While everyone is still debating whether AI can replace doctors, smarter players are already looking for the golden use cases where AI and physicians work together as the perfect team.
Healthcare AI Is Evolving from an Assistive Tool into a Core Productivity Engine
Scenario 1: Medical Imaging — “Giving Doctors a More Powerful Brain”
- AI for mammography: Beyond identifying nodules, AI can analyze the distribution patterns of calcifications, helping doctors distinguish between benign hyperplasia and early-stage breast cancer.
- AI for emergency CT: For patients with cerebral hemorrhage, AI can automatically calculate hemorrhage volume and recommend surgical options—from the moment the patient enters the scanner to the generation of the report—potentially saving 30 critical minutes in emergency treatment.
Scenario 2: Drug Development — “Cutting the New-Drug Development Cycle from 10 Years to 5”
- AI for target discovery: By analyzing massive volumes of biological data, AI can rapidly identify key cancer-related proteins. One pharmaceutical company reportedly shortened the development cycle of a new lung cancer drug by 40% using this technology.
- AI for clinical trials: By simulating drug responses across different patient populations, AI can help identify the most suitable participants for clinical trials. In one oncology drug trial, the failure rate was reportedly reduced from 35% to 12%.
Scenario 3: Hospital Management — “Turning Hospital Directors from Firefighters into Strategists”
- AI-powered bed scheduling: AI dynamically allocates ward resources based on patient conditions, surgical schedules, and staffing capacity. At one top-tier hospital, the average length of stay was shortened by two days, contributing more than RMB 10 million in additional annual revenue.
- AI-powered medical consumables management: By forecasting the use of surgical consumables, one orthopedic hospital reportedly reduced annual waste from 15% to 3%, saving more than RMB 5 million in costs.
AI is also playing an increasingly important role in rehabilitation and nursing, surgery, surgical robotics, and many other areas of healthcare.
With the continued implementation of China’s Healthy China 2030 strategy and the rollout of the New Generation Artificial Intelligence Development Plan across the healthcare sector, healthcare AI in China is moving from isolated applications toward system-level integration—and from an auxiliary tool toward a core productivity engine.
But the Road to AI Productization Is Littered with Failures
Healthcare AI has developed at remarkable speed in recent years. Yet many products have failed because of poorly chosen use cases, insufficient clinical value, or barriers to commercialization.
- BM Watson was once regarded as one of the AI-powered oncology decision-support systems closest to real clinical application. It claimed to cover more than ten types of cancer, including breast, lung, and colorectal cancer, and to provide physicians with treatment recommendations. After 2018, Watson was gradually withdrawn from a number of North American hospitals, including MD Anderson Cancer Center. In 2021, IBM sold its Watson Health assets to a private equity firm, effectively marking the failure of its commercialization efforts.
- In 2016, DeepMind partnered with the UK National Health Service (NHS) to develop an AI system that analyzed retinal scans to screen for vision-threatening diseases such as diabetic retinopathy and glaucoma. In 2019, the initiative faced scrutiny from the UK Information Commissioner’s Office amid data-privacy concerns. In 2022, the project was ultimately discontinued after failing to integrate effectively into existing clinical workflows.
- One startup launched a “10-second AI ECG analysis” device, claiming that a single-lead ECG could detect atrial fibrillation, myocardial ischemia, and more than ten other cardiovascular conditions. The product targeted primary-care clinics and health-screening centers. After launching in 2020, annual sales reached less than 10% of expectations, and the company entered liquidation in 2022.
…
At their core, these failures reflect a mismatch between technological capability and real clinical needs. Three recurring problems stand out:
- The false-demand trap: Solving pain points that barely exist—for example, consumer self-screening for skin cancer—or overengineering products, such as attempting to build an all-in-one oncology AI system.
- Poor clinical fit: Failing to integrate into actual healthcare workflows—for example, because of incompatible data interfaces—or relying on laboratory validation that does not translate into real-world clinical performance.
- A broken commercialization model: Unclear identification of who will pay—hospitals, pharmaceutical companies, or patients—or failure to demonstrate a compelling cost-benefit equation, where the time or money saved by AI is smaller than the cost of purchasing and using the product.
So, What Should We Do Instead?
The starting point for a healthcare AI product should not be:
“What technology can we build?”
It should be:
“What solution does the clinical environment actually need?”
A closed-loop process of in-depth clinical research → small-scale scenario validation → rapid iteration is essential to ensure that technology and real-world needs are closely aligned.
At the same time, regulatory pathways and business models must be considered early on. Only then can companies avoid the all-too-common outcome of having excellent technology but a failed use case.
More specifically, there are three key steps.
STEP 1 | Deconstruct the User Journey: Find the Real Pain Points
New technologies are often first adopted in obvious, widely recognized scenarios.
The real challenge, however, is identifying more specific and less visible use cases.
One of the most important tasks for R&D teams is to uncover emerging scenarios that are not immediately obvious. Once such a scenario has been identified, the next question becomes whether an existing technology—or a technology that could potentially be developed—can solve the pain points within that scenario.
This is a two-way discovery process between technology and application scenarios. It is also the process of finding the right value proposition for a technology.
Otherwise, the technology may never leave the laboratory.
This year, China Bridge collaborated with a multinational healthcare company to explore the application of AI in surgery.
From a technical perspective, the company had developed AI capable of detecting certain intraoperative abnormalities. Whereas these conditions had traditionally relied heavily on physicians’ experience and judgment, AI could now identify their precise location and quantitatively assess their severity.
But what exactly does this technology mean for hospitals and surgeons?
As we all know, surgical procedures involve numerous stakeholders, multiple micro-scenarios, long workflows, and countless pain points.
But who are the key stakeholders?
Which pain points matter most?
And how relevant are those pain points to this particular technology?
This is precisely where service design can demonstrate its value.
In the project, we worked together with the client team to return to the perspective of users—surgeons, operating-room nurses, and even procurement directors—and map out and deconstruct the entire journey of a typical surgical procedure.
Innovation always begins with deconstruction.
Only after breaking down the journey can we understand who the AI technology creates value for, and what that core value actually is.
STEP 2 | Define the Product Form: Turn the Technology into a “Surgical Copilot”
Even after identifying the right scenario and defining the technology’s core value, AI is still a long way from reaching users and the market.
It needs to take the form of a tangible product.
Which products are best suited to carry this core technology?
Once the technology is embedded, does it fundamentally transform the product?
Does it create a meaningful improvement in the user experience?
Using the same project as an example, we explored several possible directions together with the client.
- AI-Assisted Surgery: Improving Surgical Success Rates
Unexpected risks are always present during surgery, particularly when operating on vital organs such as the liver and lungs.
With AI assistance, the system can identify abnormalities in real time, alert physicians, and guide them in responding appropriately.
This can help reduce surgical risks and improve the probability of successful outcomes.
- Accelerating the Distribution of High-Quality Medical Expertise
Within China’s current healthcare system, many of the best physicians are concentrated in leading tertiary hospitals in major cities.
Meanwhile, third- and fourth-tier cities have substantial unmet healthcare demand but limited access to highly experienced specialists.
Historically, training an excellent surgeon has required many years.
Even when doctors from lower-tier cities have opportunities to train at top hospitals in major cities, it is difficult for them to accumulate sufficient clinical experience within a short period.
With AI assistance, however, a younger physician can effectively have something resembling an experienced senior doctor alongside them—providing timely reminders and guidance.
This can help extend high-quality medical expertise beyond major urban centers and make better care accessible to more patients.
- Optimizing Hospital Consumables Management
AI can support more than individual surgical procedures.
From a hospital management perspective, it can also assess the overall surgical performance of a department, analyze the use of medical consumables, and identify opportunities for optimization.
This can improve operational efficiency, reduce waste, and ultimately strengthen hospital performance.
Service design can accelerate the productization of AI technologies and help innovative technologies transition successfully into real-world applications.
STEP 3 | Design the Ecosystem Model: Turn AI into a Resource Hub
Chronic disease management—the management of chronic non-communicable diseases—is one of the key areas of healthcare today, covering highly prevalent conditions such as hypertension, diabetes, cardiovascular and cerebrovascular diseases, and chronic respiratory diseases.
Because chronic diseases typically last for long periods, require sustained intervention, and depend heavily on patient adherence, their management spans the entire cycle of:
prevention → monitoring → treatment → rehabilitation
As a result, the field involves a wide range of stakeholders, including pharmaceutical companies, digital-health platforms, medical-device companies, and insurance providers.
Although technologies such as AI, big data, and the Internet of Things have already been widely deployed, many implementations still emphasize technological sophistication over solving real problems.
As a result, both user experience and clinical value often fall short of expectations.
By studying the patient experience journey, service design can help organizations identify and bridge fragmented touchpoints.
Going further, it can map stakeholders across the value chain and, based on a company’s resources and capabilities, design the business and service models required to bring AI technologies into real-world use.
AI + Service Design: Helping Healthcare Innovation Take Root and Grow
On August 26, 2025, the Chinese government issued the Opinions of the State Council on Deepening the Implementation of the “AI+” Initiative.
The document explicitly called for efforts to:
Explore and promote high-quality health assistants accessible to all residents, advance the orderly application of artificial intelligence in assisted diagnosis and treatment, health management, medical insurance services, and other scenarios, and significantly improve the capacity and efficiency of primary healthcare services.
This means that AI is moving beyond being a piece of cutting-edge technology confined to laboratories and becoming an important part of national infrastructure—a foundational capability that every industry needs to embrace.
Healthcare is no exception.
So, how can healthcare companies and healthcare professionals participate?
For Healthcare Professionals
Get on board with AI—quickly—and become both users and creators of AI.
AI is increasingly becoming a tool embedded directly into frontline operations.
Healthcare professionals need to actively embrace it.
Throughout history, technology has repeatedly played a role in democratizing access to capabilities.
AI can combine the expertise of leading specialists with the precision of large-scale data, enabling ordinary physicians to do things that were previously beyond their reach.
Healthcare AI should ultimately become a tool that empowers every healthcare professional.
Only when healthcare professionals themselves become creators can AI unlock its enormous productivity potential.
For Healthcare Companies
While embracing AI, companies that fail to deeply understand user needs and the realities of the healthcare ecosystem can easily fall into the failure traps discussed earlier.
That is why healthcare companies developing AI products must return to the perspectives of doctors, nurses, patients, and the many other professionals involved in healthcare delivery.
They need to slow down.
Start from the user journey.
Carefully analyze specific scenarios.
Identify and distill genuine user needs.
Design the right usage contexts.
Build an integrated ecosystem.
And validate a sustainable business model.
Only then can healthcare AI move from technological potential into real-world implementation—creating tangible value for both healthcare professionals and patients.
Discover more from 桥中CBi China Bridge
Subscribe to get the latest posts sent to your email.
