How can AI solve the challenges facing elderly care services? The Taikang Senior Care and Medical Large Model 1.0 has already provided the answer.
At present, China's elderly care industry is undergoing a quiet and profound transformation.
According to the latest data released by the National Bureau of Statistics of China, by the end of 2025, the number of people aged 60 and above in China has exceeded 320 million, accounting for 23% of the total population. In accordance with the internationally accepted standard (the proportion of the population aged 60 and above exceeds 10% of the total population), China has officially entered an aging society and is in the accelerating phase of the aging process.
Right in this critical process, the 300 million silver-haired population is also undergoing profound changes in their awareness of elderly care — people are generally no longer satisfied with the simple standard of "having enough food and warm clothing", but are upgrading to a longevous life with dignity, social connection and continuous care.
However, the objective reality before us is that China's elderly care service system currently has a prominent imbalance in supply structure. Take the data barrier between medical care and elderly care as an example: due to the lack of interconnection interfaces, the health data of many elderly people is fragmented in multiple "information silos" such as hospitals, communities, elderly care institutions and home self-test devices, and can never be aggregated into a continuous, complete and traceable full-cycle health record. The value of such data has not been fully unleashed, making it extremely difficult to build a smart elderly care ecosystem that integrates "prevention, treatment, rehabilitation and long-term care".
Facing this structural dilemma, the industry is pinning its hopes on AI to break the deadlock.
On August 22, 2026, at the milestone moment of its 30th anniversary, Taikang Insurance Group officially released the 1.0 version of its self-developed vertical large model for elderly care and medical care. It is reported that this is a large model in the vertical field of elderly care and medical care built on Taikang's more than ten years of operational data from physical medical and elderly care entities and massive real service scenarios, marking that Taikang has taken another key step in the integrated innovation of "insurance + medical & elderly care + AI".
So how exactly does it operate? What kind of impact will it bring to the elderly care industry? Is the underlying logic of "AI + elderly care" becoming clear and implementable?
1. When elderly care enters the "active mode", why is AI an inevitable option?
The 14th Five-Year Health Work Report mentions that the average life expectancy of Chinese residents has reached 79 years at present, which is a sharp increase of nearly 4 years compared with 2012, and is gradually approaching the top level of high-income countries around the world.
However, longevity does not mean enjoying old age in good health — at present, the healthy life expectancy in China is only 68.7 years, and there is a "health deficit" of nearly 10 years between the two indicators. In other words, while many silver-haired elderly people have a sufficient "length" of life, their life "quality" is greatly reduced, and most of their later years are spent in a state of chronic diseases, disability or semi-disability.
So where exactly is the problem?
First, the long-standing inertia of the medical system that "prioritizes treatment over prevention" is difficult to reverse. At this stage, China's medical resources are still highly concentrated in acute treatment and high-end technology R&D, while health management services for the middle-aged and elderly such as early screening, nutrition intervention and exercise guidance have not yet formed a complete system. This leads many elderly people to miss the "window period" for health intervention, so that the burden of chronic diseases breaks out intensively and is further amplified in their later years.
Second, there is a clear gap in the support service system for healthy aging. Specifically, the elderly care service chain from home to community and then to medical institutions has not formed an effective connection, and the health needs of most elderly people are difficult to respond to in a timely manner. The services have long been in a fragmented and separated state, making it hard to achieve full-cycle coverage.
Third, the general public has insufficient awareness of personal health and active management capabilities. At present, many elderly people still hold the traditional concept of "treating serious diseases and enduring minor diseases", and do not pay enough attention to regular physical examinations, self-monitoring of chronic diseases and scientific medication. In addition, some families have low participation in elderly health management, which makes it easy to miss the best time for early intervention, further aggravating the "health deficit" in later life.
It can be said that the current silver-haired group is still in the stage of "passive elderly care", but the future trend of the industry is to move towards "active enjoyment of old age".
In the process of this model transformation, AI is making up for the gaps that manpower and systems are difficult to cover from the underlying logic level.
On the one hand, it breaks the traditional passive post-remedy mode of elderly care, moves the health management threshold forward, and alleviates the structural pressure brought by the medical system's "prioritizing treatment over prevention" through risk screening, chronic disease warning and personalized health care guidance. On the other hand, it gradually breaks down the information barriers between homes, communities and medical institutions, connects the scattered and separated elderly health services, and thus builds a closed-loop smart elderly care system covering all scenarios and the full life cycle.
In fact, this is exactly the practical logic behind the "vertical domain + physical entity" path chosen by Taikang Elderly Care and Medical Large Model 1.0 — it is not an isolated technical product, but a core driving force deeply embedded in the real scenarios and business processes of elderly care and medical services.
In this regard, Chang Cheng, Deputy General Manager of Taikang Insurance Group Technology Center and CTO of Taikang Community / Taikang Medical, clearly stated at this launch event, "Different from pure technical products on the market, Taikang Elderly Care and Medical Large Model 1.0 is built on the real service scenarios of five medical centers, 32 communities and 23,000 residents."
This is Taikang's confidence, and of course it is also the "touchstone" for AI to serve the elderly care industry, because it determines whether the large model is "guessing answers" or "making judgments". After all, only when AI is deeply embedded in the entire business process of diagnosis, treatment, nursing, rehabilitation and health management, can it truly realize the service closed loop from "perception and early warning" to "intervention and implementation" and then to "effect evaluation".
Therefore, as the industry moves into the era of "active elderly care", AI is the core engine that drives elderly care services to shift from passive response to active prediction, and Taikang Elderly Care and Medical Large Model 1.0 has taken the first step of industry practice in advance.
2. For AI + elderly care, what kind of solution has Taikang launched?
In fact, when AI tries to build a closed loop of medical and elderly care, there are no shortage of difficulties.
The first and foremost challenge is the data island: the information systems of medical institutions, elderly care institutions and community families operate independently, and health records cannot be circulated, making it difficult to make accurate intervention decisions. Then there is the trust deficit: medical scenarios have zero tolerance for "hallucinations", but the "smart answers" generated by general large models often cannot be converted into "reliable execution" in many cases. The last problem is scenario fragmentation: elderly care involves multiple links such as medical treatment, care, nursing, rehabilitation and insurance, but there is no systematic connection between scenarios, so a complete closed loop from assessment to intervention is lacking.
As a practitioner deeply engaged in the medical and elderly care industry for many years, the 1.0 version of the elderly care and medical large model launched by Taikang is a precise solution targeted at the above pain points. Specifically, this can be broken down into three key points:
The first point is to shift from "data recording" to "data-driven services", forming a moat precipitated by physical operation.
In 2024, Taikang officially started the construction of the elderly care and medical large model, which mainly relies on the real service scenarios of five medical centers under Taikang Medical, Taikang Community elderly care communities and 23,000 residents, and transforms the unique long-cycle data resources accumulated over years into core production factors that drive service upgrading.
At the implementation level, on the basis of the mainstream large model base, Taikang injects its unique elderly care and medical knowledge and business data, and superimposes self-developed innovations: it adopts 12,000 high-quality long thinking chain data for elderly care and medical care, combines the clinical knowledge of multidisciplinary experts, creates the unique "expert feedback long-range thinking chain learning for elderly care and medical care" method, and completes full-parameter fine-tuning. The model supports long-sequence understanding of health records spanning 10 years, and can cover the complete business link from enrollment, assessment, plan generation to intervention.
In this regard, Chang Cheng said, "In the past, a large amount of medical and elderly care data was mainly used for collection and recording. With the development of AI technology, these data can begin to reversely drive personalized elderly care services, realizing the fundamental transformation from 'people looking for services' to 'services looking for people'."
The second point is to suppress hallucinations at three levels to solve the "trust problem" of AI-enabled elderly care.
As we all know, medical scenarios have almost zero tolerance for "hallucinations". To this end, Taikang Elderly Care and Medical Large Model 1.0 has built a three-layer hallucination suppression system: multiple verifications are carried out at the model layer to suppress hallucinations from the source; the assessment layer conducts post-verification through rule mining and counterfactual testing; finally, the application layer realizes evidence anchoring and traceability of conclusions, so that each output can be traced back to the original data basis.
For this technical consideration, Chang Cheng also gave a detailed explanation at the launch event, "Relying on vertical real business data and discipline knowledge alignment, internalizing medical specifications, care guidelines and business rules into the model, restricting the output boundary from the base, greatly reducing the risk of wrong medical outputs, and better meeting the compliance and safety bottom line of medical business, is a better technical path to support the large-scale implementation of agents in Taikang's elderly care and medical scenarios."
The third point is to break down service scenario barriers and realize the medical and elderly care closed loop between communities and homes.
It is reported that the 1.0 version of the elderly care and medical large model mainly provides AI-native health management capabilities for the two core business scenarios of elderly care communities and home-based elderly care, and realizes an AI-driven collaborative closed loop between the two.
Specifically, in Taikang Community elderly care communities, the 1.0 version of the elderly care and medical large model can act as a health management assistant and digital assistant for medical staff, assisting medical staff in chronic disease management, functional rehabilitation and full-process nursing for the elderly, automatically generating health portraits of the elderly, and carrying out intelligent grouping to form personalized management plans.
This means that after the large model goes online, every elderly person will have a "smart family doctor" with multi-agent collaboration. For example, through real-time terminal monitoring, when the system detects that the elderly's sleep quality has declined for three consecutive days, it can automatically judge whether it is related to blood pressure fluctuations, generate intervention suggestions and push them to the nursing staff, and adjust the sports and catering arrangements for the next day at the same time.
In addition, in the home-based elderly care scenario, relying on the 1.0 version of the elderly care and medical large model, Taikang can extend high-quality medical and elderly care services from Taikang Community elderly care communities to more families. It is reported that through smart wearable devices, home health monitoring terminals and remote consultation systems, the large model can provide 7×24-hour health protection for the elderly at home.
On the one hand, the system can analyze health data such as blood pressure, blood sugar and sleep in real time, actively identify abnormal trends and push early warnings. On the other hand, the system can also link online nursing staff and family doctors to customize personalized care plans for the elderly at home, so that high-quality medical and elderly care services can truly reach more families.
It can be said that Taikang Elderly Care and Medical Large Model 1.0 is built on the data high ground precipitated by ten years of physical operation, and has completed the service closed loop from "professional community care" to "active home care", allowing AI to run through all elderly care scenarios.
In this regard, Chang Cheng said, "The 1.0 version of the elderly care and medical large model can combine the health data of the elderly to assist family doctors in generating personalized health management plans. At the same time, based on smart hardware and diagnosis and treatment data, the risk warning system can actively detect abnormal states and issue warnings, upgrading the traditional passive mode of 'elderly people actively calling for help' to a closed-loop active health management mode of 'the system actively detects abnormalities'."
3. When AI is applied to elderly care, the real competition never lies in technology
Just as AI has entered every industry in the past few years, it always provides efficiency tools, data insights and auxiliary decision-making capabilities, but it cannot truly replace human professional judgment, emotional connection and value choice.
This is especially true for the medical and elderly care industry.
Elderly care is not a standardized service, but a highly personalized life course full of uncertainties. AI can remind people to take medicine, monitor heart rate and call for emergency help, and can build a complete closed loop from early warning to response and from data to intervention. However, the tossing and turning of a disabled elderly person late at night cannot be fully interpreted by sensors; the silence of a lonely elderly person facing a smart speaker cannot be truly understood by algorithms either.
Therefore, the real value of AI in the elderly care field is not to "replace" people, but to "amplify" people's value.
Liu Tingjun, President and Chief Executive Officer of Taikang Insurance Group, also mentioned this point in his speech at the launch event. He said, "If the physical layout of elderly care, medical treatment, rehabilitation and health management is the 'confidence' of Taikang to serve the longevity era, then AI will be the 'engine' that makes this service more efficient and more caring."
In fact, Taikang has been practicing this concept in the construction of its medical and elderly care service system. Over the past 30 years, Taikang has gradually developed from a life insurance company to a physical service system covering elderly care, medical treatment, rehabilitation and health management. At present, Taikang Community has laid out 48 projects in 37 cities across the country, operated 32 communities, and the number of residents living in the communities has exceeded 23,000. In addition, among the five medical centers under Taikang Medical, Taikang Xianlin Drum Tower Hospital and Taikang Tongji (Wuhan) Hospital have successively been rated as top-level tertiary hospitals.
In this huge service carrier, Taikang Elderly Care and Medical Large Model 1.0 actually connects the medical and elderly clinical experience, community service practice and medical health data accumulated over years, relies on large model technology to intelligently upgrade the offline physical service capabilities, and realizes the deep integration of AI and medical and elderly care services. It is an important implementation of Taikang's new life insurance strategy at the technological dimension to build a "seamless full-life-cycle service system integrating elderly care, medical treatment, rehabilitation and health management".
Of course, this is just the beginning. Liu Tingjun also emphasized in his speech, "The large model is not the end point, but a starting point."
Looking to the future, when the entire medical and elderly care industry accelerates its transformation from "passive elderly care" to "active enjoyment of old age", Taikang Elderly Care and Medical Large Model 1.0 can open up the key blockages of medical and elderly care services. The greater imagination space lies in: when long-cycle health data is deeply coupled with AI's predictive capabilities, elderly care will no longer be a fragmented crisis response, but a well-prepared long journey of improving life quality.
For this proposition of the era, Taikang's answer has been very clear: technology for good, medical and elderly care for reality. Only by letting AI