"The operation was successful, and I can go home from the hospital." When they first heard this sentence, most patients and their families would breathe a sigh of relief, thinking that this serious illness would be over. But only those who have experienced it personally understand: being discharged from the hospital is the beginning of a new round of anxiety.
How to care for wounds? Do I need to adjust my medicine after taking it? When will the reexamination take place? I suddenly feel chest tightness and dizziness. Do I need to go to the hospital? During the hospitalization, doctors and nurses were watching over me, which made me feel at ease; but after I was discharged from the hospital, this "safety rope" to protect my health was gone.
A broken "safety rope": After discharge, who will take over the risk?
At the National Hygiene and Health Conference held recently, it was clearly stated that it is necessary to "strengthen the comprehensive prevention and control of chronic diseases and improve the full chain of prevention, treatment and health management services" so that the people can get medical treatment nearby and get good treatment. The four words "prevent and treat health problems" mean prevention is in the front and management is in the back. The "health" in the middle is exactly the link that was most easily ignored in the past.
On October 22, at iFlytek Global 1024 Developer Festival, iFlytek Medical released AI Patient Management Platform 3.0 and announced the launch of a special AI patient management expert consensus plan. At first glance, this sounds like industry news; but its distance from ordinary people may be much closer than imagined.
Let’s look at two real cases: Mr. Fu (pseudonym) from Sichuan, 69 years old, was discharged from Sichuan Provincial People’s Hospital after treatment for coronary heart disease. Upon discharge, he joined the AI patient management service. Based on his medical history and treatment plan, the platform automatically generated a personalized rehabilitation plan and set up a risk warning plan based on the key points of out-of-hospital management of coronary heart disease. AI "stands guard" for his out-of-hospital rehabilitation throughout the process.
On the 106th day after discharge, family members reported through the platform: Mr. Fu's chest tightness has become more frequent than before recently and his whole body has no strength. After receiving the information, the platform quickly triggered an early warning. The health manager immediately checked Mr. Fu's symptoms, medication and recovery records, and combined with his medical history, determined that the patient was at risk of severe myocardial ischemia. So I called him immediately and suggested returning to the hospital immediately. At the same time, relevant information was also sent to Mr. Fu’s attending doctor. After communicating with the family, the doctor finally decided to hospitalize him for treatment. The admission examination found that Mr. Fu's coronary stenosis worsened, and he subsequently underwent coronary stent implantation. He recovered well after the operation and was discharged successfully.
From the message sent by the family member on the mobile phone, to the triggering of the early warning, the verification by the health manager, the intervention of the doctor, and then Mr. Fu was admitted to the hospital for examination and surgery, there was no "wait and see" in this link, and there was no "check again next time".
A similar experience also happened to 55-year-old Ms. Luo (pseudonym) in Anhui. On the 17th day after discharge from the hospital after cerebral infarction, family members reported through the platform that the patient was gibbering and slurred. AI combines the past medical history to help indicate the high risk of stroke recurrence; the platform immediately starts the intervention process and synchronizes with the attending doctor to recommend return to the hospital for treatment. Subsequently, the patient underwent relevant examinations at the local emergency department and was diagnosed with recurrent cerebral infarction. He was later transferred to the First Affiliated Hospital of the University of Science and Technology of China (Anhui Provincial Hospital) for further treatment. From early warning triggering, on-site examination to referral to hospital, all links are closely connected. For stroke patients, time is life.
What these two cases have in common is that after family members report patient abnormalities, the "AI system" proactively identifies, quickly verifies, generates early warnings and promotes execution. In other words, the "safety rope" after discharge is not left to the patient to guess outside the hospital, but is caught together by "AI early warning + manual verification + doctor decision-making".
A replicable "new path": What shortcomings does AI Patient Management 3.0 make up for?
"Continuing medical services" have been included in the reform list at the national level, and the reform of DRG/DIP medical insurance payment methods is also continuing to advance. Whether patients are managed well after discharge is directly related to the quality of medical care, the security of medical insurance funds and the vital interests of patients. But the reality is: the base of discharged patients is huge, medical staff are overloaded with manual management, telephone follow-up is difficult to cover, and long-term patient compliance is difficult to ensure. At the same time, unified and standardized implementation standards for patient management have not yet been formed across the country.
The core logic of the AI patient management platform 3.0 launched by iFlytek this time is one sentence: changing out-of-hospital management from "lucky" to "standardized".
The platform has accumulated more than 160 standardized disease management paths, covering more than 30 mainstream clinical departments. AI no longer just sends a "please review" text message regularly, but dynamically adjusts the plan based on the patient's recovery progress, identifies risk signals in real time, and triggers hierarchical intervention. Behind the scenes, there is a multi-teacher co-management model of doctors, nurses, nutritionists, and rehabilitation specialists. In other words, out-of-hospital management no longer relies solely on patients’ own records and family members’ guesses. Instead, there are paths to follow, standards to follow, and people to manage.
Up to now, the platform has been implemented in more than 300 hospitals across the country, and has provided refined health management services for chronic diseases and high-risk groups to more than 500,000 people. Behind the platform capabilities is the Xinghuo Medical model. In the doctor-patient interaction scenario, the platform accumulates more than 2.2 million real service data every day, supporting the continuous iterative optimization of the large model; the data is circulated in a closed loop, and the out-of-hospital health management capabilities empowered by the large model become more accurate as they are used.
A set of real data from hospitals: Where does the confidence for out-of-hospital management come from?
Whether a technology is good or not, whether hospitals are willing to use it and how effective it is are very real test standards.
The First Affiliated Hospital of the University of Science and Technology of China (Anhui Provincial Hospital) and iFlytek Medical have jointly built a large-scale "intelligent doctor accompanying" model to open up the entire chain of intelligent guidance before diagnosis, full-process companionship during diagnosis, and patient management after diagnosis. Dean Tian Yanghua introduced that the hospital adopts a joint model of "medical care focusing on core diagnosis and treatment + platform professional operation support" to provide standardized basic services to all patients in the hospital and refined management to key specialized patient groups. At present, it has been launched in more than 30 clinical departments, covering more than 100 types of diseases, with more than 180,000 online consultation services, and a patient satisfaction rate of 97%. Specialized disease projects such as cerebrovascular intervention, integrated pregnancy and childbirth, and weight loss management have produced visible results.
Han Hui, deputy director of Qilu Hospital of Shandong University, introduced that the hospital and iFlytek Medical released the "Qilu Spark" full-disease management model, which relies on the human-machine collaboration model of "AI initial judgment + health care professional intervention + doctor control" to achieve hierarchical and classified management. Management has covered 20 clinical departments and 70 disease types, and has managed more than 270,000 patients for free, reducing 90% of the follow-up workload of medical care; patient satisfaction exceeded 98%, planned follow-up rate was 70%, and medication compliance was 97%.
Recently, the base model that supports all of this has been upgraded to another level. The Spark Medical Large Model V3.7, released on the same day, scored 99.6 in the Shanghai Artificial Intelligence Laboratory MedBench 5.0 agent evaluation; its performance is also leading in core tasks such as medical knowledge question and answer, clinical language understanding, diagnosis and treatment plan recommendation, medical document generation, multi-modal interaction, and multi-round evidence-based reasoning.
On the doctor's side, the reasonable rate of intelligent medical record generation has increased to 85%, and the image-assisted diagnosis capability has improved by more than 15% compared with industry vertical models. On the resident side, iFlytek Xiaoyi has moved from "single consultation" to "continuous management", and its capabilities such as comprehensive interpretation of medical reports, dynamic health file recording, personalized risk insights, and traditional Chinese medicine tongue and facial constitution identification have also been upgraded simultaneously.
A "standard line" jointly built by the industry: making AI services truly credible and accessible
AI patient management is actually not a new concept, but in the past, each region had its own approach, and services were uneven. It was difficult for patients to judge whether it was good or bad, and it was difficult for hospitals to replicate it on a large scale. The AI Patient Management Expert Consensus Special Plan launched at this forum united the Chinese Medical Association Magazine, China Cardiovascular Health Alliance, China Academy of Information and Communications Technology, and a number of top tertiary hospitals, combined with real clinical cases, to sort out service processes, technical specifications, quality control standards, and application boundaries one by one, filling the gap in industry standards.
This is the starting point for the industry to launch the expert consensus special plan. Tao Xiaodong, President of iFlytek Healthcare, said that the construction of Healthy China requires continuous strengthening of national health management, disease prevention and control and continuous medical service capabilities. Relying on independent and controllable domestic AI computing power and large medical models, iFlytek is promoting the extension of capabilities from in-hospital diagnosis and treatment to out-of-hospital patient management. The core goal is to make up for the shortcomings of medical service continuity.
The "whole chain of prevention and treatment of health care" falls on every family. In fact, it allows patients to not panic after being discharged from the hospital and going home. There is someone to respond when you feel uncomfortable, someone to remind you when it is time to review, and someone who can understand, judge accurately, and push when the indicators fluctuate.
