������Ƶ

The future of EHR documentation is likely to be influenced by advancements in technology, using artificial intelligence (AI), machine learning (ML), and natural language processing (NLP), patient-facing initiatives such as Open Notes, and improved teamwork and care coordination. Future advancements in EHR documentation can address a range of factors, including increased provider volumes, clinician burnout, and clerical burden (especially documentation of care and order entry).

Once viewed as a technology of the future, AI is currently being tested, trialed, and implemented at healthcare provider organizations nationwide to reduce the burden of documentation.^58,59

Ambient AI scribes use novel generative AI techniques such as automatic speech recognition and NLP to capture real-time patient-provider conversational interactions and assemble them into a structured note. Initial investigations into ambient AI scribes have found promising results—reducing clinicians’ burden and the amount of time spent constructing notes while simultaneously improving the experience of both clinicians and patients.⁶⁰

In freeing providers to spend more time and interact more directly with patients, ambient AI scribes may improve providers’ diagnostic ability. However, ambient AI scribes also introduce new risks that may lead to diagnostic error. AI-generated notes may be inaccurate, inconsistent, and biased, hindering providers’ diagnostic ability.^61-63 While the strong benefits of this new technology may lead to rapid implementation, it is imperative to carefully consider the consequences it may have on diagnostic safety to mitigate any added danger.

ML can significantly improve diagnostic safety through:

• Enhanced pattern recognition (e.g., analyzing large datasets to identify patterns that may not be apparent to human clinicians).⁶⁴ 
• ML-powered CDS to provide real-time recommendations based on patient data.⁶⁵
• Electronic trigger tools to identify signals of diagnostic error.⁶⁶   
• Predictive analytics to help prioritize diagnostic testing and interventions.^67,68

More specifically, NLP, a type of ML, can be used to enhance EHR documentation through clinical note summarization and analysis. NLP is defined as any computer-based algorithm that manages, enhances, and converts natural language to a form suitable for computational analysis.⁶⁹ NLP technologies can “read” unstructured documentation and convert it into discrete data, including automatically summarizing lengthy clinical notes and documentation, extracting key findings, diagnoses, and treatment plans to create concise and structured summaries. 

Clinical summarization tools can improve documentation efficiency, enhance readability, and facilitate information retrieval for providers, supporting better clinical decision making and patient care.⁷⁰ Studies show that NLP has better sensitivity than ICD codes at identifying common patient symptoms, particularly when the symptom burden is high.⁷¹ 

Because NLP technologies can analyze free-text notes to extract relevant diagnostic information, such as symptoms, findings, and provisional diagnoses, specially designed NLP algorithms can glean important insights from notes. By automatically identifying and coding diagnoses from unstructured text, NLP streamlines the documentation process, reduces manual effort, and improves the accuracy and consistency of diagnosis documentation. 

NLP techniques can perform semantic analysis of clinical notes to identify key concepts, relationships, and contextual information related to diagnoses. By mapping clinical terms to standardized medical terminologies (e.g., SNOMED CT, ICD-10), NLP facilitates the recognition and normalization of diagnoses, ensuring consistency and interoperability in EHR documentation. 

Lastly, NLP-powered CDS systems can assist providers in documenting diagnoses by offering real-time suggestions, alerts, and recommendations based on the patient’s clinical data and documentation. NLP decision support tools can help providers consider differential diagnoses, adhere to clinical guidelines, and ensure thorough and accurate diagnosis documentation. Examples illustrating future implications for clinical practice include a model to predict in-hospital mortality using notes in the first 24 hours of a patient admission⁷² and a model to identify prediabetes discussions in clinical documentation.⁷³

Patients play a role in both providing and reviewing data to ensure its accuracy and completeness. An incomplete history, due to a lack of patient disclosure of important symptoms or family history or when a provider fails to capture relevant information, will result in inaccurate records regardless of the design or quality of the EHR system. 

Government initiatives, beginning with legislation in 2009, incentivized health systems to offer patients electronic access to their own data via secure electronic patient portals.⁷⁴ Opportunities continue to emerge for patients to review their data to ensure accuracy and facilitate shared decision making.⁷⁵ With the growing awareness of the importance of patient input in achieving diagnostic excellence, there is great interest in the utility of patient access to their notes. Codified by the 21st Century Cures Act, “the open notes” movement now legislates immediate patient access to their notes. 

Open Notes is an international movement designed to promote transparency in healthcare and is endorsed by the American College of Physicians.⁷⁶ Studies show that when patients read their notes, they identify a large number of errors, especially related to diagnosis.^24,77,78 Common errors include mistakes in diagnoses, medical history, medications, physical examination, and test results, notes on the wrong patient, and errors on which side of the body was the site of the injury or symptom.

Patient engagement with their notes is not universal, with significant disparities observed.^79,80 One factor is the complexity and structure of standard medical notes. Numerous studies suggest that typical notes are not at an appropriate reading level for most patients and patients often misconstrue or misinterpret even some of the most standard phrases in their notes.^81-84 However, expecting providers to manually change their documentation practice is not only unreasonable but will likely increase providers’ already heavy documentation workload. 

The rapid advancement of large language models now affords the ability to automate the creation of simplified patient-centric notes from existing provider documentation without negatively affecting provider documentation burden. Preliminary studies suggest it is not only feasible, but in controlled settings, also improves patient comprehension of the documented information.⁶⁰ Future studies will be needed to determine the scalability of this technology and the impact of automated conversion of standard notes on patient engagement with their EHR.

The National Academies of Sciences, Engineering, and Medicine (NASEM) report Improving Diagnosis in Healthcare described successful diagnosis in the 21st century. It is a team-based, patient-centric model leveraging the knowledge and skills of all interprofessional staff and expanding the diagnostic team to include pathologists, radiologists, allied health professionals, medical librarians, and others.^6,85 

Encouraging collaboration among members of the care team, including physicians, nurses, specialists, and allied health professionals, promotes a multidisciplinary approach to diagnosis documentation. Effective communication among care team members fosters a shared understanding of the patient’s clinical presentation, diagnostic evaluation, and treatment plan. By facilitating open dialogue and information exchange, communication helps align the care team’s efforts and priorities, leading to more cohesive and coordinated diagnosis documentation in the EHR.⁸⁶ 

Collaborative documentation allows each team member to contribute their unique perspectives, clinical insights, and expertise to ensure comprehensive and accurate documentation of the patient’s diagnosis. Expanding the diagnostic team will bring both challenges and opportunities for improving diagnostic documentation by facilitating effective teamwork. Implementing administrative changes, such as providing documentation assistance and fostering empowered teamwork, can alleviate the burden on clinicians by redirecting data entry responsibilities.