Ant Group CEO Han Xinyi Outlines AI and Healthcare Vision: Ant Afu Tops 150 Million Users, Deepening Doctor-Patient Integration

Deep News
34 mins ago

Ant Group CEO Han Xinyi recently delivered a speech at the 2026 Ant Afu Healthcare Ecosystem Conference held in Guangzhou, outlining four key reflections on the intersection of AI and healthcare. He emphasized that every advancement in medical technology responds to three fundamental expectations: patients want continuous, compassionate care; doctors want a reliable, handy assistant; and hospitals want to handle growing health demands with limited resources.

Han stated that Ant Afu's founding mission has always been aligned with this direction—starting from real clinical needs and refining tools step by step, hoping Afu can help doctors practice medicine better, support young doctors in growing more effectively, and assist hospitals in organizing medical services more efficiently. He described four areas of thinking around this mission.

Extending professional care beyond the clinic

The first reflection focuses on how to extend professional care beyond the examination room and provide patients with continuous health support. Han noted that a single outpatient visit or surgery is not the end of the treatment journey. Whether patients take medication on time, how they recover after surgery, and whether abnormalities require a return to the hospital—these out-of-hospital conditions directly determine whether in-hospital plans can succeed. AI can serve as a reliable link connecting doctors and patients, sustaining care beyond the clinic walls.

He cited Dr. Huang from the Anorectal Department of Sichuan Provincial People's Hospital, who performs approximately 600 surgeries annually. In the past, post-operative follow-ups relied on a notification form and a few reminders, making it difficult for doctors to track patients after they went home. Now, with Afu, key post-operative milestones are set in advance, and the system proactively inquires about recovery conditions. Afu has already helped Dr. Huang complete hundreds of continuous follow-ups. Dr. Che from Shanghai First Maternity and Infant Hospital also uses Afu for proactive follow-ups, promptly detecting risks such as patients stopping medication on their own or experiencing abnormal post-operative bleeding, enabling immediate evaluation and guidance, and directing patients back to in-person visits when necessary.

Han identified three core elements for extending professional care beyond the hospital: first, connecting information so that out-of-hospital recovery data can safely and compliantly return to doctors; second, reshaping processes by turning follow-ups into preset nodes in the treatment chain, forming standardized actions that can be proactively initiated and graded in response; third, proactively capturing risks so that when abnormal signals appear, they are immediately pushed to doctors for timely intervention. He acknowledged that these explorations are just beginning and that Afu is only a fulcrum—the real leverage comes from process reengineering and multi-party collaboration within hospitals. He expressed hope to work with hospitals to connect information pathways and streamline processes so that continuous care becomes a service capability that hospitals can reliably provide.

Making AI a trusted assistant for clinical judgment

The second reflection addresses how to make AI a truly valuable assistant for doctors' clinical judgment when dealing with medical complexity. Han observed that diseases are dynamic: how symptoms evolve, how treatments respond, and how medication and lifestyle habits interact all affect final outcomes. The value of AI in clinical settings should not be making decisions for doctors, but ensuring that every decision doctors make is built on more complete information and more solid evidence.

He outlined two expectations. First, AI should help doctors connect disease course information so they can see the "full course" rather than "a single slice." Second, AI should provide a trustworthy evidence-based knowledge foundation. By integrating authoritative clinical resources such as UpToDate and medical databases, AI functions as a portable, searchable, traceable knowledge base that quickly provides evidence-based references, labels sources of evidence, and explains its own limitations, making every assistive output rigorous and reliable.

However, Han stressed that no matter how models iterate, AI remains only an auxiliary tool, and clinical decisions must be left to professional physicians. The stronger the model's capabilities, the more prominent the value of human expertise. Doctors should become verifiers of trustworthy information, explainers of complex trade-offs, and facilitators of shared decision-making. At the same time, AI recommendations must withstand scrutiny—doctors should be able to understand, verify, and modify them. He also noted another side of human-machine collaboration: not only does AI assist doctors, but doctors' clinical experience also shapes AI, making it more professional and reliable. He cited Dr. Chi's team from the Urology Department of Shanghai Renji Hospital, who deeply participated in Afu's specialty co-creation, deconstructing cases and calibrating the knowledge base to instill clinical thinking—such as differential diagnosis and trade-off considerations—into the model. Han expressed hope for more such clinical co-creation, allowing technological capabilities and clinical experience to nourish each other.

Turning daily clinical practice into new medical knowledge

The third reflection centers on distilling everyday clinical practice into new medical insights that benefit more patients through real-world experience. Han noted that medical progress relies on countless clinical observations, summaries, and inquiries. Which treatment plans suit which populations, how long efficacy cycles last, and what are the high-risk factors for recurrence—many valuable research questions are hidden in day-to-day clinical practice and follow-ups.

He expressed the hope that every clinical encounter not only cures a patient but also becomes a building block for understanding disease. First, AI can automatically transform unstructured information from outpatient records and follow-up conversations into structured research materials, helping doctors capture easily overlooked micro-variables and discover new clinical patterns from multi-dimensional data. Second, AI can promote real-world research that naturally occurs during daily care. In the past, the hardest part of real-world research was "finding people and keeping track of them"—costs were high, and participants easily dropped out. When follow-ups become part of daily care, continuous disease course data naturally accumulates into high-quality research cohorts. AI can automatically screen and match research subjects, connecting scattered treatment records, medication changes, and recovery progress into complete datasets, and highlighting patterns worth deeper investigation.

He cited Dr. Zhang from Ningbo University Affiliated Women and Children's Hospital, who used Afu to build over ten follow-up plans for different diseases and post-operative recovery scenarios. Within just two months, these covered multiple patients, and plans are already underway to conduct real-world research on post-operative recurrence based on long-term follow-up data. When one patient's treatment experience can safely and compliantly bring value to more patients, continuous care transcends the meaning of a single clinical visit. Han noted that related explorations have already begun, aiming to create a one-stop research empowerment tool for doctors through Afu, covering follow-up plan design, automatic data aggregation, cohort screening and matching, and trend analysis, helping doctors capture real-world clinical changes at any time and improve research output efficiency.

Passing on medical experience through AI

The fourth reflection focuses on how AI can help pass on medical experience and support the solid growth of young doctors. Han emphasized that the accumulation and inheritance of medical knowledge ultimately centers on people. The growth of a qualified clinician depends on theoretical study, clinical practice, and especially the mentorship of senior physicians. The ability to differentiate complex cases comes from repeated questioning, differentiation, weighing, and review. He expressed hope to transform senior physicians' experience into repeatable training scenarios, making AI a vehicle for medical inheritance.

First, AI can turn senior physicians' experience into interactive, repeatable practice scenarios. Teaching goals are set by supervising teachers, de-identified teaching cases are selected, and AI provides step-by-step guidance: first presenting chief complaints and basic examinations, guiding young doctors to propose differential diagnoses; then asking what medical history needs to be supplemented and which high-risk factors cannot be overlooked; then layering in changes in condition and re-evaluating plans and explaining adjustment logic. Based on years of validated training programs, Han hopes AI can ask good questions and guide young doctors to develop rigorous clinical thinking rather than simply finding an answer from AI.

In traditional apprenticeship models, a renowned physician's energy could only mentor a few students. But when experience becomes standardized training scenarios, one expert's approach can reach hundreds or thousands of young doctors—this is the scalable value technology brings to medical inheritance. Second, AI can make every practice session form a feedback-driven, measurable growth loop. After a simulation ends, AI reviews the complete reasoning chain, pointing out leaps and omissions in thinking. Young doctors' weak points are clearly identified, and training content and difficulty are dynamically adjusted. In the past, "how well one practiced" mainly depended on supervising teachers' observations; now, every simulation provides immediate, specific feedback, making growth paths clear and trackable, with training intensity able to be further increased.

However, Han firmly stated that simulation training can never replace real ward experience, and that teaching content and feedback standards must be reviewed by clinical experts. Today's "good doctors" all grew from young doctors 20 or 30 years ago; today's young doctors will be the backbone of "good doctors" in ten years. Han expressed hope to accompany them through every step of study, residency training, and practice, helping them master ways to collaborate with AI, maintain independent critical thinking while using AI tools, and let AI become an accelerator for young doctors' growth and an amplifier of professional capabilities.

Han concluded by noting that as the economy and society develop, public health demands will continue to grow, and in the short term, the supply of quality medical care cannot keep up with changes in demand. All of Afu's explorations point to the same vision: caring for today's patients, supporting today's doctors, and helping today's young doctors grow faster into tomorrow's renowned physicians. He announced that Ant Afu's doctor-side platform has launched two major AI assistants—"Medical" and "AI Avatar"—supporting doctors in completing patient management and research retrieval in a one-stop manner through AI, and has connected with the Ant Afu app, which has over 150 million users, achieving deep integration between doctor and patient ends.

Han candidly acknowledged that Afu still has many shortcomings and needs repeated testing and continuous iteration in real clinical scenarios. He sincerely invited doctors to participate in Afu's design and effect verification, offer valuable suggestions, and jointly refine the tool and improve the service system. He expressed hope to work with peers to make doctors practice more comfortably, young doctors grow more solidly, and hospitals organize services more efficiently, so that every technological innovation translates into real value that patients can perceive, doctors can trust, and hospitals can sustain.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10