Summary of Key Points
The rules for entering the market and the profit-making models for AI+medical products are becoming clearer. Experts are discussing various issues, such as who will pay for these products (health insurance, hospitals, commercial insurance companies, etc.), how hospitals should choose them, the main barriers to adoption (compliance and trust, data incentives), and how companies can overcome these challenges (integrating software and hardware, collaborating to share costs). The government has also introduced relevant policies (e.g., including 12 types of AI services in health insurance coverage and guiding principles for clinical evaluations) to promote the practical application of AI in healthcare.
Detailed Explanation
1. Payment Sources Are Not Limited to Health Insurance and Hospitals; New Investors Are Emerging
In the past, it was thought that AI in healthcare could only rely on health insurance reimbursement or hospital purchases. However, now there are more sources of funding:
- Health Insurance: In March this year, 12 types of AI-assisted diagnostic services, including imaging and pathology, were included in the secondary category of health insurance coverage (allowing for partial reimbursement). However, for AI products to be covered by health insurance, they must demonstrate their ability to improve efficiency, ensure quality, and offer innovation.
- Hospitals: Many hospitals are willing to invest in AI solutions that can reduce costs and enhance patient care—for example, AI that helps doctors quickly analyze images, reducing labor costs and the risk of misdiagnosis. Hospitals will actively purchase such products.
- Outpatient Services: Intelligent services like online consultations with AI can be purchased by patients at their own expense.
- Potential New Players: Commercial insurance companies and pharmaceutical companies are interested in paying for compliant medical data (e.g., using AI to analyze data for insurance product development or new drug research). Industry platforms and investors may also become funding sources.
In short, there are many new ways to generate revenue from AI in healthcare beyond just relying on hospitals and health insurance.
2. Hospitals Choose AI Products Based on Their Practical Value
Hospitals have their own criteria for purchasing AI products; not all AI tools are suitable:
- Clinical Value: Mao Xinsheng from Shukun Technology emphasizes that AI products must meet three requirements: efficiency, quality control, and innovation. For instance, an AI system that can analyze CT scans faster than a doctor, reduces misdiagnosis rates, and identifies lesions that doctors might miss is considered valuable.
- Maturity: Le Ying from Peking University Health Science Center categorizes AI products into two types: those with proven effectiveness through data (already used in multiple hospitals with good results) can be purchased directly; those without clinical validation require joint development with the company.
- Payment Models Have Changed: Previously, hospitals would buy AI software or hardware once and for all. Now, payment is based on the value provided—e.g., the duration of product use or the number of patients treated. This encourages companies to update their products regularly, and hospitals can see the actual benefits before making further investments.
3. Barriers to Adoption: Lack of Compliance and Trust, Insufficient Data Incentives
Two key issues remain before AI in healthcare can be widely adopted:
- Compliance and Trust: There are concerns about data security (e.g., the risk of patient records and imaging data being leaked) and the accuracy of AI diagnosis results. These trust issues have not yet been fully addressed, so hospitals and patients are hesitant to use AI.
- Data Incentives: AI requires large amounts of medical data for training, but current data providers (hospitals, patients) do not have clear benefits from sharing their data, preventing the optimization of AI models.
4. How Companies Can Overcome These Challenges?
Companies are finding innovative solutions:
- Integration of Software and Hardware: Sun Xiaoyu from Medtronic notes that selling AI software separately to hospitals is challenging (as hospitals don’t know how to charge for it). By integrating AI with medical devices (e.g., a pacemaker with remote follow-up functionality), companies can create a viable revenue model.
- Cost Sharing through Collaboration: Medtronic is willing to purchase AI algorithms from domestic companies or co-establish digital innovation centers to share research and development costs.
- Expanding Outpatient Services: For example, using AI for postoperative patient follow-ups can generate both revenue and valuable data.
Conclusion
The path forward for AI in healthcare is becoming clearer, with support from policies and a variety of funding sources. Hospital purchasing standards are becoming more defined, but challenges in compliance and data incentives still need to be addressed. To succeed, companies must integrate AI with hardware, collaborate to share costs, and explore new revenue streams. For the general public, AI-based healthcare services (such as assisted diagnostics and remote follow-ups) will likely become more common in the future, with some even covered by health insurance, making them highly practical.