Management for stroke intelligent early warning empowered by big data

Abstract

Global aging population, especially with the global pandemic outbreak of the Corona Virus Disease 2019 (COVID-19), has endangered human health security. Digital information technology through big data empowerment and intelligent application is widely considered a key element to solve the problems. Stroke is a life-threaten disorder. We studied individual health management and disease risk perception using human health assessment model and make full use of wearable wireless sensor, Internet of Things, big data, and Artificial Intelligence for potential risk monitoring and real-time stroke warning. We proposed an effective method of monitoring, early warning and rescue to improve the stroke treatment. The result shows that the health management empowered by big data can generate new opportunities and ideas to solve early detection and warning of stroke.

1. Introduction

The application of digital information technology is now widely recognized as a key factor in addressing the imbalance between healthcare supply and demand. Through big data empowerment and intelligent applications, the deep reconfiguration of health management can generate new opportunities and ideas for solving the development dilemma of the healthcare industry.

Advocating the early warning management of “moving the gate forward” and “health first” is a scientific and effective strategy to improve the comprehensive prevention and control of chronic diseases. It is also the consensus of global health model change. Traditional Chinese Medicine (TCM) is a theoretical system of disease prevention and health management, which is to take preventive and therapeutic measures to stop the development of diseases before they occur or when they are in the bud. Based on the theory of treating the disease in Chinese medicine, combined with wearable devices that continuously monitor important physiological indicators of the body, researchers can study various hidden abnormal data of physiological indicator changes. And with the help of big data analysis, they can seek the correlation with the potential disease risk of internal organs of the body and perceive the health status of individuals in a more refined way. It is very important and promising to explore the establishment of a new intelligent pathological model for stroke prediction to provide a simple, easy, safe and reliable pre-hospitalization monitoring and early warning emergency system for stroke patients [1].

Progress and bottlenecks of domestic and international digital information technology in health alert management. With technology innovation becoming increasingly popular in everyday life, consumers are embracing consumer-grade software and hardware devices to manage their health. Smart wearable devices are consumer-grade and connected electronic devices that can be worn as accessories or embedded in clothing, including smart watches, rings, and wristbands. They all have strong processing power and a multitude of sophisticated sensors that can collect health information. The research and application of artificial intelligence (AI) technology and wearable medical devices to prevent diseases and health status assessment and early warning by medical gateway forward is in the initial stage [2]. A systematic study and high-quality evidence-based medical research verification are lacking, and the big data research on the prediction of related diseases by wearable medical devices has not been formally carried out. With the help of artificial intelligence and big data technology, innovative research is needed to make up for the lack of cardiovascular and cerebrovascular disease monitoring and management system by mining the potential risks of personal physiological data. It is necessary to break through the bottleneck to build intelligent organ models to carry out systematic research of intelligent early warning management. The research results can prevent and control the risk of stroke in a better and accurate way, which is the development direction of health management of cardiovascular and cerebrovascular diseases and an important option to reduce the risk of cardiovascular and cerebrovascular diseases.

This paper is composed as follows: Section 2 describes how to collect and obtain human physiological index data. Section 3 describes big data technology and intelligent health management. Section 4 discusses various intelligent treatment technology and stroke early warning management method. Section 5 describes development and progress of stroke prevention and treatment using digital intelligence. Section 6 describes deep mining of TCM diagnosis and treatment data. Section 7 describes the intelligent application of TCM data for stroke early warning. Section 8 describes the three typical experiments and their verification results. After the discussion about application prospects in Section 9, the concluding remarks are provided in Section 10.

2. Data acquisition and application of medical and health intelligent applications

We provided the wearable physiological watch ANDUN smart watch (as shown in Fig. 1 ) to carry out the continuous collection of body index data by the smart device and form the real-time physical health monitoring big data. The data studied mainly include individual physiological index continuous monitoring data (as shown in Table 1 ), life environment monitoring data, time and location data, TCM diagnosis data, background and behavior data, etc. The individual physiological index monitoring data is mainly based on the TCM midnight-noon ebb-flow theory of organ correlation and TCM pulse-taking digital technology, and is collected in a time-series manner using wearable smart devices. Historical diagnostic data and desensitized empirical data are used directly for TCM diagnostic data. For those data, this project classified them into two categories, structured data and unstructured data. While structured data used convolutional neural network for feature extraction and unstructured data used graph convolutional neural network combined with probabilistic graphical model for feature extraction and disease risk classification. In the area of organ modeling and health risk assessment, the TCM midnight-noon ebb-flow theory was used to model the five viscera and six organs and construct a dynamic empirical model of health status, design a pathological model and use it for clinical evidence-based assessment. On that basis, the data was collected through questionnaires, health consultation and health checkups. And then the association between behavior and physical health was explored to establish an intelligent system to support the development and application of the industry, such as a classification model and a recommendation system for the recreation industry.

Fig. 1.

Wearable devices and health management visualization.

Table 1.

Data of heart rate, blood pressure and oxygen by ANDUN watch.

Time	Heart rate (bpm)	Time	Blood pressure (mmHg)	Time	Blood oxygen (%)
09/02/2022 23:56:24	79	09/02/2022 23:56:24	128/84	09/02/2022 23:56:24	98
09/02/2022 23:54:23	81	09/02/2022 23:48:23	128/89	09/02/2022 23:48:23	99
09/02/2022 23:52:23	86	09/02/2022 23:28:22	138/94	09/02/2022 23:36:23	99
09/02/2022 23:50:23	84	09/02/2022 23:20:25	131/85	09/02/2022 23:28:22	98
09/02/2022 23:48:23	81	09/02/2022 23:02:20	132/86	09/02/2022 23:20:25	98
09/02/2022 23:46:23	84	09/02/2022 22:52:20	139/93	/09/02/2022 23:12:21	96

Note:beat per minute (bpm).

Table 1. An example of human physiological data collected intensively by ANDUN health watch that the interval of data sampling can be set remotely and the minimum time interval can be 30 s.

3. Big data technology and intelligent health management

The health management of "moving the gate forward" and "health first" is a scientific and effective strategy to actively meet the needs of healthcare and improve comprehensive prevention and control of chronic diseases, which is also an inevitable trend of global health model transformation. Health management is based on modern health concepts and biological, psychological, and social adaptability, applies modern medical and management knowledge, monitors, analyzes and evaluates the health of individuals or groups, intervenes and manages disease risk factors, and provides continuous services in order to prevent and control diseases and improve life quality with minimal cost.

The United States (U.S.) health management model is mainly inclined to commercial health insurance, and about 77 million people enjoy medical services in 650 health management organizations. More than 90 million Americans have become the beneficiaries of preferred provider organization (PPO) plans, forming a 90% and 10% model of health management, i.e. 90% of individuals and businesses that manage their health have reduced their medical costs to 10% of the original costs, while 90% of individuals and businesses that do not manage their health have increased their medical costs by 90%. And more than 90 million Americans are PPO plan users, which means that 7 out of 10 Americans have accessed to health management services [3].

An information services group states recently that insurers are adopting new technologies such as AI and drones to better meet customer needs [4]. In a survey conducted by Salesforce for the average patients in the United States, the future trend of U.S. smart industry may move toward wearable health monitor devices, mobile healthcare services with an app connected to the monitor devices, and telemedicine [5,6] (Fig. 2 ).

Fig. 2.

Personal health management online platform and telemedicine.

4. Intelligent treatment technology and stroke early warning management

4.1. Correlation of biological information and heart and brain diseases in traditional Chinese medicine

The chronological correspondence of internal organs in Traditional Chinese Medicine (TCM) is related to the flow sequence of internal organs and meridians, and each of twelve meridians corresponds to twelve hours. For example, chronic cholecystitis and other gallbladder diseases tend to attack or worsen at midnight (23–1 o'clock), when the body's Qi and Blood come from the triple energizer meridian, enter the gallbladder meridian, and then flow to the liver meridian, meanwhile, Qi and Blood compete with evil Qi of the diseases, therefore the diseases develop; predawn diarrhea occurs at dawn (5–7 o'clock), which is the main time of large intestine meridian, and at this time Yang is born. If Yang is deficient in spleen and kidney, Yin energy is strong and Yang energy is too insufficient to warm intestines, and diarrhea occurs [7].

Ji Li and Tozhi Wang [8] investigated the treatment of hypertension from etiology and pathogenesis of hypertension, and infusion of internal organs by meridian flow. They concluded that Chen hour is the time of much Qi and Blood when nature Yang and body Yang are both flourishing, and the two Yang energies merge to cause dizziness. Yin hour is the time when kidney meridian is in order, which time Yang energy is getting weak and Yin energy is getting strong. When kidneys are weak at the time of crossing, there will be an imbalance between Yin and Yang and vertigo will occur.

Zihua Wang et al. [9] analyzed the peak time of various types of arrhythmias in 520 patients with coronary artery disease and their different TCM dialectical classifications. The results showed that the peak time of arrhythmia attacks in patients with coronary artery disease and different TCM dialectical classifications corresponded to midnight (23–1 o'clock).

Liang Jinfeng, Wang Lei, et al. [10], [11], [12], [13] discussed the relevance of midnight-noon ebb-flow to biological rhythms from three aspects: understanding of biological rhythms in TCM, relevance of midnight-noon ebb-flow to biological rhythms, and biological rhythms in modern science. It combines time and space factors with biological rhythms to carry out clinical applications in TCM (Fig. 3 ).

Fig. 3.

Sequence of correspondence between midnight-noon ebb-flow injection and TCM organs.

4.2. Relevance of circadian rhythms to heart and brain diseases in modern medicine

Circadian rhythms are a biological rhythm with a period of 24 h. Circadian mechanisms have been discovered in almost all living organisms and molecular mechanisms of circadian rhythm in humans and other mammals have been elucidated. Circadian rhythms are regulated by circadian clocks (also called "biological clocks"), which are composed of central biological clock (consisting of about 20,000 neurons located in the supraoptic nucleus of hypothalamus) and peripheral biological clock (which is found in almost every tissue). The central biological clock coordinates the peripheral biological clock of cells, tissues, organs, and systems through autonomic neuropeptides, endocrine, and mediators. The peripheral biological clock responds to tissue-specific synchronizers (e.g., diet and activity) in addition to the central biological clock signals.

In the clinic, patients with sleep disorders, people with jet lag, and people working in shifts are often unable to sleep at night or rest during daytime, because external light environment is not synchronized with their internal biological clock, showing circadian rhythm disorders, which is associated with an increased risk of cardiovascular risk factors such as hypertension, metabolic syndrome, and cardiovascular disease, and even with an increased risk of cardiovascular mortality, compared to people with regular rest (Fig. 4 ).

Fig. 4.

Schematic representation of the human biological clock.

5. Development and progress of stroke prevention and treatment using digital intelligent

Stroke is a life-threaten health problem included in the "Health China Action". Stroke is a general term for brain tissue necrosis caused by narrowing or occlusion of brain blood-supplying arteries (carotid and vertebral arteries). Many clinical studies suggest that changes in vital signs can provide a useful early warning.

5.1. Stroke incidence

Global Burden of Disease (GBD) 2019 shows that China stroke incidence decreased from 222/100,000 in 2005 to 201/100,000 in 2019. Among them, ischemic stroke incidence increased from 117/100,000 in 2005 to 145/100,000 in 2019, and hemorrhagic stroke incidence decreased from 93/100,000 in 2005 to 45/100,000 in 2019 (Fig. 5 ) [14].

Fig. 5.

Ischemic and hemorrhagic stroke incidence in China, 2005–2019 (Data from GBD) [14].

5.2. Stroke mortality

According to China Health Statistics Yearbook 2019, gross stroke mortality in China in 2018 was 160/100,000 in rural residents and 129/100,000 in urban residents. The sixth census shows that about 1.94 million people died of stroke in China in 2018. Stroke has become the second cause of death in rural residents (24.16% of all causes of death) and the third cause of death in urban residents (20.53% of all causes of death) (Fig. 6 ) [14].

Fig. 6.

Prevalent stroke mortality in urban and rural residents in China, 2005–2018 (Data from the China Health Statistics Yearbook 2019) [14].

6. Deep mining of TCM diagnosis and treatment data and intelligent application

Based on TCM theory, diagnosis and treatment of diseases are achieved through analysis and integration of evidences. The penetration and integration of TCM evidence-based thinking with AI technology help to create an intelligent, objective, and repeatable standardized system model, which can provide theoretical support for the scientificity of TCM and the more efficiency and convenience of diagnosis and treatment.

Chunsheng Wu et al. [15] proposed an unified intelligent TCM framework based on an edge cloud computing system, which built a model incorporated with deep learning algorithms, and then validated it with the identification and typing of hypertension and colds. Computer-aided evidence identification and prescription recommendation were finally achieved. Yin and Yang are a general outline of eight syndromes, which are two platforms for identifying the attributes of diseases. Disorder of Yin and Yang leads to diseases. Therefore, identification of Yin and Yang is crucial for clinical diagnosis in TCM.

Yanmin Qian et al. [16] inputted unstructured text from medical records and conducted experiments using two end-to-end algorithms. It confirmed the feasibility of end-to-end text classification algorithms for Yin and Yang identification in unstructured health records with an accuracy of 92.55%. In TCM, which has been effective in treating advanced lung cancer and prolonging life, accurate identification of evidence plays a pivotal role in the treatment.

Qingchen Zhang et al. [17] also built a TCM diagnosis model for lung cancer identification and typing by inputting unstructured text medical records using an end-to-end model. It can maximize medical record utilization, and through data augmentation and model fusion, deep learning-based multi-label classification methods can better model TCM discriminative evidence for complex diseases such as advanced lung cancer.

Since the complex and nonlinear relationship between symptoms and signs in TCM, Qian Hu et al. [18] accordingly constructed a discriminative model for chronic gastritis diagnosed by TCM using deep learning with multi-label. The results showed that deep learning improved evidence recognition accuracy and at the same time provided a reference for clinical practice.

7. Intelligent application of TCM data for stroke early warning

7.1. Systematic algorithm of real-world diagnosis and treatment data based on transformation from TCM theory to practice

To build a smart health management system for TCM clinical transformation, the ability of processing complex data by relevant intelligent devices, quality of output images, and degree of user cooperation are key elements. With deepening integration of AI and medical intervention, the application of AI technology in TCM is flourishing.

Digital TCM diagnosis, intelligent decision-making systems, and TCM theory modernization have all made great progress. Smart TCM formation cannot be achieved without the support of big data, cloud platforms, and the Internet. Typical features of TCM big data are continuous cycle, dynamic processing, holistic system, and visible value, establishing micro connections from macro data and transforming cognitive information from algorithmic evolution. AI technology empowering TCM big data forms an integrated modern TCM discriminative thinking strategy of data collection-analysis-diagnosis-treatment.

Our group intends to collect pulse diagnosis data, important physiological index (such as body temperature, heart rate, blood pressure, blood oxygen, etc.), corresponding time and position, and living environment data by wearing smart wearable devices. Using flow sequence method for fine classification of data collection and intensive monitoring by period, we will explore the establishment of organ health state assessment based on the probability distribution of the original input data in the form of dividing intervals. The model is shown in Fig. 7 .

Fig. 7.

Health monitoring based on big data collection and organ health assessment model.

7.2. Intelligent pathology model for stroke early warning management

The 5 G (5th Generation Mobile Communication Technology) wearable device is used as a collection and transmission channel of important physiological indicators and health-related factors of individuals, and relevant data from classical theory of TCM midnight-noon ebb-flow of human internal organs, extract and classify stroke disease features based on medical diagnosis text [19,20], and use AI machine learning algorithms to extract correlation features and build a multidimensional health model of human internal organs. To design intelligent pathological models for stroke, establish research directions, explore sensors and monitors for early warning are the key technologies that need to be addressed.

Based on the organ health assessment model, our group explored the construction of a stroke intelligent pathological model using stroke clinical diagnosis pathological data set and Recurrent Neural Network (RNN) algorithm based on the TCM midnight-noon ebb-flow and 12 meridians transmission relationship of internal organs (Fig. 8 ).

Fig. 8.

Intelligent stroke pathology model based on circulating neural network.

8. Experiment and verification

The paper took elderly population and permanent residents with health risks as the research experimental subjects, and 100 cases of high-risk people be selected for monitoring and warning. The data collection period is 12–24 months. Specific enrollment criteria are as follows: (1) age 50–75 years; (2) any gender; (3) at least one or more stroke risk factors (obesity, hypertension, diabetes, hyperlipidemia, history of smoking, family history of early-onset cerebrovascular disease); (4) ability to cooperate with proper completion of the wearable device for continuous monitoring and data collection.

Firstly, the data was collected by ANDUN smartwatch, analyzed in the background with an intelligent big data monitoring system, and then the health status of the patient was assessed. Finally, the assessment results were compared with the diagnosis results performed by professional doctors in regular hospitals with the equipment to verify the scientificity and validity of the assessment.

8.1. Case 1: stoke monitor

Participant 1, male, 59 years old with medical history of lacunar infarction, high blood sugar, hypertension, and diabetes, started to wear the smart watch from February of 2022. During the time from September 24 to 30, 2022, the big data monitor system caught abnormal indicators (Fig. 9 ). Based on the data, the system sent an assessment result: “Heart and brain blood supply insufficiency, which may induce cardiac-cerebro vascular diseases.” It also suggested the participant to seek medical attention promptly (Fig. 10 ).

Fig. 9.

Abnormal physiological indicators monitored by big data monitor system for Case 1.

Fig. 10.

Big data risk analysis and early warning for Case 1.

Participant 1 was hospitalized and examined on October 1, 2022. It reported: multiple infarction areas in the brain, brain atrophy. The system gave the same diagnosis as doctor did. As shown in Table 2 .

Table 2.

Test results and doctor's diagnosis for Case 1.

Image report from Dandong Traditional Chinese Medicine Hospital
Out patient#: 044,256	Bed#:
Name: XXX	Gender: Male	Test#: S-202,210,014,428
Age: 59y	Equipment: CT	Department
Test location	Test date: 2022–10–1
Description: Multi-point lamellar low-density foci were seen in the bilateral semiovale center, lateral ventricle, and basal ganglia areas. Periventricular white matter density decreased. There was no abnormal density imaging in other areas of brain. Ventricles, sulci, and fissures widened and deepened, and midline structures were centered.
Diagnosis: multiple cerebral infarctions, leukoaraiosis, brain atrophy
Report Doctor: XXX	Report date: 2022–10–01

Note: this report is only for doctor's reference.

The doctor's diagnosis is highly consistent with the assessment result from the ANDUN big data analysis system.

8.2. Case 2: cardiovascular monitor

Participant 2, female and 72 years old, is from Beijing. She started to ware ANDUN smart watch from June 19, 2021. On September 1, 9, 18, and 25 of 2021, the system monitored abnormal (Fig. 11 ) and launched early warning “risk of cardiovascular disease attack” (Fig. 12 ). On September 24, 2021, the participant saw a doctor in Capital Medical University Beijing Friendship Hospital, and the diagnosis was shown in Table 3 .

Fig. 11.

Abnormal physiological indicators monitored by big data monitor system for Case 2.

Fig. 12.

Big data risk analysis and early warning for Case 2.

Table 3.

Test results and doctor's diagnosis for Case 2.

Capital Medical University Beijing Friendship Hospital
CT report
Test#: 202,109,141,000**** Register#:13,447,273
Name: XXX Gender: Female Age: 72y
Department: Cardiology Diagnosis: Hypertension
Test location: coronary artery Test date: 2021–09–24
Technology: Scan area: bottom - apex of heart; Scan parameters: layer thickness, layer interval
Image Findings:
1. The total coronary calcification score is 1110.03.
2. The coronary artery is right-dominant.
Calcified plaques can be seen on the wall at the beginning of the left main trunk, and the lumen was slightly narrowed.
Calcified plaques were seen on the wall of proximal anterior descending, and the lumen was moderately to severe narrowed. Non-calcified plaques were seen on the wall of the middle part, and the lumen was slightly narrowed. The distal lumen of the artery had no stenosis.
The first diagonal branch was well visualized. No plaque and stenosis were seen.
Non-calcified plaques were seen in the proximal section of the circumflex branch, and the lumen was narrowed by 25–49%, and no plaques and stenosis were seen in the distal section.
The middle branch was well visualized, and no plaque and stenosis were seen.
Calcified plaques were seen in the proximal and distal sections of the right coronary artery, and the lumen was moderately narrowed; calcified plaques were seen in the middle section, and the lumen is severely narrowed.
The posterior descending branch was well visualized, no plaque was seen on the wall, and no stenosis was seen in the lumen.
3. Myocardial density was uniform, no hypertrophy or thinning was seen, and no abnormal imaging was seen in the heart chamber.
4. Calcified plaques were seen on the aortic wall within the scanning range.
Impression:
1. The total coronary calcification score is 1110.03.
2. The coronary artery is right dominant.
3. Coronary atherosclerosis manifestations: the degree of stenosis of the lumen is shown above.
4. Atherosclerotic changes in the aorta.
Report doctor: XXX Review doctor: XXX Report date: 2021–09–26
Note: this report is only for doctor's reference

The doctor's diagnosis is lumen stenosis and aorta atherosclerotic changes, which is highly consistent with the assessment result from ANDUN big data analysis system.

8.3. Case 3: heart rate monitor

Participant 3, male and 67 years old from Weifang has the medical history of hypertension. He started to wear the ANDUN smart watch from September 19, 2021, and physiological indices presented abnormal frequently (Fig. 13 ), and atrial fibrillation warning triggered (Fig. 14 ). On July of 2022, by a return visit call from ANDUN big data stage, the participant said “he hadn't uncomfortable in his heart.”. Customer service told the participant that many hear problems have no feeling at the beginning and suggest the patient to seek medical attention and prevent any heart diseases. On August 1, 2022, the participant did a physical examination in the hospital and the report shows sporadic atrial premature heartbeat and paroxysmal tachycardia. The diagnosis is in Table 4 .

Fig. 13.

Abnormal physiological indicators monitored by big data monitor system for Case 3.

Fig. 14.

Big data risk analysis and early warning for Case 3.

Table 4.

Test results and doctor's diagnosis for Case 3.

Weifang Traditional Chinese Medicine Hospital
Name: XXX Gender: Male Age: 67y Department: Cardiology Report doctor: XXX Outpatient#: Hospital register#: Bed#:
Diagnosis
Heart rate The slowest heart rate-4 interval: 49 bpm at 22:03 The fastest heart rate-4 interval: 49 bpm at 22:03 Average heart rate-24 h: 61 bpm Average the slowest heart rate/hour: 53 bpm at 22:00 Average the fastest heart rate/hour: 71 bpm at 8:00 Analyze heart beat: 87,784 Analyze minute: 1421 Dynamic electrocardiogram record time: 23 h and 41 m	Ventricular rhythm Total premature ventricular beat: 0 Total paired premature ventricular beat: 0 Total ventricular tachycardia:0 The longest ventricular tachycardia: No Ventricular tachycardia with the fastest heart rate: No Tachycardia with the slowest heart rate: No Per heart beat /number of premature ventricular beat per hour: 0/0 % of premature ventricular beat: 0.00 R on T: No	Heart rate variability SDNN-24 h: 111 SDANN:104 SDNN Index: 32 Rmssd: 22 Pnn50: 2 Frequency domain power-24 h: 970.1 The smallest frequency domain power per hour: 371.3 The biggest frequency domain power per hour: 1834.4
ST segment analysis Total ST minute CH1:0(I) Total ST minute CH2:0(II) Total ST minute CH3:0(III) Maximal Delta ST depression: No Maximal Delta ST elevation: No The longest ST segment: No The fastest heart rate on ST segment: N/A Total ischemic burden: 0.001	Atrial rhythm Total atrial premature beat: 34 Total paired atrial premature beat:0 Total atrial tachycardia: 2 (last 5.7 S) The longest atrial tachycardia: 3.6 s 132 bpm (15:45:16) Atrial tachycardia with the fastest heart rate: 3.8 s 132 bpm (15:45:16) % of premature atrial beat: 0.04% Total junctional beats/junctional rhythm : 0/0 % of Atrial fibrillation/atrial flutter: 0	Bradycardia Heart beat >2.00 s: 0 The longest pause: No The longest R-R interval: 1.234 s (05:29:07) QT The biggest QT: 413 ms The biggest QTc: 448 ms The biggest QTc interval: at 15:06 72 bmp Ventricular escape: No
Conclusion 87,784 heart beats were analyzed, in which average heart rate was 61 bpm, the slowest heart rate was 49 bpm happened at 22:03, and the fastest heart rate was 94 bpm happened at 17:41. There were 34 atrial premature beats, of which 23 were single atrial premature beats, and 2 paroxysmal atrial tachycardia. The longest R-R interval was 1.234 s happened 05:29:07. Sinus rhythm, occasional atrial premature beats, short and paroxysmal atrial tachycardia, ST-T changes. Report doctor: XXX Report date: 2022–08–01
standard deviation of NN intervals (SDNN), root-mean-square of successive difference of NN intervals (Rmssd), percent of NN (Pnn),hour(h), minute (m), second (s), millisecond (ms)

Shown in Table 4, the doctor's diagnosis is highly consistent with the assessment result from ANDUN big data analysis system. The timely and early warning from ANDUN monitor system attracted the watch-wearer a special attention. Through equipment test and doctor's diagnosis in the hospital, a report of high risk for abnormal heart rhythm diseases was sent to the participant, in order to help the participant to timely prevent the diseases.

9. Application prospects

Directed by the theory of AI and big data empowering TCM, we will study the personal health management system and risk perception system using the human health state model, make full use of big data distributed storage, rapid processing, refinement insight, visualization display and other technologies to monitor the potential risk of stroke and real-time early warning, through clinical detection and evaluation, successfully improve stroke treatment by big data support system for comprehensive prevention, control, and emergency treatment. We will also analyze the correlation between behavior and health, support the development and upgrading of recreation, medical care, residence, and pension industries, provide scientific and technological support for the recreation industry, and provide a basis for decision-making on the industrial layout.

By operating the physical health management system and risk perception system, our group will use telephone interviews, cell phone programs, questionnaires, etc. to obtain data related to the recreation and residence population: gender, age, income, length of stay, travel mode, expenditure, information access, motivation, preference, recreation experience, etc., and use in-depth interviews to obtain the core and deep elements of recreation needs for statistical processing. The relationship changes between travel experience and health index will be obtained, and the analysis between travel experience and health index changes will be conducted for the results of health index changes influenced by travel experience in recreation and recreation; the healthcare needs of different people in the process of travel and recreation will be analyzed in combination with interviews. The scientific and technological elements of recreation and recreation product development support and service innovation model will be obtained in this way (Fig. 15 ).

Fig. 15.

Evidence-based study on the association between physical health and wellness travel behavior.

With the development of sensors, 5 G communication, and other technologies, new applications of wearable mobile health monitoring devices and personal health management systems based on cloud services are beginning to emerge to make up for the lack of cardiovascular and cerebrovascular disease monitoring and management systems by tapping into the potential risks of personal physiological data.

In the long run, TCM combines with continuous monitoring devices of body indicators for various hidden abnormal changes of physiological indicators, in order to seek the correlation with potential disease risk of body organs. Refined perception of individual abnormalities and precise implementation of stroke risk prevention and control with the help of artificial intelligence and big data technology is the development direction of cardiovascular and cerebrovascular disease management. This is the way to reduce the risk of cardiovascular and cerebrovascular diseases.

From the holistic concept of "Heaven and Man" to the thinking system of diagnosis and treatment, artificial intelligence technology can help to break through the shackles and barriers of TCM, and build a modernized TCM diagnosis and treatment model in the post-COVID-19 era [21].

10. Conclusion

Focusing on the problems of insufficient data, limited technology, and lack of professional evaluation methods by physicians with medical theory in health management and application, this study explored a holistic health management approach based on physiological data collected by wearable smart watch, Narrow Band Internet of Things (NB-IoT) transmission, health big data archives management, and twelve o'clock regimen of the TCM midnight-moon ebb-flow. Through the professional health medical management team and mobile APP, a health management big data service backstage has been established for individual health diagnosis, assessment, disease prediction, and early warning. Meanwhile, the comparison of the clinical medical diagnosis with the analysis and evaluation of the big data intelligent monitor system proved that the method described in this study is scientific and effective, and significant to full-cycle monitoring of human health and early warning of diseases. Especially, we established the monitoring model of human viscera and organ health based on TCM midnight-moon ebb-flow by big data and artificial intelligence algorithm, which provides a new idea and method for health management and maintenance in the post-COVID-19 era.

CRediT authorship contribution statement

Xiaoyong Chen: Conceptualization, Software, Methodology, Writing – original draft. Boxiong Yang: Conceptualization, Software, Data curation, Writing – original draft. Shuai Zhao: Data curation, Investigation, Writing – original draft. Wei Wei: Writing – original draft. Jialu Chen: Resources, Visualization. Jie Ding: Writing – review & editing. Hong Wang: Validation. Peng Sun: Validation. Lin Gan: Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no conflicts of interest to report regarding the present study.

Biographies

Xiaoyong Chen received the B.S. degree in Medicine from Hunan University of Chinese Medicine in 1992. He was employed as a master's tutor of Guangzhou University of Chinese Medicine in 2012. His area of research interests include digital diagnosis and treatment, application of traditional Chinese medicine data and health management.

Boxiong Yang received the B.S. degree in Computer Software from Huazhong Normal University, China, the M.S. degree in Geophysics from the Institute of Seismology, Seismological Bureau, China, and the Ph.D. in Geophysics from Institute of Geophysics, China Seismological Bureau in 1997, 2000 and 2005, respectively. His area of research interests include Artificial Intelligence and Big Data.

Shuai Zhao received the B.S. degree in Acupuncture and Massage from Hubei University of Chinese Medicine, China, and the M.S. degree in Orthopedics and Traumatology of Traditional Chinese Medicine from Guangzhou University of Chinese Medicine, China in 2014 and 2017, respectively. Her area of research interests include traditional Chinese medicine rehabilitation and chronic disease health management.

Wei Wei received a bachelor's degree in Traditional Chinese Medicine from Beijing University of Chinese Medicine in 2014 and a master's degree in Acupuncture and Massage from Gansu University of Chinese Medicine in 2020. Her research interests include traditional rehabilitation and acupuncture therapy.

Jialu Chen received the B.S. degree in Medical Laboratory Changsha Medical College in 2015 and a health management qualification certificate in 2020. Her area of research interests include medical testing and health information management.

Jie Ding received MD in 1986, MS and PhD in Surgery in 1989 and 1997 from Tianjin Medical University, China, PhD in Biochemistry in 2003 from University of Kagawa, Japan. She completed PDF training in Biochemistry in 2007 at University of Alberta, Canada. Her research focuses on the cellular and molecular mechanism of wound healing and hypertrophic scarring.

Hong Wang received the B.S. degree in Traditional Chinese Medicine from Hunan University of Traditional Chinese Medicine, China, the M.S. degree in Health Services Management from the Business School, Manchester University, England in 1992 and 2004, respectively. His area of research interests include clinical practice, health services management and administration.

Peng Sun received the B.S. degree in Department of Medicine from Qingdao Medical College, China, and the M.S. degree in Surgery from Weifang Medical University, China in 1987 and 2002, respectively. His area of research interest includes diagnosis and treatment of cerebral vessels.

Gan Lin (tutor at the University of Sanya, China) received her B.Sc. in President University and Columbia University in 2017, the M.S. from Hong Kong Polytechnic University, in 2019. Her research interests include text-mining, sentiment analysis, and media computing.

Data availability

• The authors do not have permission to share data.