AI Scribes are emerging as exciting tools to enhance patient care and streamline clinical workflows in the rapidly evolving healthcare technology landscape. With dozens of solutions to choose from in the market (Nuance DAX, AWS HealthScribe, DeepScribe, and others), we aim to provide guidance and perspective to help healthcare organizations (HCOs) select, pilot, and scale a solution successfully.
This article outlines perspectives and strategies for selecting the right solution that addresses your organization’s specific context and challenges and fulfills the needs of your patients and providers.
What is an AI Scribe?
An AI Scribe is an advanced software solution that leverages artificial intelligence to assist healthcare professionals in documentation and administrative tasks. Its capabilities often include:
- Real-time transcription of patient-doctor conversations
- Automatic generation of encounter summaries and draft notes
- Extraction of key medical information from dialogues
- Integration with Electronic Health Records (EHR) systems
Generative AI, specifically Large Language Models (LLMs), has increased the potential of these solutions to reduce the administrative burden for clinicians. An LLM’s ability to generate human-like text allows some to produce a reasonable first draft of a clinical note by extracting summaries of the chief complaint, diagnosis, and treatment plan based on the transcribed interaction between clinician and patient. The LLM can be configured to follow a provider’s preferred format, such as the standard Subjective, Objective, Assessment, and Plan (SOAP) note structure.
This allows clinicians to spend less time facing their workstations and more time facing their patients. A clinician’s confidence in an AI Scribe’s ability to create a helpful summary or draft is critical to unlocking this return to a human-centric patient experience in the exam room while also reducing the clinician’s “pajama time” (the time they spend updating and closing their notes from home). Latest studies have shown that AI scribe can reduce time spent on paperwork by 70% to 90%.
Key strategic objectives for AI Scribe implementation
Given recent advancements in AI, HCOs are overwhelmed by researching the increasing number of AI scribes in the market and identifying ways to integrate the technology into their ecosystem. Attempts to roll out these solutions often result in higher costs, lackluster adoption, and disappointing outcomes.
Before adopting AI Scribes, HCOs must:
- Contextualize potential AI Scribe use within the HCO’s innovation approach
- Identify relevant use cases where AI Scribes can provide significant benefits
- Evaluate their readiness to deploy the solution (e.g., capabilities, infrastructure, and data)
While many hospitals are exploring AI solutions, implementation approaches and outcomes vary widely. Organizations should view AI Scribes as a strategic investment in innovation, acknowledging the novelty of these solutions. When considering implementation, hospitals must decide between buying off-the-shelf solutions or building custom ones based on their organization’s capabilities and preferences. Understanding your limitations and constraints is as critical as understanding the solutions’ potential and capabilities.
Choosing the right approach to AI Scribes in healthcare
AI Scribes have come a long way since Pariveda first developed a proof-of-concept with an innovative HCO client years ago. Today, HCOs have options that span the build vs. buy continuum, with some provided as features of their existing health information systems. The outpatient exam use case has spawned a plethora of productized solutions, and innovative health systems have internal teams experimenting with the latest large language models.
A key to a successful AI Scribe rollout is to start with a realistic assessment of your organization’s desire to pursue the development of a novel solution, piloting and scaling an emerging product, or using the features of existing technology partners as they are released. This upfront clarity will provide an initial filter and avoid spending time and resources exploring and implementing solutions that are incompatible with your organization’s strategy and capabilities.
Identifying the best use cases for AI Scribes
AI Scribes can be applied in various primary, outpatient, and acute care settings to reduce time spent on documentation and improve the patient-provider experience. HCOs must assess which unit, provider, specialty, and care pathway would benefit most from AI scribe capabilities to ensure adoption and scalability of the solution post-pilot.
When selecting use cases, consider these critical criteria:
- Operational constraints and capacity for change (particularly for clinical functions)
- Risk assessment, including identification of risks and valid mitigations
- Quantifiable value and the ability to measure ongoing benefits
Essential steps to successfully pilot and scale AI Scribes
Regardless of the innovation approach or targeted use case, HCOs will want to pilot the solution before a costly rollout across the organization. These pilots can confirm the solution’s value to the hospital and improve the ROI of scaling the solution by adapting the rollout plan based on feedback from stakeholders (clinicians, patients, and staff). Critically, these pilots can also reveal the true needs of staff and clinicians, reshaping the requirements and possibly resulting in a calculated decision to pivot away from the initially selected solution.
A pre-pilot readiness evaluation gives the pilot design team a set of hypotheses to test based on the most significant risks to a successful AI Scribe implementation. Many pilots only focus on the solution’s outputs. Envisioning all the changes needed by IT, clinical, and administrative teams empowers HCOs with a more realistic total cost of ownership and a lens through which to test the solution that goes beyond a mere feasibility study.
Defining your strategic posture for AI Scribes
Before engaging in any formal vendor selection, it is critical to conduct a candid assessment of your organization’s approach to innovation.
- What is our appetite for experimentation?
- Do we have the capacity to test and learn?
- How will we capture and apply lessons learned from proofs-of-concept?
- Are we ready to support and maintain something we build?
The answers to these questions will provide qualification criteria for the vast universe of vendor and technology options. This filter will align your selection approach to your context and constraints.
Academic medical centers may engage in research with the latest models, which will require engineering to test them in a valid and secure environment. Innovative systems may have a mature vendor piloting capability to trial new solutions from nascent technology companies. Other providers might let the market be their pilot and wait patiently for leading solutions and practices to emerge as new standard practices. These can all be valid approaches based on HCO’s resources and priorities.
Regardless of an organization’s innovation posture, HCOs will want to start with a pilot whenever they are ready to integrate AI Scribes into their patient care and use lessons from that pilot to provide insights to inform a comprehensive rollout and whether the solution and the organization are ready to scale beyond the pilot.
