Artificial Intelligence-Driven Clinical Documentation and Data Extraction with ChatGPT
Healthcare companies across the nation are on a mission to employ AI and ChatGPT as tools to automate Electronic Health Record processes, allowing doctors to spend less time with their EHR and more time with their patients. According to a study by Stanford Medicine, 7 out of 10 doctors stated that using their EHR takes valuable time away from their patients.[1] Most of that time is not even patient-facing. It is clerical. In fact, a typical physician will spend 62% of their time per patient reviewing electronic health records.[2] Of that time spent staring at their screens, Physicians reported that documentation, data in, and chart review, data out, account for more than half of that time.[3]
Saving Time, Saving Lives
Studies by the Mayo Clinic report that physician burnout is at an all-time high, with 63% of physicians reporting at least one sign of burnout per week.[4] Physician burnout is a significant concern in healthcare, and electronic health records have emerged as a primary contributor to this problem.[5] EHR, while designed to streamline medical documentation and improve patient care, has added administrative burdens while increasing the time spent on clerical tasks. According to a survey by Wakefield Research, 91% of clinicians agreed that the most urgent need to improve patient care was to improve EHR-related administrative tasks.[6] That is where AI could benefit clinicians by speeding up documentation and data extraction processes in EHR systems.
An Artificial Scribe with AI and ChatGPT
Since OpenAI launched ChatGPT in November, every industry has sought to harness AI to improve efficiency. Healthcare is no exception. The app has already performed “at or near the passing threshold” for all three parts of the United States Medical Licensing Exam, so it’s easy to imagine how ChatGPT and many AI tools like it, could house the potential for clinical decision support and predicate analysis.[7] However, AI’s most tangible benefit to healthcare right now is not its ability to make decisions like a doctor, but instead to assist a doctor by serving as their AI Scribe.
AI for Data Documentation
The first phase of AI incorporation into EHR is in documentation. In April, Microsoft and Epic announced a partnership to integrate generative AI into EHR software. The partnership focused on AI-drafted message responses to patients and enhanced conversational support in dictation.[8] That same day eClinicalWorks announced its own investment into Microsoft Azure, integrating its EHR and PM solutions with ChatGPT and machine learning models from Azure OpenAI, including an AI-based dictation service.[9]
Most momentum in AI clinical documentation stems from technology companies utilizing ambient clinical intelligence (ACI) in EHR. ACI uses voice-enabled AI to operate in the background of a provider-patient interaction to capture audio, video, and other data sources with everyday objects like cell phones or wearable audio-capturing devices. Through voice recognition, natural language processing, and other advanced technologies, ACI understands and interprets the main points in provider-patient conversations and transcribes those notes into the EHR. The goal of ACI is to create a more natural provider-patient experience, enabling healthcare providers to focus more on patient interaction rather than spending excessive time on data entry.[10]
Already healthcare companies are testing ACI in clinical settings. In March, Abridge, a powerhouse in AI-powered documentation, announced its partnership with The University of Kansas Health System to implement ACI tools in the clinical field, generating draft summaries of provider-patient conversations within a minute.
“With Abridge, we have found a powerful solution that addresses the biggest challenge facing our providers — excessive time spent on documentation,” said Dr. Gregory Ator, Chief Medical Information Officer and Head and Neck Surgeon at The University of Kansas Health System. “Our partnership with Abridge represents a major step forward in reducing burnout, improving provider satisfaction, and ultimately enhancing the delivery of patient care.”[11]
Later in March, Nuance, a Microsoft company, announced Dragon Ambient eXperience, a similar documentation tool combining ACI with ChatGPT-4.
“Our state-of-the-art blend of conversational, ambient, and generative AI will accelerate the advancement of the care delivery ecosystem beyond what Nuance or Microsoft could have achieved separately – expanding our comprehensive portfolio of solutions that fulfill our vision to improve care quality and support enhanced outcomes for generations to come,” Mark Benjamin, CEO of Nuance, said.[12]
In May, PartnerMD, a concierge medicine network, also announced their rollout of a voice artificial intelligence system, the Suki Assistant. Suki Assistant listens to the provider-patient interaction in real time and generates a clinical note summary that syncs back to the EHR.[13]
AI is changing clinical documentation, replacing scribes and hours of manual notetaking with voice-generated background documentation. According to a KLAS report, ambient speech technology in healthcare has shown such initial promise that the AI-speech recognition software is rapidly becoming the go-forward method of clinical documentation.[14] The market for such tools seems to grow busier each day. And it only seems to be the beginning.
AI for Data Extraction
AI is changing the landscape for clinical documentation, but could more data captured by AI, lead to more time wasted sifting through that data? Of the time physicians spend in their EHR, they already report more wasted time reviewing charts, 33%, than documenting data, 24%.[16] Surely as AI captures more data, doctors will need more automated tools to review that data. If the industry does not adapt accordingly, more data could perpetuate the problem of too much data, ultimately leading to an information overload.
AI-powered data visualization tools are popping up across the market, but not at the rate documentation tools are. One study at Stanford University tested an AI-powered data visualization tool on 12 physicians. The physicians reported that the AI tool saved them 18% of their time compared to their standard review records. Eleven of the twelve physicians favored the AI tool over their standard review.[17] While the prospect is positive, only a few AI-based tools on the market address the information overload concern. AI will help EHR capture more and more data, but how will we keep up and review all that data?
A Clinical Data Visualization Tool That Works
DHRpro is a patient visualization dashboard that sits on top of existing EHR/PM systems. For the first time, an entire patient history is displayed on a fully customizable dashboard with all the information no more than a single click away. See clinical and financial data, medications, diagnostics, and imaging—all in a historical context in an easy-to-follow longitudinal format. With so much data in our electronic health records and only more to come, now more than ever before, doctors need reliable data visualization software to reduce wasteful chart review time. DHRpro is that software.
