Vertical AI Startup: Medical Communication AI Abridge
In this article, we introduce the initiatives of Dr. Shiv Rao, co-founder and CEO of Abridge, a startup gaining attention for transforming medical communication, who is also a practicing cardiologist. Based on a conversation with Jason Calacanis (hereinafter "Jason"), host of the popular American podcast "This Week in Startups," we will explain how Abridge utilizes generative AI to resolve major issues in the medical field—such as the administrative burden on doctors and inefficiencies in information transfer—while also playing a role in deepening patient understanding. We will examine how Abridge's approach can serve as a solution to deep-seated challenges in the healthcare sector, including physician shortages, overwork among medical staff, and the assurance of medical data security and reliability, while incorporating specific examples and quotes.
1. The Turning Point for Healthcare and AI
1-1. Changes in the Medical Industry and Rapid AI Adoption
At the beginning of the conversation, Jason expressed surprise, saying, "I didn't think the medical field would adopt technology at such a fast pace in 2023." In response, Shiv pointed out, "The medical industry has been moving really fast over the last six to eight months." Since the pandemic, the exhaustion of medical professionals has reached a peak, and traditional methods are no longer sufficient to support doctors and nurses. This is where generative AI, including ChatGPT, has emerged.
The medical field has long been said to be "slow to innovate," but the COVID-19 pandemic triggered a rapid spread of online consultations, and the practical application of generative AI has led many hospitals and clinics to accelerate AI adoption all at once. Shiv analyzes the reason for this as "the sense of crisis regarding the burden of record-keeping and paperwork, as well as the inflation of medical costs, reaching its limit among medical professionals."
1-2. Abridge Focusing on "Conversation"
According to Shiv, "Healthcare is inherently an activity based on human-to-human conversation." Conversations take place in every situation, such as interactions between patients and doctors, or with call center staff, as well as in telemedicine (online consultations), and they are filled with important information related to diagnosis and treatment plans. Abridge contributes to significant time savings for doctors and promotes patient understanding by recording and analyzing these "conversations" through technology and generating medical documents efficiently and accurately.
2. Abridge The Mechanism of AI Utilization
2-1. Automatic Creation of SOAP Notes and Multi-Channel Support
Abridge automatically generates medical documents called "SOAP notes." This is a standard organizational method used by doctors when examining patients, consisting of the following four items.
S (Subjective): Subjective symptoms and complaints of the patient
O (Objective): Objective findings and test data
A (Assessment): Evaluation of diagnosis and findings
P (Plan): Treatment plan and next steps
Abridge records the conversation between the doctor and the patient and converts it into text in real-time or within a short period. Then, generative AI organizes and summarizes it according to the SOAP note format, accurately extracting diseases and test names, and coding them (assigning medical codes such as ICD codes). In this process, it automatically structures important information while cross-referencing past medical records and the patient's medical history.
Furthermore, Abridge features "omni-channel support that can be used not only in hospital examination rooms but also in call centers and online consultations." For example, even when a doctor is receiving a phone consultation, the AI captures the conversation and generates notes on the spot. It is also actively working on multilingual support, and there are already cases where conversations mixed with multiple languages, such as English, Portuguese, and Creole, can be automatically summarized into English-based notes.
2-2. Patient Summary Function and Terminology Explanation
Abridge provides easy-to-understand summaries not only for medical professionals in the form of SOAP notes but also for patients. Explanations are automatically attached to technical terms, and it is designed so that patients can "correctly grasp their own medical condition and treatment plan." Jason also highly praised this, stating, "The mechanism of presenting difficult medical terms to patients in an easy-to-understand way narrows the distance between doctors and patients."
Shiv emphasizes that "while physician notes and patient notes serve different purposes, generating both simultaneously can ensure both the accuracy of information and the promotion of patient understanding." The goal of Abridge is to contribute to improving the overall quality of healthcare by providing the benefit of reduced administrative work for physicians and the benefit of easy-to-understand treatment explanations for patients.
3. Future Outlook and Challenges
3-1. The Potential of AI to Transcend Medical Barriers
Shiv expects that in the near future, solutions like Abridge will provide significant value even in "regions and countries with scarce medical resources." In areas lacking physicians and medical facilities, there are many cases where patients cannot see a specialist in the first place. However, if generative AI can be used to remotely triage symptoms or suggest appropriate tests and initial responses, more lives could potentially be saved.
Furthermore, a future vision is being drawn where Abridge incorporates vital data from wearable devices (such as heart rate, blood pressure, and blood glucose levels) and coordinates with physicians. Jason mentioned, "I have seen physicians who had patients reach a diagnosis thanks to an Apple Watch detecting atrial fibrillation," and Shiv also expressed high expectations for the further application potential of AI, such as "automatically analyzing sensor information to detect urgent symptoms and sending alerts to physicians."
3-2. Concerns Regarding Data Management and Reliability
On the other hand, as long as medical data is handled comprehensively, ensuring privacy and security is an essential challenge. If recorded conversations or automatically generated records are leaked, the risk of exposing patient personal information becomes immense. Additionally, from the perspective of medical litigation, some medical professionals express concerns about "how recordings will be treated as evidence." Shiv addresses these issues as follows:
“Abridge meets the requirements of HIPAA (the U.S. Health Insurance Portability and Accountability Act), and patient information is encrypted and stored securely. Furthermore, it is important to clearly define the scope of implementation with medical institutions and handle only the necessary data.”
To ensure reliability, Abridge actively publishes research results demonstrating the accuracy and safety of its algorithms through various papers and academic presentations. In the field at client sites (hospitals and clinics), efforts are also being made to reduce the risk of errors or the generation of "hallucinated information" by establishing a system where physicians ultimately review notes and can make corrections or additions as necessary.
The complexity of administrative work in the medical industry and the deepening shortage of physicians have reached a level that can no longer be ignored. Amidst this, the generative AI platform of Abridge is bringing innovation to the medical field by going beyond simple voice transcription to "maximize the time physicians spend on the care and patient interaction they were originally meant to perform," and further "deepening patient understanding."
What emerged through the conversation with Jason is that "conversation" is the core of healthcare, and that the evolution of generative AI holds great potential to complement and enhance human capabilities. Especially in the medical field, because it is a sensitive area involving diagnosis and treatment, accuracy, transparency, and reliability are required. Abridge is attempting to balance high-level security and practicality through its proprietary models, multilingual support, and strict adherence to regulations such as HIPAA.
Moving forward, we can expect not only further optimization of solutions for physicians, but also real-time vital data integration from devices, remote medical support, and the deployment of solutions aimed at correcting medical disparities around the world. In the digital transformation of healthcare, Abridge is truly a "next unicorn candidate" that we cannot take our eyes off of.
Related Articles