Facing High-Incidence Severe Diseases: Ping An's Breakthrough Battle with Medical AI

For a patient newly diagnosed with breast cancer, what lies ahead of her is not only the complexity of medical diagnosis and treatment but also the uncertainty of medical expenses.

Some experts may recommend a seemingly high - tech solution, "surgery + overseas cell therapy", which costs up to 1 million yuan and is completely outside the scope of medical insurance reimbursement. On the other hand, another option is the "standard plan" based on clinical guidelines, which involves surgery combined with radiotherapy, chemotherapy, and endocrine therapy, with a cost of about 150,000 yuan.

Facing various examination reports dozens of pages thick and filled with obscure terms, patients not only face the anxiety of "treating the illness" but also the real risk of "falling into poverty".

This is a microcosm of the medical journey of countless Chinese cancer patients. A global medical AI revolution is trying to dispel this "information decision - making blind spot".

In January 2026, the medical AI startup OpenEvidence announced the completion of its Series D financing, and its valuation soared to $12 billion. This company, which has only been established for a few years, has an AI tool that can search authoritative literature in real - time and basically achieve "zero hallucination", and more than 40% of practicing doctors in the United States are using it.

This wave of enthusiasm quickly spread to China. Internet giants have followed up, entering the field with their respective strengths, hoping to become the Chinese version of OpenEvidence. While the giants are still competing in chatting and finding literature, Ping An of China is not satisfied with being just a "useful AI Q&A tool". Relying on its long - term deep cultivation in the financial and large - health fields, the Chinese platform is striving to build an AI - MDT (Multidisciplinary Team) platform that integrates "in - depth diagnosis and treatment" and a "risk - control brain". Of course, this is a more difficult and in - depth path.

Path divergence: Dialogue, tools, and treating diseases

The core value of OpenEvidence is "cognitive enhancement". It provides highly reliable information in a more efficient way, helping doctors integrate the latest evidence - based medical evidence and make clinical decisions. Its customers are often pharmaceutical companies seeking doctors' attention. In essence, it is more like a vertical media in the AI era.

Participants in the Chinese medical AI track are also conducting opportunistic marketing. The high valuation of OpenEvidence verifies the prospects of this track. While emphasizing that they have entered the game, they then shift the topic to emphasize their own capabilities and have taken two different business paths.

The first is the "dialogue path", represented by Internet giants and large - model companies. Internet giants have advantages in developing basic large models. With a large user base and C - end traffic, they try to build a general dialogue entrance in the medical field with overwhelming investment. However, in the serious medical scenario with extremely low error tolerance, the hallucination problem of general large models remains a challenge. Moreover, language ability does not equal clinical decision - making ability. The lack of front - line clinical experience is not easy to make up quickly. Without in - depth training with their own case data, such large - health AI services are difficult to generate complete diagnosis and treatment plans that meet current clinical needs and often stop at "literature retrieval".

The second is the "vertical tool path", represented by some emerging medical industry startups. These products are most similar to OpenEvidence in form. Through detailed data collection and governance, they build their own medical knowledge graphs and have established a good reputation in the field of "evidence - based search". However, these services often only provide information and do not solve patients' medical problems. Analyzing the business models of some of these companies, we can find that their core revenue still depends on providing digital infrastructure and marketing services for pharmaceutical companies.

Ping An of China, which has been deeply involved in the medical field for many years, is taking a different path: a closed - loop of "diagnosis and treatment + payment".

As a payer of medical services, Ping An's core demands are "value - based medicine" and "risk control". This determines that when developing medical AI, Ping An cannot just be a search tool but must be able to assist in decision - making, standardize diagnosis and treatment, solve difficult and severe diseases, and further reduce medical costs.

Picking the "crown jewel" of serious medical care: Ping An AI - MDT focuses on severe diseases

For a long time, Internet medical services have faced an embarrassment: platforms can increase the service frequency by optimizing the service experience and introducing more general health consultation services to make the user activity data look good. However, for users, consultations for daily minor illnesses are often characterized by "low pain points and low memory", which are difficult to convert into in - depth recognition of the platform's professional ability.

What can truly establish deep - seated trust with users is the ability to diagnose and treat complex diseases, which is also the reason why famous doctors and top - tier tertiary hospitals are irreplaceable. Ping An, which has handled countless critical illness insurance and medical insurance claims and found green channels for countless severe patients to seek medical treatment, is now combining its medical resource advantages with AI and striving to pick the "crown jewel" of serious medical care and enter the field of oncology AI - MDT (Multidisciplinary Team).

In the deep - water area of serious oncology medical care, the complexity of treatment plans is extremely high. The experience of a single doctor is often limited. MDT is the golden key to solving complex clinical problems. However, limited by the scarcity of top - level expert resources, it has always been difficult to popularize on a large scale, which easily leads to deviations in the diagnosis and treatment path. Patients do not benefit, and it also triggers disputes over "over - treatment" and huge insurance expenditures.

In the field of serious medical care, it is difficult for any party to achieve victory with a single move. Ni Yuan, the general manager of the AI product team of Ping An Technology's medical services, introduced that Ping An AI - MDT has built a unique "three - layer capability advantage":

Authoritative evidence - based medical foundation: Benchmarked against the global leading level, Ping An has built a knowledge graph that includes authoritative clinical guidelines, journal literature, and textbooks, ensuring that each piece of advice generated by the AI can be accurately traced, achieving the rigor of "no generation without evidence".

In - depth diagnosis and treatment decision - making logic: This is the key advantage that differentiates Ping An from general large models. On the one hand, there is a data advantage. Ping An's vertical model training relies on its own massive real - world oncology cases. On the other hand, there is an expert advantage. Ping An has a network of 50,000 cooperating famous doctors, providing tens of thousands of critical illness multidisciplinary consultation services every year. The expert team collaborates deeply with Ping An. Through the RLHF (Reinforcement Learning from Human Feedback) mechanism, the diagnosis and treatment conclusions of the model are aligned with those of top - level experts. At the same time, Ping An also collects data on the "Chain - of - Thought (CoT)" of interdisciplinary experts (the clinical reasoning process, training the model to simulate the decision - making path of experts), which greatly improves the model's reasoning ability.

Commercial insurance payment and risk - control system: Ping An serves more than 200,000 oncology claim customers every year. This huge payment scenario is not only a touchstone for testing AI capabilities but also a key lever for building a commercial closed - loop. AI - MDT can not only fill the gap in medical resources but also truly build a "risk - control defense line" to standardize the diagnosis and treatment path and curb unreasonable expenses in the annual hundreds of billions of oncology treatment expenditures.

Oncology diagnosis and treatment is a touchstone for testing medical strength and also a great proof of Ping An's "hard - core medical strength". When Ping An can properly solve critical illness problems related to life and death, it can firmly establish its industry authority of "understanding medicine and being truly professional" and build a professional reputation that competitors can hardly surpass.

For doctors: Creating a top - level diagnosis and treatment assistant

In the medical ecosystem, doctors are the core nodes and the key to the larger - scale application of AI - MDT. Ping An hopes to gather the top wisdom in the industry by creating a co - pilot - level doctor tool and focus on attracting top - tier oncology specialist doctors for high - frequency use.

Facing the complex clinical data in oncology, AI first acts as an "efficiency expert". Ping An's medical AI has the ability to conduct a panoramic analysis of patients' conditions, automatically sort out messy medical record information, and enable doctors to instantly grasp the overall situation of the illness. For difficult problems, AI - MDT will activate the DeepResearch mode, conduct in - depth retrieval and reasoning of global authoritative literature, and assist doctors in formulating accurate diagnosis and treatment plans.

Ni Yuan emphasized that AI is by no means intended to replace doctors but to be their partner. The distribution of high - quality medical resources in China is uneven. Through the "human + machine" collaborative model, AI can empower primary - level doctors to have a diagnosis and treatment ability close to that of experts in tertiary hospitals. This means that whether in prosperous cities or remote rural areas, patients can enjoy standardized and high - quality medical services, truly realizing the universal availability of high - end medical resources.

For patients and insurance: Value feedback, authoritative second - opinion diagnosis, and risk - control brain

When AI - MDT aggregates the wisdom of top - level doctors, it can provide value feedback to the patient side (To C) and the insurance side (To B), truly building a closed - loop of "in - depth diagnosis and treatment + commercial insurance cost control".

In response to the confusion and anxiety of patients facing high - cost treatment plans, Ping An AI - MDT has launched an "human - machine collaborative" authoritative second - opinion diagnosis service. The AI will generate a structured "Multidisciplinary Diagnosis and Treatment Report", which includes in - depth analysis of diagnosis and treatment plans, comparison of the advantages and disadvantages of different plans, and subsequent rehabilitation suggestions. This report will finally be reviewed and confirmed online by well - known experts to ensure the authority of the plan and give patients more confidence when making decisions.

Decision - making is only the first step, and implementation is the key. Ping An's medical AI has the ability of "precise recommendation", achieving a leap from "providing a plan" to "providing a channel". Based on the determined optimal plan, the AI will recommend suitable doctors and hospitals for patients and can guide patients to enter Ping An's cooperative network hospitals. These hospitals are carefully selected and have the characteristics of "high cost - performance and high - quality service", thus truly opening up the "last mile" of patients' medical treatment.

The operation of the medical system cannot do without the support of "money". A useful and benevolent AI is not only good at treating diseases but also helps to "save money". The commercial insurance side has a natural motivation to pay for "healthier customers". In this model, AI is no longer a product that needs to be separately promoted to patients but is internalized into various infrastructures for insurance products to reduce costs and increase profits. The involvement of commercial insurance payment enables medical AI to truly complete the closed - loop from technological investment to commercial return.

The diagnosis and treatment plans and cost - assessment capabilities provided by AI - MDT do not sacrifice medical quality to reduce costs but find the optimal solution for treatment and accurately intercept unnecessary medical expenditures, thus truly achieving precise cost control for commercial insurance. It not only effectively protects the interests of patients but also builds a sustainable medical AI closed - loop.

What Ping An is trying to solve is a long - standing difficult problem in the medical industry: how to "reduce costs and improve quality at the same time". Looking to the future, under Ping An Group's dual - wheel - drive strategy of "comprehensive finance + medical care and elderly care", medical AI is not only a "connector" linking huge assets and services but also an "accelerator" for improving service efficiency and quality.

Conclusion: Medical AI needs to do difficult but right things

The competition in medical AI will ultimately return to "solving problems". Good technology enables every ordinary person to have the confidence to "get the right treatment and not waste money" when facing diseases.

Ni Yuan revealed that in 2026, Ping An AI - MDT will take this "difficult but right" perseverance to a new level of breadth and precision. In terms of breadth, it will expand the scope of AI - MDT services to more high - incidence critical illnesses. In terms of precision, it will increase the accuracy of diagnosis and treatment plan recommendations to 90% and ensure 100% traceability of evidence citation.

When the medical track is booming again, Ping An's medical AI presents a unique Chinese model: not only aiming to be the Chinese OpenEvidence but also committed to becoming a "doer" in the whole - process diagnosis and treatment of difficult and severe diseases. It uses top - level scientific research tools on one hand to "build a nest to attract phoenixes" and attract experts, and on the other hand, uses authoritative second - opinion diagnosis to "provide value feedback" to serve patients and insurance payers, taking a serious medical closed - loop path of "in - depth diagnosis and treatment + commercial insurance cost control".