Summary of Key Points
This article dispels the misconception that AI in healthcare is merely about using technology to diagnose diseases, emphasizing that its true value lies in optimizing administrative processes within medical systems (such as registration, scheduling, reimbursement, and pre-authorization). The development of AI in healthcare is divided into three stages:
1. Single-point tools phase: These tools address specific challenges faced by individual doctors (e.g., converting speech to medical records or searching for medical information). They are easy to implement but have limited value.
2. Deep-water zone phase: These tools facilitate complex coordination among multiple stakeholders and systems (e.g., cross-hospital referrals or insurance pre-authorization). The barriers to adoption are high, but the potential benefits are significant.
3. New paradigm phase: The company that successfully integrates all these processes will become the “operating system” of healthcare AI, potentially replacing traditional Electronic Health Records (EHR) systems as the central data platform.
The Value of AI in Healthcare: Not Diagnosing Diseases, but Managing Funds and Processes
Many believe AI is used for diagnosing diseases, but the article argues that the most costly aspect of healthcare is not the actual diagnosis itself, but the administrative processes. For example, the annual total healthcare expenditure in the United States is $4.9 trillion, of which $740 billion (about 15%) goes towards administrative costs, while only $63 billion is invested in IT software. The opportunity with AI lies in automating these administrative tasks, such as processing insurance pre-authorization and scheduling appointments.
#### Three Stages of AI Development
- Single-point tools phase: These tools solve small, individual problems for doctors without requiring coordination with other stakeholders.
- Clinical evidence retrieval tools (e.g., OpenEvidence) are widely used and generate revenue through targeted advertising.
- Speech transcription tools (e.g., Abridge) free doctors from paperwork, but customers often switch them (67% of users).
- Deep-water zone phase: These tools require coordinating multiple roles and systems (doctors, nurses, insurance companies), representing a significant challenge. For instance, scheduling a specialist procedure involves verifying insurance eligibility, contacting doctors, coordinating with the operating room, and applying for pre-authorization. Success in this phase requires understanding the needs and rules of all parties involved.
- New paradigm phase: The ultimate goal is to create a seamless process from patient intake to payment, making AI the central hub of healthcare operations. Traditional EHR systems may become data repositories, with new AI systems serving as the primary user interfaces for doctors.
Three Critical Elements in the Deep-water Zone
The article uses a metaphorical “body” to describe these critical stages:
- Intake (the entrance): The first point of contact for patients, such as answering phone calls, registering information, and providing follow-up reminders. Companies like Assort Health use AI to improve patient experience.
- Workflow (the brain): The central hub that manages the entire process, including triage, referrals, and scheduling. Companies like Notable Health are leading in this area by coordinating cross-hospital referrals and automating insurance pre-authorization.
- Billing (the heart): This phase is directly related to hospital revenue, involving pre-authorization, coding treatments for reimbursement, and handling claims. Companies like Commure have entered this market because it represents a major source of income for hospitals.
The Final Battle: Who Will Become the “Operating System” of Healthcare AI?
The article identifies three key factors:
1. Interconnection of components: Different tools will gradually integrate with each other, forming a cohesive system.
2. Winners won’t be those developing new EHRs: Directly creating new EHR systems is difficult, but companies that build on existing technologies and connect all the necessary processes can replace traditional EHRs.
3. Counteraction from giants: Traditional EHR providers (e.g., Epic) are integrating AI features into their systems, making it more attractive for customers to use their services. New players must quickly establish a foothold in hospital systems before these giants can counteract them.
The Importance of the Local Healthcare System
The potential of AI in healthcare varies significantly depending on the complexity of the local healthcare system:
- United States: With its complex insurance landscape and high administrative costs, there are ample opportunities for AI.
- Australia: With universal healthcare, administrative tasks are simpler, but AI is mainly used for appointment scheduling and patient communication.
- EHR monopolies: Companies like Epic dominate the EHR market, posing a significant barrier for new entrants.
In summary, AI in healthcare is about streamlining administrative processes rather than developing sophisticated diagnostic tools. The deep-water zone represents a critical opportunity, and the company that successfully addresses these challenges will become the next generation of healthcare “operating systems.”
(Note: Data cited in the article comes from authoritative reports such as Menlo Ventures and KLAS; the views expressed do not constitute investment advice.)