Where AI innovation has taken one EHR vendor in 2023
![Where AI innovation has taken one EHR vendor in 2023 Where AI innovation has taken one EHR vendor in 2023](http://www.healthcareitnews.com/sites/hitn/files/AH_Headshot_Paul-Brient_HITN.jpg)
Healthcare IT News sat down with Paul Brient, chief product officer of athenahealth, to gain his insights into how artificial intelligence can improve patient experiences and healthcare delivery, simplify healthcare interactions, and make healthcare more accessible, as well as hear how the electronic health record vendor is using AI today.
He said that new AI-driven features in athenaOne are already helping to improve healthcare efficiency and address clinician burnout by cutting administrative tasks, such as the proactive identification of missing prior authorization information achieving an 18% reduction in task completion across EHR users.
Looking ahead to 2024 and beyond, Brient said AI-generated reports will be available to help caregivers prepare for care meetings with providers, while generative AI holds the potential to soon reach patients well beyond the clinical setting.
Predictive analytics is one area that could guide patient care by suggesting additional services and treatment modalities that similar patients have utilized.
It’s been shown to spot patient medication nonadherence and help decrease the in-patient sepsis death rate, but it’s often hard to get clinicians on board with data-driven medicine. Some say predictive analytics in EHRs aren’t yet effective enough for clinical decision support at the point of care, and physicians must be the sole decision-makers.
As swiftly moving AI technologies hold vast potential to improve healthcare delivery and outcomes, how we ensure their safe use in clinical care might just be the issue of the year for 2023, if not the decade.
“It’s important to note that while AI has the potential to improve care delivery, it should always be used in conjunction with clinical expertise and human judgment,” Brient said.
“AI algorithms should be transparent, explainable and continuously validated to ensure their accuracy, reliability and ethical use in healthcare settings.”
Q. How can AI be used to improve predictive analytics in clinical use?
A. There are a variety of ways that AI can help improve predictive analytics in a clinical setting.
Risk Stratification and Early Detection: AI models can be trained to identify high-risk patients. By analyzing historical patient data, AI algorithms can identify patterns that suggest a patient is at high risk and might benefit from an immediate intervention such as care management, rather than waiting for their next scheduled visit.
Clinical Decision Support: AI-powered CDS can provide real-time guidance to healthcare providers by analyzing patient data and offering evidence-based recommendations. These systems can assist in coding diagnoses, identifying gaps in care, accelerating orders and alerting clinicians to potential drug interactions.
Resource Optimization: AI can help healthcare organizations enhance resource allocation by predicting patient demand and adjusting schedules and resource availability to better match demand and reduce un-utilized appointment slots. This can lead to improved operational efficiency, reduced wait times and improved access to care.
Q. How will AI improve patient experiences and outcomes as time goes on? Can it help to make healthcare more accessible in the future?
A. AI can fundamentally change the physician experience with an EHR by making it an intelligent partner for the clinician. With AI, the EHR can “understand” the patient record, digest/parse unstructured data and present this information to clinicians in the context of the visit, patient situation and provider’s preference.
For example, if a provider is seeing a Medicare patient for an annual wellness visit, AI can wade through all of the information about the patient, understand which preventative measures have been taken and what is missing so the physician can quickly and easily order these services, and then spend the majority of the visit engaging with the patient, and ensuring there are no other underlying issues that need to be addressed.
From a patient perspective, we will almost certainly start to see AI-enabled triage and chatbots that help patients better understand where they should go to best seek care – akin to the “ask a nurse” triage lines.
Additionally, our long-term outlook envisions genAI, like ChatGPT, bridging communication gaps to improve accessibility to healthcare, simplifying medical information and further enhancing the patient-provider experience.
For example, there is enormous potential to use ChatGPT for communicating with patients beyond the clinical setting and to overcome communication barriers or even to close care gaps.
In a large study conducted with Spanish speakers living in the U.S., about 25 million people reported receiving a third less healthcare than other Americans. In addition, the study found that Spanish speakers had 36% fewer outpatient visits compared to non-Hispanic adults. This clearly demonstrates the need for technology to improve language barriers.
ChatGPT, or other AI-based language translation systems, can serve as a resource for multilingual interaction and simultaneous translation, and can help to communicate a message in a patient’s first language, reducing the language-based gaps in healthcare and improving the patient’s access to healthcare.
Q. Athenahealth has long used machine learning to enhance electronic health record offerings. How has EHR automation increased efficiencies and reduced administrative burdens for providers over time?
A. Athenahealth has been leveraging various forms of ML and AI for nearly a decade to simplify the user experience for EHR and practice management systems. Our use of ML is focused on solving critical pain points for our customers, streamlining work and removing administrative burdens that get in the way of focusing on patient care.
For example, we’ve enabled automatic selection of insurance packages from a photograph or scan of a patient’s insurance card. Using optical character recognition and advanced ML to instantly select an insurance package and confirm the patient’s eligibility, this feature eliminates the need to manually enter data, and improves both accuracy and efficiency for front-desk staff, while improving the patient experience.
This feature is already delivering a 31% reduction in insurance-related claim holds across practices using the capability, saving practice staff more than 6,500 hours of administrative time in the last 12 months.
To simplify patient record keeping, we use voice commands to allow providers to navigate and interrogate the athenaOne mobile app quickly and easily. For example, rather than typing in an order or prescription, the provider can simply say, ‘Order 20 mg of Lipitor once per day.’ The app also predicts the most likely next action for a provider to take in response to an inbox item and suggests it as a one-click action at the top of the provider’s list.
To simplify document management, we utilize ML and natural language processing to classify and file inbound patient documents across our network. This ensures the patient’s chart is the most complete and as easily accessible as possible.
Q. How has automation evolved in the past year since ChatGPT burst onto the scene? What do athenahealth’s customers want and how has their feedback shaped new offerings?
A. We continuously assimilate customer – and especially provider – input to improve their experience and satisfaction. While athenahealth has used traditional AI models to streamline administrative work for years, this year’s Codefest [a homegrown week-long coding event that focuses on design, development and testing of new HIT features] focused on ensuring that our engineering teams are fully up to date on genAI and implementing four genAI-enabled features that are top priorities for our customers.
Two of these new features are available to select athenahealth customers today and show significant, quantifiable results. They are:
Proactive identification of missing prior authorization information: As many as 10% of prior authorization tasks are sent back to providers due to missing clinical information, adding work and causing delays in getting authorization approvals. New capabilities embedded into athenaOne identify missing or incorrect information before the prior authorization is submitted and suggest the correct inclusions to maximize the chances the authorization will be approved, saving time and reducing costs for practices, while improving patient experience.
Patient case response drafts: Providers on the network respond to about four million patient cases each month and spend more than 35% of clinical inbox time managing patient case documents. This capability enables providers to have pre-drafted responses available for consideration, review and editing, to enhance productivity without replacing the provider’s expert judgment.
In addition, we have identified more than 40 more potential features that generative AI could enable. We are actively evaluating these for deployment in future product releases to reduce the administrative burden providers and their staff encounter, while also providing new tools that enable providers to deliver high-quality care for their patients.
Q. How is athenahealth using genAI to help providers surface relevant clinical information at the point of care?
A. Providers today have access to an unprecedented amount of information about their patients from a variety of sources. Unfortunately, there are times when documents within these records are not labeled in an intuitive or helpful way – causing providers to have to open each document in the hopes that it might contain the information they are searching for.
One of athenahealth’s newly deployed genAI features solves this problem by summarizing the contents of these documents intelligently so providers can quickly and easily find the right one and in many cases glean the needed information without having to open and read the entire document.
In addition, athenahealth care managers will soon receive AI-generated “Huddle Reports” to prepare for weekly care meetings with providers.
These reports help facilitate conversations between care managers and providers, and are a critical tool to maintaining an open flow of information to improve patient care. Generating these reports automatically will streamline conversations between care managers and physicians, enabling clinicians to deliver more personalized care across the healthcare continuum. The time saved by care managers will allow them to provide individual care to a greater number of patients.
Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS Media publication.