A closed-loop medical system and method based on ternary state space and field regulation

Abstract

The application discloses a closed-loop medical system and method based on a three-element state space and field regulation, and belongs to the technical field of intelligent medical treatment and medical engineering. The system comprises a multi-modal sensing and collecting layer for collecting physiological, psychological and environmental data of a target object; a three-element state space modeling layer connected with the sensing and collecting layer, which maps low-dimensional collected data into a three-element state vector containing a manifesting extreme state value, a returning extreme state value and a running central axis sequence parameter through an encoder, and constructs an individualized health state space; a field regulation and treatment layer connected with the state space modeling layer, which is internally provided with a field balance controller, calculates a deviation value through a single composite operation composed of exponential operation, logarithmic operation and difference operation, and generates a regulation and control instruction; and a treatment execution layer connected with the field regulation and treatment layer, which is used for applying precise physical field or chemical field stimulation to the target object according to the regulation and control instruction. The application firstly integrates a three-element state space diagnosis model and a field regulation and treatment model into a unified closed-loop medical architecture, and is suitable for health state evaluation and individualized treatment of various biological systems such as human bodies, animal bodies, isolated organs, tissue engineering constructs and organoids.

Description

Technical Field

[0001] This invention belongs to the field of intelligent medical and medical-engineering interdisciplinary technology. Specifically, it relates to a closed-loop medical architecture that integrates the "ternary state space" diagnostic model with the "field regulation" treatment model. It is applicable to the health status assessment, disease diagnosis and personalized treatment of various biological systems such as human bodies, animal bodies, ex vivo organs, tissue-engineered constructs and organoids. Background Technology

[0002] The current medical system has long faced a paradigm shift: First, the holistic approach of Traditional Chinese Medicine (TCM) and the reductionism of Western medicine lack a unified mathematical language. TCM, through the four diagnostic methods of observation, auscultation, inquiry, and palpation, maps the complex physiological and pathological states of the human body onto the eight principles of differentiation: Yin and Yang, exterior and interior, cold and heat, deficiency and excess. Its diagnostic logic is essentially high-dimensional system modeling, but it has long lacked objective and quantifiable mathematical expression, leading to insufficient consistency in diagnosis and difficulty in interoperability with the data systems of modern medicine. Western medicine, on the other hand, obtains microscopic indicators through laboratory tests and imaging, attributing diseases to specific gene mutations, protein abnormalities, or pathogen infections. Its diagnostic logic is essentially reductionism, but it lacks a unified description of the overall state of the system, making it difficult to explain holistic medical phenomena such as "treating different diseases with the same method" and "treating the same disease with different methods."

[0003] Secondly, diagnosis and treatment are independent and lack a continuous, quantitative closed loop. In existing clinical pathways, the diagnostic system outputs qualitative conclusions (such as "diabetes" or "hypertension"), and the treatment system applies standardized interventions based on these conclusions (such as fixed doses of hypoglycemic or antihypertensive drugs), lacking a continuous, quantitative closed loop from diagnosis to treatment. Doctors cannot track the dynamic changes in the system's state after intervention in real time, nor can they dynamically adjust intervention strategies based on the direction and magnitude of state deviations. This "open-loop" medical model leads to treatment effectiveness being highly dependent on the doctor's personal experience, making it difficult to achieve precise and personalized dynamic control.

[0004] In 2026, the HAPLN1 biomechanical switch study published in *Science* revealed that whether tissue injury leads to "regeneration" or "scarring" is not determined by the cells themselves, but by the "softness and stiffness" of the extracellular matrix—a biomechanical field. This discovery confirms the effectiveness of the "field regulation" principle at the molecular level—by maintaining a "soft" field, the inherent self-healing potential of life can be unlocked. However, this research remains at the level of biological discovery and has not yet formed a clinically applicable closed-loop medical system integrating diagnosis and treatment.

[0005] Therefore, there is an urgent need for a novel closed-loop medical architecture that can provide a unified mathematical language to integrate traditional Chinese and Western medical theories, and integrate diagnostic accuracy and personalized treatment into a continuous, quantitative closed loop. The technical concept of this invention is inspired by ancient Chinese numerology and symbolism, and is combined with cutting-edge modern regenerative medicine, neuroscience, and artificial intelligence technologies. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a closed-loop medical system and method based on ternary state space and field control, in order to solve technical problems such as the lack of a unified mathematical language in traditional Chinese and Western medicine, and the lack of continuous quantitative closed-loop in diagnosis and treatment, which are independent of each other.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A closed-loop medical method based on ternary state space and field regulation includes the following steps: S1: Multimodal perception acquisition steps—collect physiological, psychological, and environmental data of the target object and convert them into standardized multidimensional feature vectors.

[0008] S2: Triadic state space modeling step - Input the multidimensional feature vector into the pre-trained encoder, map it into a ternary state vector containing explicit extreme state values, hidden extreme state values ​​and dynamic axis order parameters, and construct an individualized health state space.

[0009] S3: Field Deviation Calculation Steps—The ternary state vector is input into the field balance controller, and the field deviation value is calculated through a single composite operation consisting of exponential operation, logarithmic operation, and difference operation. The functional form of the single composite operation is: Field deviation value H = e^(current sequence parameter - healthy baseline sequence parameter) - ln(steady-state baseline value).

[0010] S4: Regulation instruction generation steps—When the H value is positive and exceeds the threshold, a regulation instruction to enhance the storage pole suppression signal is generated; when the H value is negative and the absolute value exceeds the threshold, a regulation instruction to enhance the manifest pole activation signal is generated.

[0011] S5: Closed-loop iterative steps—Repeat the above steps until the ternary state vector returns to a healthy steady state.

[0012] Furthermore, the target object includes any one of the following: human body, animal body, isolated organ, tissue-engineered structure, or organoid.

[0013] This invention also provides a closed-loop medical system for performing the above-described methods, comprising a multimodal sensing and acquisition layer, a ternary state space modeling layer, a field-controlled treatment layer, and a treatment execution layer. The field-controlled treatment layer incorporates a single composite computing hardware module with a response period not exceeding 50 microseconds. Detailed Implementation

[0014] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can refer to and implement it.

[0015] Example 1: System Basic Architecture and Core Hardware Implementation The closed-loop medical system of the present invention consists of a four-layer architecture, and the complete closed loop from data acquisition to treatment execution is completed within a millisecond timescale.

[0016] The multimodal sensing and acquisition layer integrates three types of data interfaces: wearable biosensor interface (acquiring physiological data such as heart rate, blood pressure, blood oxygen, skin conductance, and body temperature), medical imaging equipment interface (acquiring imaging data such as ultrasound, CT, and MRI), and electronic health record data interface (acquiring clinical data such as medical history, medication records, and examination results). All data undergoes standardized preprocessing and is output as a multidimensional feature vector.

[0017] The ternary state space modeling layer incorporates a pre-trained encoder network. This encoder consists of three cascaded parts: a ResNet50 feature extraction network (for extracting high-level semantic features from image and temporal data), an LSTM temporal modeling network (for extracting temporal evolution features from continuous physiological signals), and an Attention Pooling layer (for weighted fusion of multimodal features). The encoder compresses the high-dimensional multimodal feature vector into a Yin-Yang binary basis vector (explicit weights and implicit weights), then expands it through a Five Elements weight modeling layer into a five-element state vector containing the weights of the five elements (wood, fire, earth, metal, and water), and finally fuses them into a ternary state vector: explicit extreme state value, implicit extreme state value, and the axial order parameter of the transformation.

[0018] The field-controlled therapy layer incorporates a single composite computing hardware module. This module consists of three cascaded hardware circuits: the first stage is a CORDIC exponential calculation circuit, which directly calculates e^(current sequence parameter - healthy baseline sequence parameter) at the hardware level using a coordinate rotation digital calculation method; the second stage is a lookup table logarithmic calculation circuit, which calculates ln (steady-state baseline value) using a preset lookup table; the third stage is a 32-bit fixed-point difference calculation circuit, which calculates the final field deviation value H = e^(Δ sequence parameter) - ln (steady-state baseline value). The entire hardware module has a delay of no more than 50 microseconds from receiving the ternary state vector to outputting the H value.

[0019] The treatment execution layer integrates four types of execution modules: an electrical stimulation array (outputting electrical pulses of specific frequency and intensity through a microelectrode array), an ultrasonic transducer (outputting mechanical vibrations of specific frequency through focused ultrasound), a magnetic stimulation coil (generating an alternating magnetic field of specific intensity through electromagnetic induction), and a micro-drug pump (achieving precise delivery of drugs at the nanoliter level through a microfluidic chip). Based on the control commands output by the field-controlled treatment layer, the treatment execution layer selects and activates the corresponding execution module to apply precise physical or chemical field stimulation to the target object.

[0020] Example 2: Closed-loop diagnosis and treatment process based on TCM syndrome differentiation Taking a patient with liver qi stagnation syndrome as an example, the closed-loop diagnosis and treatment process of the present invention is fully demonstrated.

[0021] In the multimodal perception acquisition process, the system collects multimodal data through four dimensions: visual inspection (acquiring facial and tongue images and extracting color, texture, and morphological features), auscultation (acquiring speech signals and extracting spectral, speech rate, and sound intensity features), inquiry (collecting symptom temporal relationships, lifestyle habits, and environmental factors through standardized questionnaires), and palpation (acquiring pulse waveforms through a pulse sensor and extracting features such as superficiality, depth, slowness, and rate). After standardized preprocessing, the collected data is output as a 256-dimensional multimodal feature vector.

[0022] In the ternary state space modeling step, the pre-trained encoder maps the 256-dimensional feature vector to a ternary state vector. The output is as follows: Manifested extreme state value = 0.65 (liver qi stagnation, obstructed outward flow, corresponding to a "full syndrome" tendency in traditional Chinese medicine), Returning to storage extreme state value = 0.35 (insufficient yin and blood nourishment, corresponding to a "deficiency syndrome" tendency), and the order parameter of the transformation axis = 0.42 (abnormally high weight of wood element, imbalance of the five elements). At the same time, the system extracts an individualized health baseline vector based on the patient's historical health data (baseline values ​​of physiological indicators from multiple physical examinations): Health Manifested Extreme = 0.50, Health Returning to Storage Extreme = 0.50, and Health Order Parameter = 0.25.

[0023] In the field deviation calculation step, the field balance controller subtracts the healthy baseline sequence parameter (0.25) from the current sequence parameter (0.42) to obtain the Δ sequence parameter = 0.17; substituting this into a single composite operation: H = e^(0.17) - ln(1.0) ≈ 1.185 - 0 = 1.185 (where the steady-state baseline value is preset to 1.0, and ln(1.0) = 0). Since the H value is positive and exceeds the preset threshold (15% of H_safe, where H_safe = 0.15), the system is determined to be in an overexposed state.

[0024] In the regulation command generation step, the field balance controller generates a regulation command to enhance the inhibition signal of the Gui Zang pole based on the judgment result that the H value is positive and exceeds the threshold. The specific parameters of the regulation command include: acupoint selection (Taichong, Xingjian, and Qimen, these three acupoints are key acupoints of the liver meridian, and stimulation can adjust the weight of the wood element), stimulation mode (a specific combination of electrical pulses with a pulse width of 90μs and a frequency of 130Hz to inhibit chronic liver qi hyperactivity), and upper limit of current intensity (not exceeding 3.5V to ensure safety).

[0025] During the treatment process, the treatment execution layer activates the electrical stimulation array, which outputs specified electrical stimulation pulses to the selected acupoints via microelectrodes. Simultaneously, a micro-drug pump delivers a traditional Chinese medicine preparation (an extract of the modified Chaihu Shugan San formula, containing 12g of Bupleurum, 9g of Cyperus, 9g of Citrus aurantium, 12g of Paeonia lactiflora, 9g of Ligusticum chuanxiong, and 6g of Glycyrrhiza uralensis) at a specified dose.

[0026] In the closed-loop iterative process, patient multimodal data was recollected 7 days after treatment. The ternary state space modeling layer updated the state vector: explicit pole = 0.58, hidden pole = 0.42, and order parameter = 0.35. The field balance controller recalculated the H value: H = e^(0.35-0.25) - ln(1.0) ≈ 1.105 - 0 = 1.105, which is lower than the pre-treatment value of 1.185, but has not yet returned to steady state. The system automatically reduced the set amplitude of the stimulus intensity and continued treatment. After another 7 days of treatment, the order parameter was reassessed, and H = e^(0.28-0.25) - ln(1.0) ≈ 1.030 - 0 = 1.030. At this time, the H value was close to the threshold, and the system further reduced the stimulus intensity, entering the maintenance regulation phase. When the H value drops to within ±0.15, the system is considered to have returned to a healthy steady state, and treatment enters the closed-loop monitoring and maintenance phase.

[0027] Example 3: Closed-loop field regulation for tissue regeneration and repair This system is applied to tissue regeneration and repair after myocardial infarction. The multimodal sensing acquisition layer monitors the mechanical stiffness (expressed as Young's modulus E, in kPa) of the myocardial tissue in the infarcted area in real time using ultrasound elastography.

[0028] The ternary state space modeling layer maps mechanical stiffness values ​​and related biological indicators into ternary state vectors: the manifest extreme state value corresponds to collagen deposition density (fibrosis tendency), the storage extreme state value corresponds to the concentration of stem cell resting state markers (regeneration potential), and the transport sequence parameter corresponds to the tissue stiffness value E.

[0029] Under normal circumstances, the stiffness of healthy myocardial tissue is approximately 5-10 kPa. After a myocardial infarction, the stiffness of the infarcted area can rise to 20-50 kPa within weeks, indicating that fibrotic scarring is forming. When the field-modulated treatment layer detects that the tissue stiffness exceeds a preset "soft" benchmark (e.g., E_benchmark = 12 kPa), it calculates: H = e^(measured E - 12) - ln(1.0). When E = 25 kPa, H = e^13 - 0 ≈ 442413, far exceeding the threshold.

[0030] The field balance controller immediately generates control commands, driving the treatment execution layer to apply a specific pattern of microcurrent stimulation (frequency 10 Hz, intensity 0.5 mA / cm²) to the infarct area. This specific stimulation frequency and intensity inhibits excessive activation of fibroblasts and excessive collagen deposition, while promoting the proliferation and differentiation of endogenous cardiac stem cells. Throughout the treatment, the system continuously monitors changes in tissue stiffness and dynamically adjusts the stimulation parameters until the tissue stiffness returns to its normal steady-state range.

[0031] The treatment strategy in this embodiment mimics the biological function of the HAPLN1 protein in maintaining the "softness" of the tissue's mechanical field—restoring the regenerative homeostasis of the tissue microenvironment through external regulation, rather than directly replacing the cell's inherent repair procedures. This embodies the core concept of this invention: the goal of treatment is to restore the system's own dynamic balance, rather than forcefully combating symptoms externally. Beneficial effects

[0032] For the first time, the holistic concept of "syndrome differentiation and treatment" in traditional Chinese medicine and the reductionist technique of "precision targeting" in Western medicine are unified under the mathematical framework of "three-dimensional state space" and "field regulation", providing a unified technical platform for the integration of traditional Chinese and Western medicine.

[0033] By connecting diagnosis and treatment into a continuous quantitative closed loop, the system no longer outputs isolated diagnostic labels, but continuously tracks the changing trajectory of the ternary state vector and dynamically generates individualized control strategies based on the offset direction and magnitude.

[0034] The goal of treatment is to restore the overall dynamic balance of the system. By using the principle of "minimum effective intervention" (triggering the system-level self-zeroing with minimal external energy), the risks of "overtreatment" or "treatment dependence" in traditional therapies can be avoided.

[0035] The system has good adaptability and scalability. It does not require changes to the architecture when facing different target objects (humans, animals, ex vivo organs, organoids, etc.). Only the default values ​​of the sensing method, reference vector and stimulation parameters need to be adjusted according to the characteristics of the object. Attached Figure Description

[0036] Figure 1 Overall architecture diagram of a closed-loop medical system Figure 2 Flowchart of ternary state space modeling Figure 3 Field balance controller hardware structure diagram Figure 4 Closed-loop control flowchart.

Claims

1. A closed-loop medical method based on ternary state space and field control, characterized in that, Includes the following steps: S1: Multimodal perception acquisition steps—collect physiological, psychological, and environmental data of the target object and convert them into standardized multidimensional feature vectors; S2: Triadic state space modeling step - Input the multidimensional feature vector into the pre-trained encoder, map it into a ternary state vector containing explicit extreme state values, hidden extreme state values ​​and the axial order parameter of the transport, and construct an individualized health state space. S3: Field Deviation Calculation Step - The ternary state vector is input into the field balance controller. The controller calculates the field deviation value between the current ternary state vector and the preset health benchmark vector through a single composite operation consisting of exponential operation, logarithmic operation and difference operation. S4: Regulation instruction generation step - When the field deviation value exceeds the preset threshold, the field balance controller generates a regulation instruction, which is used to drive the treatment execution module to apply precise physical or chemical field stimulation to the target object. S5: Closed-loop iteration step - Repeat S1 to S4 until the ternary state vector returns to the preset steady-state range of the health benchmark vector.

2. The method according to claim 1, characterized in that, In step S2, the encoder compresses the multidimensional feature vector into a yin-yang binary basis vector, then expands it into a five-element state vector containing the weights of the five elements, and finally fuses it into the ternary state vector.

3. The method according to claim 1, characterized in that, The single composite operation in step S3 uses the following function: Field deviation value H = e^(current sequence parameter - health baseline sequence parameter) - ln(steady-state baseline value), where the health baseline sequence parameter is an individualized baseline value extracted from the historical health data of the target object, and the steady-state baseline value is a preset system constant.

4. The method according to claim 3, characterized in that, When the field deviation value H is positive and exceeds a preset threshold, it indicates that the manifest pole is overactive, and the field balance controller generates a control command to enhance the suppression signal of the homing pole; when the field deviation value H is negative and its absolute value exceeds a preset threshold, it indicates that the homing pole is overactive, and the field balance controller generates a control command to enhance the activation signal of the manifest pole.

5. The method according to claim 1, characterized in that, The physical field stimulation in step S4 includes at least one of electrical stimulation, magnetic stimulation, ultrasonic stimulation, light stimulation, or mechanical stimulation of a specific frequency and intensity; the chemical field stimulation includes the precise delivery of traditional Chinese medicine prescriptions, chemical drugs, or biological agents.

6. A closed-loop medical system based on ternary state space and field control, used to execute the method according to any one of claims 1 to 5, characterized in that, include: The multimodal sensing and acquisition layer is used to collect physiological, psychological, and environmental data of the target object. The ternary state space modeling layer is connected to the multimodal perception and acquisition layer and has a built-in pre-trained encoder to map low-dimensional acquired data into ternary state vectors and construct an individualized health state space. The field-controlled treatment layer is connected to the ternary state space modeling layer. It has a built-in single composite computing hardware module consisting of an exponential operation circuit, a logarithmic operation circuit, and a difference operation circuit, which is used to calculate the field deviation value and generate control commands. The treatment execution layer, connected to the field modulation treatment layer, is used to apply precise physical or chemical field stimulation to the target object according to the modulation instructions.

7. The system according to claim 6, characterized in that, In the single composite computing hardware module of the field-controlled treatment layer, the exponential computing circuit is implemented in hardware using a coordinate rotation digital calculation method, the logarithmic computing circuit is implemented in hardware using a lookup table, and the difference computing circuit is a 32-bit fixed-point computing circuit; the response period of the entire hardware module does not exceed 50 microseconds.

Images