How an AI-powered clinical notes API could boost telehealth

Healthcare providers invest significant hours each week in composing clinical notes to document patient visits.

Kwindla Hultman Kramer, co-founder and CEO at Daily, the developer of WebRTC, an open-source tool for developers to work with video and audio, has collaborated with healthcare professionals and telehealth application engineers to develop a technological solution aimed at reducing documentation time by 80% or more.

Daily's new APIs facilitate the seamless integration of SOAP and other clinical notes workflows into virtual care platforms. Daily's WebRTC developer platform offers a suite of HIPAA-compliant APIs and supports Safari and iOS for one-click, no-download telehealth video calls.

We had the opportunity to interview Kramer to discuss the role of AI in telehealth, documentation during virtual care visits, and the new documentation API.

Q. Why do you believe telemedicine is a fertile ground for the application of artificial intelligence?

A. From a technological perspective, telehealth interactions inherently capture audio digitally, making it ready for transcription and summarization. This makes telemedicine an ideal starting point for integrating new AI tools into healthcare workflows.

The category of AI tools receiving substantial attention and adoption currently are known as large language models (LLMs). For example, ChatGPT is a user interface built on top of OpenAI's large language models, GPT-3.5 and GPT-4.

The latest state-of-the-art LLMs are proficient at taking unstructured text as input and generating structured data as output. This capability is a significant advancement in computing and has generated considerable excitement among technologists.

A transcript of a telehealth visit is a prime example of "unstructured data" containing valuable information. Until recently, making sense of a telemedicine transcript required manual reading and data extraction for input into a medical records system. Large language models can now automatically create clinical documentation in specific formats.

The rapid growth of telemedicine during the COVID-19 pandemic accelerated regulatory, billing, and technological changes that were already underway. As telemedicine becomes a staple of healthcare delivery, there is an opportunity to thoughtfully incorporate AI tools to enhance access to care, save clinicians time, and improve patient outcomes.

Q. What can AI do for caregivers' SOAP and other clinical notes that caregivers cannot achieve themselves?

A. In general, the goal is not to find tasks for AI that caregivers cannot perform but to identify areas where AI can save time and enable caregivers to focus on patient care. Healthcare professionals typically do not choose their careers to spend time writing clinical documentation. SOAP notes and similar tasks are crucial but not particularly creative and do not fully leverage a caregiver's expertise and human touch, especially in comparison to patient interactions.

An AI tool that can automatically generate a first draft of a SOAP note for caregiver approval or editing has the potential to save a significant amount of time, potentially more than ten hours per week. The aim is to make the AI-generated note as similar as possible to what the caregiver would produce themselves.

While there may be room for AI to do things differently or better than humans in the long term, the immediate focus is on tasks that consume a significant amount of time with minimal loss of quality. This allows caregivers to spend more time with patients or with their families, which is a significant benefit.

Q. Your company has recently released the AI-Powered Clinical Notes API for Telehealth. Who is the API targeted at, and how can it be integrated with virtual care technology? What outcomes is the API designed to achieve?

A. Our customers are software developers who create telehealth applications, specializing in video and audio components. Our APIs are designed to be a component of a complete product or service.

The goal of everything we do is to enable the development of software that adds value to the world. While we work across various industries, healthcare is particularly important because of the potential to improve access to quality medical care.

The new AI-powered Clinical Notes APIs were developed in response to feedback from our customers. Healthcare providers typically spend over 10 hours per week on clinical documentation, prompting our customers to ask for ways to reduce this burden.

We worked closely with several customers to develop and refine this new offering, ensuring it meets high standards in data privacy, output quality, and service reliability.

Q. How do you envision artificial intelligence being applied in telemedicine in the future?

A. In the near term, we anticipate AI playing a role in simplifying preparation for visits for both patients and caregivers, conducting detailed analyses of care outcomes, and offering real-time "copilots" that provide access to information during sessions.

In the longer term, there's potential in "multi-modal" AI models, which are trained on vast datasets containing text, images, audio, and data from various sensors. These models could use the data collected by our digital devices for passive and "always on" health monitoring and diagnostics.

While privacy concerns exist, healthcare data is typically well-regulated and protected. This expansion of telemedicine could encompass a broader range of health monitoring and diagnostics, benefiting both patients and caregivers.