms to better highlight relevant parts of a medical record. They also hope to develop the system, which they designed with an emergency department in mind, to help primary care physicians as well.

The MIT and Beth Israel Deaconess researchers also want to create an EHR system that physicians can contribute knowledge to, which would help provide more information for all users, the press release concluded.