Research on Traditional Chinese Medicine Intelligent Diagnosis

Based on Arterial Blood Detection and Pulse Diagnosis

Jianjin Tang

Mechanical Engineering and Automation, Northeastern University, Shenyang, Liaoning, China

Keywords: Artificial, Intelligence, Intelligent Diagnosis, Traditional Chinese Medicine.

Abstract: Traditional Chinese Medicine (TCM) pulse diagnosis has always been regarded as an important means to

reflect the operation of qi and blood in the body and the functions of the zang-fu organs. However, the

traditional method has long relied on the subjective experience of doctors and lacks objective and quantitative

standards, which limits its scientific promotion. This paper reviews the current research progress and technical

approaches related to the combination of arterial blood detection technology and TCM pulse diagnosis. This

article explores the use of modern instruments and equipment such as ultrasonic flowmeters and optical

sensors to obtain key physiological indicators, including hemodynamics, pulse waveforms, vascular elasticity,

and heart rate variability, and to establish a quantitative model of TCM pulse conditions. At the same time,

multi-modal data fusion and machine learning methods are introduced to conduct feature extraction,

classification, and intelligent diagnosis of pulse diagnosis signals. Moreover, a human-machine interaction

system is integrated to optimize the display and practicality of the diagnostic results. The research shows that

this technical solution is expected to significantly improve the accuracy and repeatability of TCM diagnosis,

providing new ideas for the coordinated development of TCM and Western medicine.

1 INTRODUCTION

Over thousands of years, the diagnosis and treatment

system of Traditional Chinese Medicine (TCM) has

developed unique theories and practices. Among

them, the four diagnostic methods of "inspection,

auscultation and olfaction, interrogation, and pulse

taking and palpation", especially the part of pulse

diagnosis, embodies TCM's unique insights into life

and diseases. Although traditional pulse diagnosis has

its unique charm, it often relies on the personal

experience of doctors, and it is difficult to establish a

quantitative evaluation standard. This has imposed

certain limitations on its application within the

modern medical system. In recent years, how to

transform the "empirical judgment" in TCM pulse

diagnosis into data-driven indicators and achieve

objective diagnosis through modern technologies has

gradually become an important issue that urgently

needs to be addressed in the field of integrating

traditional Chinese and Western medicine.

As the core of TCM pulse diagnosis, the pulse

condition not only reflects the physical movement of

the blood but also reveals the overall information

about the qi and blood of the body as well as the

functions of the zang-fu organs. With the rapid

development of artificial intelligence, signal

processing, and biosensing technologies, an

increasing number of scholars are attempting to

transform traditional pulse diagnosis into quantifiable

data. For example, some studies have utilized multi-

point sensors to achieve the synchronous collection of

pulse signals from the three locations of "cun",

"guan", and "chi". This breaks through the limitations

of single-point collection in the past and provides data

support for the pattern recognition of pulse conditions.

However, most of the current pulse diagnosis

instruments are based on methods such as pressure

sensors, and there is a significant difference between

the pulse diagnosis results obtained by these

instruments and those of clinical physicians (Hsieh et

al., 2021). Many scholars have also attempted to

introduce more meticulous physiological

measurement methods. For instance, they use

Doppler ultrasound technology to explain the

characteristics of traditional pulse conditions,

providing a basis of physical quantities for

descriptions like "stringy", "slippery", and "thin" in

pulse diagnosis. The research on pulse diagnosis is

currently in a rapid development stage where tradition

Tang, J.

Research on Traditional Chinese Medicine Intelligent Diagnosis Based on Arterial Blood Detection and Pulse Diagnosis.

DOI: 10.5220/0014320400004718

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on Engineering Management, Information Technology and Intelligence (EMITI 2025), pages 66-70

ISBN: 978-989-758-792-4

and modernity converge, and technology and

medicine integrate. The CiteSpace software can be

used to understand the research status quo and

development trends in this field by measuring the

literature in the relevant domain (Hou & Hu, 2013,

Chen et al., 2015). This paper combs through the

research status quo and hotspots of TCM pulse

diagnosis, and points out that future research will

focus more on intelligent diagnosis, integration of

multimodal information, and quantitative extraction

of pulse characteristics.

At present, most of the research is still limited to

signal collection and the extraction of single

characteristics, and the correlation between the pulse

condition and the overall physiological and

pathological state of the human body has not been

fully explored. Combining modern precision

detection technologies such as arterial blood testing

with pulse diagnosis can not only fill the information

gap of traditional pulse diagnosis but also provide a

basis for biological verification of TCM diagnosis. As

an important flowing tissue in the body, arterial blood

has rich indicators and mature collection techniques,

and it has become an important data source for

modern medical testing.

This study aims to construct an intelligent

diagnosis model that integrates modern detection

techniques and TCM theories. By means of high-

precision pulse condition collection and standardized

arterial blood testing, a multimodal health data

system will be established. Moreover, machine

learning and data mining techniques will be applied

to extract representative features, so as to achieve

accurate classification of disease states and the

construction of a TCM syndrome differentiation

model. It is hoped that this will build a bridge between

pulse condition data and biochemical indicators,

enhancing the persuasiveness of TCM in the context

of modern medicine.

2 INTELLIGENT DIAGNOSIS

SCHEME AND DISEASE

CONDITION PREDICTION:

SENSORS AND DATA

ANALYSIS

2.1 Design and Implementation of the

Sensor System

The scheme envisioned in this paper adopts a brand-

new design of the sensor module, striving to

organically integrate arterial blood testing with the

collection of pulse diagnosis signals. With high-

precision pressure sensors and photoelectric sensors

as the core, this module combines microfluidic

technology and wireless data transmission to achieve

real-time monitoring of the pulse waveform and

blood biochemical indicators. For the arterial blood

testing part, a color Doppler ultrasound diagnostic

instrument is used to collect the blood flow

spectrograms of normal pulse, superficial pulse, deep

pulse, stringy pulse, slippery pulse, weak pulse and

unsmooth pulse (Gao et al., 2024). At the same time,

for the pulse diagnosis signal collection part, a high-

resolution pressure sensor array is employed to

accurately record the subtle changes in the pulse

waveform, and with the help of a high-speed data bus,

real-time data transmission with the host computer is

achieved. The entire module adopts a modular and

standardized design, thus ensuring the synchronism

and consistency of signal collection in each part,

laying a solid foundation for subsequent data

processing (Tian et al., 2025).

2.2 Methods of Data Collection and

Preprocessing

During the data collection process, this scheme

adopts a dual-channel synchronous collection

strategy: on the one hand, it conducts quantitative

determination of the biomarkers in the blood, and on

the other hand, it performs continuous and dynamic

collection of the pulse signals. A high-precision timer

is used to ensure the precise synchronization of the

sampling time. After the collected raw data are

processed by hardware signal amplification and

filtering, they are further subjected to combined low-

pass and high-pass filtering according to the

measurement principles of traditional pulse graph

parameters (such as h1, t1) (Zhang et al., 2025). A

band-pass filter is used to eliminate noise and

baseline drift in the pulse signals. For the positioning

and extraction of the pulse waves, the Shannon

energy envelope line and Hilbert transform methods

are adopted (Cui et al., 2018) to filter out external

noise and interference. Subsequently, at the software

level, through real-time normalization and smoothing

algorithms, short-term mean value correction is

applied to the continuous curve, and time-frequency

joint analysis methods such as wavelet transform are

used to extract key features, providing high-quality

input for multimodal data fusion and ensuring that the

results of intelligent diagnosis are objective and

clinically practical.

Note：Observation Indicators of the Pressure-based

Pulse Diagnosis Instrument and Their Interpretations

Research on Traditional Chinese Medicine Intelligent Diagnosis Based on Arterial Blood Detection and Pulse Diagnosis

Modern Pulse Diagnosis of TCM has provided

detailed definitions of the pulse graph parameters

(Figure 1), which mainly include the following

contents: h1: The height of the main wave; h3: The

height of the pre-dicrotic wave; h4: The height of the

mid-diastolic notch; h5: The height of the dicrotic

wave; t1: The time value from the starting point of the

pulse graph to the peak point of the main wave; t3:

The time value from the starting point of the pulse

graph to the pre-dicrotic wave; t4: The time value

from the starting point of the pulse graph to the mid-

diastolic notch; t5: The time value from the mid-

diastolic notch to the ending point of the pulse graph;

t: The pulsation cycle, that is, the time value from the

starting point to the ending point of the pulse graph

(the heart rate can be indirectly calculated); w1: The

height at 1/3 of the main wave; w2: The height at 2/3

of the main wave; w: The time duration during which

the pressure in the left ventricle and the large arteries

remains at a high level; w1/t and w2/t: The elasticity

at the fixed position on the amplitude of the main

wave and the magnitude of the peripheral resistance;

t1/t and t1/t4: Cardiac ejection; h4/h1 and h5/h1: The

peripheral resistance of the blood vessels; h3/h1: The

elasticity of the arterial vessels and the peripheral

resistance; U angle: The viscosity of the blood and the

elasticity of the blood vessels; s1: The slope of the

ascending branch; s2: The slope of the rapid

descending segment; s3: The slope of the slow

descending segment (Fei, 2003).

Figure 1: Main measurement parameters of the pulse graph

(Fei, 2003).

2.3 Data Fusion and Intelligent

Analysis Algorithms

In the data fusion stage, the scheme adopts the

popular multimodal data fusion technology in the

current objective research of TCM pulse diagnosis to

achieve precise docking between the arterial blood

test data and the pulse diagnosis signals. Firstly, a

time alignment algorithm is used to ensure the

consistency of the two types of data at the collection

moment. Secondly, a multi-layer feature extraction

framework is constructed by using a Convolutional

Neural Network (CNN) and a Support Vector

Machine (SVM). Then, time-domain, frequency-

domain, and time-frequency joint analysis is carried

out on the preprocessed data to extract key parameters

that can quantitatively describe the characteristics of

pulse diagnosis (Yi et al., 2024). The constructed

feature vectors are then used as inputs and fed into the

intelligent diagnosis model that has been repeatedly

trained and optimized with large-scale data samples.

Supported by historical data, this model can achieve

accurate identification of TCM syndromes and

diseases (Yi et al., 2024). At the same time, the

ensemble learning model of the TCM-assisted

diagnosis system based on artificial intelligence

(iterative algorithms (Luo et al., 2022), Gradient

Boosting Decision Tree can be applied to avoid their

respective weaknesses (Yan et al., 2022)), and

conduct disease condition analysis to ensure the

accuracy of the diagnosis. The ensemble learning

model used for syndrome prediction can utilize four

existing prediction methods (Backpropagation,

Random Forest, Extreme Gradient Boosting, and

Support Vector Classifier) to achieve consistently

high prediction accuracy (Zhang et al., 2020). The

overall data processing flow not only takes fully into

account the nonlinear characteristics of the sensor

data, but also incorporates the quantitative indicators

from the TCM diagnostic methods of "inspection,

auscultation and olfaction, inquiry, and pulse-taking"

into the model, thus achieving a deep integration of

traditional experience and modern intelligent

technology (Chen et al., 2025).

2.4 Result Feedback and System

Optimization

After completing the data fusion and feature

extraction, the intelligent diagnosis system will

provide feedback on the diagnostic results to the

physicians in the form of intuitive charts and detailed

reports. It not only displays the changing trends of

key pulse condition parameters but also presents an

analysis of the preliminary pathological state. The

system also has a built-in feedback mechanism. By

comparing the real-time data with the historical

standard database, it will regularly retrain the model

and calibrate the parameters, so as to ensure the

accuracy and stability of the detection indicators. In

clinical trials, the system compares and validates its

results with those of traditional pulse diagnosis

through multiple evaluation indicators such as

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

accuracy rate, recall rate, and F1 value, demonstrating

its high clinical reference value. In addition, the

device supports remote data storage and multi-center

sharing, enabling multi-dimensional data integration

and big data analysis. This provides technical support

for the continuous optimization of the objectivity and

intelligence of TCM pulse diagnosis, and also lays a

scientific and standardized foundation for future

clinical decision-making assistance.

3 APPLICATION PROSPECTS

AND CHALLENGES

3.1 Application Prospects

The intelligent diagnostic technology that combines

arterial blood testing with pulse diagnosis has broad

development potential in the field of TCM. With the

progress of sensor, artificial intelligence, and big data

technologies, the intelligent diagnosis system of TCM

will be improved in many aspects. Firstly, the

integration of multimodal physiological data has

become a trend. By using high-precision sensors to

collect various types of information such as pulse

waveforms, hemodynamic parameters, and blood

biochemical indicators, and combining them with the

syndrome differentiation elements such as the

patients' symptoms and the information obtained

from inquiry, comprehensive objective data can be

provided for diagnosis. Signal processing and pattern

recognition algorithms can explore the correlations

between pulse conditions and disease states from a

large number of clinical samples.

Secondly, the development of remote monitoring

and wearable devices will drive the intelligent

diagnosis of TCM out of the consulting room and

expand it into community and home scenarios.

Patients can collect health data at any time through

wearable pulse sensors and portable detectors,

providing data support for the monitoring of chronic

diseases and health management. In communities or

remote areas with limited medical resources, such

technologies can improve the accessibility of

diagnosis and treatment.

Thirdly, the in-depth application of artificial

intelligence technology empowers the intelligent

diagnosis of TCM. Researchers can utilize deep

learning models to train diagnostic models and

continuously optimize the algorithms. Technologies

such as knowledge graphs combine TCM theories

with data-driven methods, improving the

interpretability of the system's conclusions regarding

pulse diagnosis. This integration helps to reduce the

dependence on empirical judgment.

In addition, the development of intelligent

diagnosis in TCM is closely related to the digital

transformation of the medical system. In the context

of electronic health records and intelligent medical

environments, TCM pulse data can be integrated with

Western medicine indicators, providing multi-

dimensional references for clinical decision-making.

In the future, the intelligent TCM diagnosis system is

expected to play a role in fields such as collaborative

diagnosis and treatment between TCM and Western

medicine, as well as personalized health

management. The data-driven auxiliary diagnosis

system can provide reference opinions for doctors,

promoting the standardization of the diagnosis and

treatment process; at the same time, it helps with the

inheritance of experience and improves the quality of

medical care.

3.2 Challenges

Despite its broad application prospects, the actual

implementation of intelligent diagnosis in TCM still

faces multiple challenges, which require

comprehensive responses from aspects such as

technology, ethics, policies, and user behavior.

From the technical aspect: The pulse diagnosis

signals in TCM are dynamically complex, and there

are significant differences in pulse conditions among

different individuals, making it difficult to establish a

unified standard. Current sensing devices are limited

by precision and environmental factors, which affect

the stability of the data. The fusion and modeling of

multi-source heterogeneous data remain a difficult

problem, and the insufficient compatibility of data

from different devices also restricts the utilization of

large-scale data. Training diagnostic models requires

a large amount of annotated data. However, the lack

of a unified annotation and sharing mechanism TCM

diagnostic data restricts the generalization ability of

the algorithms. Finally, the polysemy of TCM

theories poses a challenge to the interpretability of the

models. It is necessary to combine expert knowledge

with data analysis to improve the traceability of

diagnostic decisions.

From the ethical aspect: The intelligent diagnosis

system involves patients' sensitive health data. How

to protect privacy, prevent leakage and abuse, and

ensure informed consent are issues that must be

resolved. Algorithmic decision-making may affect

the trust between doctors and patients. If

misdiagnosis or bias occurs, it will have a negative

impact on this trust. Therefore, a transparent

Research on Traditional Chinese Medicine Intelligent Diagnosis Based on Arterial Blood Detection and Pulse Diagnosis

evaluation and feedback mechanism should be

established to continuously monitor and correct the

diagnostic results, and the auxiliary position of

artificial intelligence in clinical practice should be

clearly defined to prevent excessive dependence.

From the policy aspect: Currently, there is a lack

of unified supervision standards for the technologies

and devices related to intelligent diagnosis in TCM.

For the promotion and application, it is necessary to

improve the policies and regulations regarding

software certification, data security, and the

qualifications of medical devices. It is also essential

to coordinate the differences between TCM and

Western medicine in terms of diagnosis and treatment

records as well as the indicator systems, so as to

achieve data intercommunication and collaborative

diagnosis and treatment. In addition, research,

development, and application require policy and

financial support. The lack of supportive policies and

talent cultivation plans will also affect the

development of this field.

From the user behavior aspect: New technologies

need to gain the trust of both doctors and patients.

Some TCM practitioners may have reservations about

intelligent diagnostic tools, believing that machines

are difficult to replace experience. The trust level of

patients towards machine-assisted diagnosis also

needs to be improved. Therefore, it is necessary to

provide training to enhance the understanding and

acceptance of the technology by both doctors and

patients, and optimize the system design to meet the

needs of different users. Only by taking into account

both technical and humanistic needs can intelligent

diagnosis in TCM truly be integrated into clinical

practice and contribute to social health services.

4 CONCLUSION

The development of the intelligent diagnosis system

of TCM provides a new way for the modern

transformation of TCM. By integrating artificial

intelligence, sensing technology, and the concept of

the integration of TCM and Western medicine, it can

not only improve the diagnostic accuracy but also

help TCM go global. With the continuous

advancement of technology, intelligent diagnosis in

TCM is expected to play an increasingly important

role in the global medical system and make greater

contributions to the public health cause.

REFERENCES

Chen, Y., Chen C. M., Li, Z. Y., et al.: 'The Methodological

Functions of the Knowledge Map of CiteSpace',

Scientia Sinica (Studia Naturalia), 2015 (2): 242-253

Chen, Z., Zhang, D., Nie, P. F.: ‘Developing the Artificial

Intelligence Method and System for "Multiple Diseases

Holistic Differentiation" in Traditional Chinese

Medicine and Its Interpretability to Clinical Decision.’

EVIDENCE-BASED MEDICINE, 18(2), 2025

Cui, J., Tu, L. P., Zhang, J. F., et al.: 'Analysis of pulse

signals based on array pules volume', Chin J Integr Med

,2018, 25(2): 103-107

Fei, Z. F.: 'Modern Pulse Diagnosis in Traditional Chinese

Medicine', Beijing: People's Medical Publishing House,

2003: 5-30

Gao, L. S., Li, Y., Yu, X. F., et al.: 'Screening of

Identification Indicators of Traditional Chinese

Medicine Pulse Patterns Based on Doppler Ultrasound

Technology', Journal of Binzhou Medical University,

2024, 47(05): 385-388

Hou, J. H., Hu, Z. G.: 'Review and Prospect of the Research

on the Application of CiteSpace Software', Modern

Information, 2013(4): 99-103

Hsieh, T. C., Wu, C. M., Tsai, C. C., et al.: 'Portable

Interactive Pulse Tactile Recorder and Player System',

Sensors (Basel), 2021, 21: 4349

Luo, Z. Y., Cui, J., Hu, X. J., et al.: 'A study of machine-

learning classifiers for hypertension based on radial

pulse wave', Biomes Signal Proc Control, 2022

Tian, H. Y., Huang L. Q., Xian, N, X., et al.: 'Automatic

Pulse Signal Acquisition System Based on a Six-Array

Sensor', Transducer and Microsystem Technologies,

2025, 44(01): 83-87

Yan, J. J., Cai, X. L., Zhu, G. Y., et al.: 'A non-invasive

blood pressure prediction method based on pulse wave

feature fusion', Biomed Signal Proc

Control,2022,74;103523

Yi, K., Zhang, M. D., Guo, S., et al.: 'Review of the

Research on Data Annotation of Intelligent Equipment

for Pulse Diagnosis in Traditional Chinese Medicine',

Shanghai Journal of Traditional Chinese Medicine,

2024, 58(10): 5-10

Zhang, H., Ni, W. D., Li, J., Zhang, J. J.: ‘Artificial

Intelligence–Based Traditional Chinese Medicine

Assistive Diagnostic System: Validation Study’, JMIR

Medical Informatics, 8(6), 2020

Zhang, J., Liu M. H., Song, X. H., et al.: 'Clinical Research

and Analysis of Pulse Diagnosis Instruments and Their

Detection Indicators', Chinese Journal of Basic

Medicine in Traditional Chinese Medicine, 2025,

31(03): 425-430

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence