Application of ketamine in treatment of severe depressive disorder

Abstract

The invention discloses an application of ketamine in treatment of severe depressive disorder. The method comprises the following steps: S1, continuously collecting and preprocessing physiological data and behavior data of a patient by using intelligent wearable equipment and a hospital monitoring system; s2, analyzing the data by adopting a time crystal theory, and identifying a hidden periodic mode; s3, in combination with Markov chain Monte Carlo and Hamiltonian Monte Carlo technologies, performing dynamic modeling to generate a personalized drug dose adjustment scheme; s4, further optimizing and adjusting the scheme through the graph convolutional network; s5, monitoring the reaction of the patient in real time, and dynamically feeding back and adjusting the drug dosage and the drug administration time; s6, a treatment scheme is evaluated and updated regularly, and a closed-loop personalized optimization process is formed; and S7, selecting an oral tablet, a nasal spray, a sublingual tablet or an injection form for administration according to the requirements of the patient. The traditional Chinese medicine composition remarkably improves the treatment effect, reduces side effects, and is high in adaptability and safety.

Description

Technical Field

[0001] The present invention relates to the technical field of medicine, and in particular to application of ketamine in treating major depressive disorder. Background Art

[0002] First, traditional ketamine treatment methods lack a personalized dosage adjustment mechanism. Each patient has significant differences in physiological responses and drug metabolism, resulting in the same dose of ketamine having different efficacy and side effects in different patients. In the prior art, fixed doses are usually used, which may result in some patients not being able to obtain the best treatment effect or risking drug overdose.

[0003] Secondly, existing ketamine treatment methods fail to make full use of modern technology, such as smart devices and data analysis tools, to monitor and adjust patients' physiological data in real time. Although ketamine's rapid onset is significant, the balance between its efficacy and side effects is very delicate. The lack of a dynamic monitoring and adjustment mechanism makes it difficult to optimize the drug dosage at any time during treatment, thereby increasing the risk of side effects.

[0004] Third, existing technologies lack in-depth analysis of patients’ physiological signals in the evaluation of ketamine’s therapeutic effects. For example, existing monitoring methods usually only focus on basic physiological indicators such as heart rate and blood pressure, while ignoring more complex physiological signal patterns, such as brain waves and heart rate variability. These complex signals often contain important information that can reveal changes in the patient’s physiological state, thus providing a basis for precise adjustment of drug dosage.

[0005] In summary, existing ketamine treatment methods have significant defects in personalized treatment plans, real-time monitoring and adjustment, and in-depth analysis of physiological signals. These deficiencies limit the widespread application of ketamine in the treatment of major depressive disorder and also increase the risk of side effects and drug dependence. Therefore, how to provide the application of ketamine in the treatment of major depressive disorder is an urgent problem to be solved by those skilled in the art. Summary of the invention

[0006] One purpose of the invention is to propose the application of ketamine in the treatment of major depressive disorder. The present invention makes full use of advanced technologies such as smart wearable devices, time crystal theory, Markov chain Monte Carlo simulation technology, graph convolutional network, etc., and describes in detail the scheme of personalized drug dosage adjustment and dynamic treatment optimization, which has the advantages of accurate treatment effect, strong real-time monitoring ability and low side effects.

[0007] The use of ketamine in treating major depressive disorder according to an embodiment of the present invention comprises the following steps:

[0008] S1. Use smart wearable devices and hospital monitoring systems to continuously collect patients’ physiological and behavioral data and perform data preprocessing.

[0009] S2. Use time crystal theory to perform periodic structure analysis on preprocessed physiological data to identify hidden periodic patterns in patients’ physiological signals, providing a basis for subsequent dynamic modeling;

[0010] S3. Based on the identified periodic patterns, combined with Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulation techniques, the patient's physiological response and drug dynamics are dynamically modeled to generate personalized drug dosage adjustment plans;

[0011] S4. Further optimize the drug dosage adjustment scheme output by the dynamic model through graph convolutional network;

[0012] S5. During the treatment process, the patient's physiological response is monitored in real time, and the Markov Chain Monte Carlo method is used to dynamically feedback and adjust the drug dosage and administration time;

[0013] S6. Regularly evaluate and update treatment plans based on patients’ real-time feedback and treatment effects to form a closed-loop personalized treatment optimization process to maximize treatment effects and reduce side effects;

[0014] S7. Ketamine is administered in the following forms: oral tablets or capsules, nasal spray, sublingual tablets, intravenous injection, subcutaneous injection or intramuscular injection. The form of administration is selected according to the individual needs of the patient.

[0015] Optionally, S2 includes the following steps:

[0016] S21, the preprocessed physiological data is represented as time series data {x(t)}, where x(t) represents the physiological signal data collected at time t;

[0017] S22. Perform a short-time Fourier transform on the time series data {x(t)} by converting the time series data into a time-frequency domain representation {X(t,f)}:

[0018]

[0019] Among them, w(τ-t) is the window function, f is the frequency variable, and τ is the integral variable;

[0020] S23. Based on the time crystal theory, construct the Hamiltonian based on the time-frequency domain representation {X(t,f)}

[0021]

[0022] Among them, Pi and Q i are conjugate momentum and conjugate coordinates respectively, i and j represent different modes in the frequency domain;

[0023] S24. Hamiltonian Perform quantization processing, through the quantization condition Determine the stability of the periodic pattern, where and is the quantized operator, is the reduced Planck constant, δ ij is the Kronecker delta function;

[0024] S25. Calculating Phase Space in the Frequency Domain The periodic changes in the phase space trajectory are used to identify the hidden periodic patterns {P k (t)}, where P k (t) represents the kth periodic mode at time t, and the topological structure of the phase space trajectory can be expressed by the following topological invariant:

[0025]

[0026] Among them, Γ k represents a closed trajectory in phase space, is the phase gradient, r is the position vector on the path;

[0027] S26, the identified periodic pattern {P k (t)} and its corresponding topological invariant π k Output, as input parameter for subsequent dynamic modeling.

[0028] Optionally, S3 includes the following steps:

[0029] S31, the identified periodic pattern {P k (t)} and its corresponding topological invariant π k Input into the Markov chain Monte Carlo model to initialize the model's state parameter X(t), where X(t) represents the patient's multidimensional physiological state vector at time t;

[0030] S32. Model the state transition of the system and define the state transition probability as P(X t+1 |X t ), where X t and X t+1 They represent the physiological state vectors at time t and t+1 respectively, and the transition probability is defined as:

[0031]

[0032] The system refers to a comprehensive computational model used to simulate changes in the patient's physiological state and pharmacokinetic responses. represents the change of Hamiltonian of state vector X at two moments, is the Hamiltonian corresponding to the physiological state vector X, k B is the Boltzmann constant, T is the system temperature;

[0033] S33. In the Markov chain Monte Carlo sampling process, the Hamiltonian Monte Carlo simulation technology is combined to dynamically model the patient's physiological state. The Hamiltonian Monte Carlo simulation simulates the state evolution of the system by solving the following Hamiltonian equation:

[0034]

[0035] Among them, X i is the physiological state component of the i-th dimension, p i is the corresponding conjugate momentum, The Hamiltonian of the system is:

[0036]

[0037] Among them, m i is related to physiological state X i The related virtual mass parameter, V(X), is the potential energy function associated with the physiological state vector X, representing the energy distribution in the patient's physiological state;

[0038] S34, through the Hamiltonian Monte Carlo sampling process, each step result is corrected, and the sampling step length ∈ and the total evolution time T are dynamically adjusted according to the individual differences of the patients:

[0039]

[0040] Among them, α and β are constants adjusted according to the individual characteristics of the patient, P k (t) is the kth periodic pattern at time t, p i and m i are the conjugate momentum and virtual mass parameter of the physiological state component of the i-th dimension, respectively;

[0041] S35. Combining the results of Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulations, a personalized drug dosage adjustment plan D(t) is generated for the patient, and the calculation formula is:

[0042]

[0043] Where D0 is the initial drug dose, λ is the drug dose adjustment coefficient, π k is the topological invariant of the kth periodic mode, P k(t) is the kth periodic pattern at time t, Indicates the energy distribution of the patient's current physiological state;

[0044] S36, outputting the generated personalized drug dosage adjustment plan D(t), and dynamically adjusting and optimizing it according to the patient's real-time physiological data and feedback during the actual treatment process;

[0045] Optionally, S4 includes the following steps:

[0046] S41. Based on the generated personalized drug dosage adjustment plan D(t), a multi-dimensional dynamic graph structure representation G(V,E) of the patient's physiological state is constructed, where:

[0047] The node set V represents the components of the patient's physiological state vector X(t), including the conjugate momentum p i , virtual mass parameter m i and drug dosage adjustment parameter D(t);

[0048] The edge set E represents the spatiotemporal associations between different nodes, and the edge weights are calculated by dynamic time warping, specifically:

[0049] w ij (t) = exp(-γ·DTW(X i (t),X j (t)));

[0050] Among them, γ is the control coefficient, DTW(X i (t),X j (t)) is the node v i and v j The dynamic time-warping distance at time t;

[0051] S42. Perform multi-layer spatiotemporal convolution operations on the node features in the graph convolutional network, and analyze the spatiotemporal associations in combination with the temporal convolutional network, focusing on extracting spatiotemporal features. The final representation of the node-level features is:

[0052]

[0053] in, is the adaptive weight of spatiotemporal association, is the weight matrix of the spatiotemporal convolution kernel, and δ is the time delay parameter;

[0054] S43, combined with the Hamiltonian Monte Carlo simulation results, the spatiotemporal characteristics and the periodic pattern of physiological signals P k (t) Fusion to generate graph-level global feature representation h G (t), the specific formula is:

[0055]

[0056] Among them, π k is the topological invariant of the kth periodic mode, P k (t) is the kth periodic pattern at time t, is the final feature representation of the node at time t;

[0057] S44. Representing h based on graph-level global features G (t) and the patient's real-time biofeedback signal {f b (t)}, the drug dosage is further adjusted, and the adjustment formula is:

[0058]

[0059] Where D(t) is the initial drug dosage adjustment plan, β is the feedback control coefficient, is the final feature representation of the i-th node, f b (t) is the biofeedback signal of the patient at time t;

[0060] S45. Output and deploy the optimized drug dosage adjustment plan D opt (t), combined with the patient's real-time physiological feedback for dynamic adaptive adjustment.

[0061] Optionally, S5 includes the following steps:

[0062] S51. During the treatment process, the patient's core physiological data X(t), including heart rate variability, brain waves and blood oxygen saturation, are collected in real time through intelligent monitoring equipment and constructed into a time series {X(t)};

[0063] S52, performing noise elimination and standardization processing on the time series data {X(t)} to generate a standardized physiological feature vector F(t);

[0064] S53, using the Markov chain Monte Carlo method to dynamically sample the standardized physiological feature vector F(t) and calculate the state transition probability P(X t+1 |X t ), the drug dose D(t) is based on the patient's current physiological state changes and the global feature representation h in claim 4 G (t) is adjusted, and the adjustment formula is:

[0065]

[0066] in, is the Hamiltonian change of the normalized eigenvector, k is a weighted sum that combines periodic patterns with global feature representations. Bis the Boltzmann constant, T is the system temperature;

[0067] S54, based on the patient's real-time feedback signal f b (t), further optimize the administration time t g and drug dose D(t), the adjustment formula is:

[0068]

[0069] Among them, Δt is the basic dosing time interval, α is the adjustment coefficient, and h G (t) is the graph-level global feature representation at time t, f b (t) is the patient’s real-time feedback signal;

[0070] S55. Combining the patient's real-time physiological features F(t) and the graph-level global feature representation h G (t) changes in drug dose D opt (t+1) and administration time t g Make further adjustments.

[0071] The beneficial effects of the present invention are:

[0072] (1) The present invention proposes the application of ketamine in the treatment of major depressive disorder. First, the patient's physiological data and behavioral data are continuously collected and preprocessed through smart wearable devices and hospital monitoring systems. This continuous data collection and preprocessing method can capture the patient's physiological state and behavioral changes in real time and comprehensively, providing reliable data support for subsequent treatment decisions. Combined with the time crystal theory, the periodic structure analysis of the preprocessed physiological data can effectively identify the hidden periodic patterns in the patient's physiological signals. These hidden patterns are often closely related to the patient's drug response, providing a scientific basis for personalized drug dosage adjustment, thereby significantly improving the accuracy and effectiveness of treatment.

[0073] (2) The present invention combines Markov chain Monte Carlo (MCMC) and Hamiltonian Monte Carlo (HMC) simulation techniques to dynamically model the patient's physiological responses and pharmacokinetic responses and generate personalized drug dosage adjustment plans. MCMC technology can effectively explore and capture individual differences in patients in complex physiological state spaces, while HMC simulation technology can accurately describe the patient's physiological responses and pharmacokinetic processes. This dynamic modeling method can not only reflect the patient's real-time status, but also predict the patient's future response, thereby providing an accurate and personalized drug dosage adjustment plan. In order to further optimize the treatment effect, the present invention introduces graph convolutional network (GCN) technology to optimize the generated drug dosage adjustment plan. GCN can utilize the complex correlations between physiological data to deeply optimize the treatment plan, thereby effectively reducing the occurrence of side effects and ensuring the safety of treatment and patient compliance.

[0074] (3) During the treatment process, the present invention monitors the patient's physiological response in real time and combines the MCMC method to dynamically feedback and adjust the drug dosage and administration time. This real-time feedback and adjustment mechanism can ensure that the drug dosage always matches the patient's current state, avoiding the risk of overdose or underdose. In addition, by dynamically analyzing the patient's real-time physiological data, the present invention can continuously optimize the treatment plan during the treatment process, forming a closed-loop personalized treatment optimization process. This closed-loop optimization can not only maximize the treatment effect, but also significantly reduce the drug dependence and side effects common in traditional ketamine treatment methods, further improving the safety and effectiveness of the treatment.

[0075] In summary, the present invention has significant advantages in improving therapeutic effects, reducing side effects and enhancing patient compliance, and opens up a new path for the application of ketamine in the treatment of major depressive disorder. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0077] Figure 1 A flow chart of the use of ketamine in the treatment of major depressive disorder proposed by the present invention;

[0078] Figure 2 A schematic diagram of a process for generating a personalized drug dosage adjustment plan by dynamic modeling based on Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulation techniques in the present invention;

[0079] Figure 3This is a structural schematic diagram of optimizing the drug dosage adjustment scheme output by the dynamic model through a graph convolutional network in the present invention. DETAILED DESCRIPTION

[0080] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0081] refer to Figure 1-3 The use of ketamine in the treatment of major depressive disorder includes the following steps:

[0082] S1. Use smart wearable devices and hospital monitoring systems to continuously collect patients’ physiological and behavioral data and perform data preprocessing.

[0083] S2. Use time crystal theory to perform periodic structure analysis on preprocessed physiological data to identify hidden periodic patterns in patients’ physiological signals, providing a basis for subsequent dynamic modeling;

[0084] S3. Based on the identified periodic patterns, combined with Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulation techniques, the patient's physiological response and drug dynamics are dynamically modeled to generate personalized drug dosage adjustment plans;

[0085] S4. Further optimize the drug dosage adjustment scheme output by the dynamic model through graph convolutional network;

[0086] S5. During the treatment process, the patient's physiological response is monitored in real time, and the Markov Chain Monte Carlo method is used to dynamically feedback and adjust the drug dosage and administration time;

[0087] S6. Regularly evaluate and update treatment plans based on patients’ real-time feedback and treatment effects to form a closed-loop personalized treatment optimization process to maximize treatment effects and reduce side effects;

[0088] S7. Ketamine is administered in the following forms: oral tablets or capsules, nasal spray, sublingual tablets, intravenous injection, subcutaneous injection or intramuscular injection. The form of administration is selected according to the individual needs of the patient.

[0089] In this implementation, S2 includes the following steps:

[0090] S21, the preprocessed physiological data is represented as time series data {x(t)}, where x(t) represents the physiological signal data collected at time t;

[0091] S22. Perform a short-time Fourier transform on the time series data {x(t)} by converting the time series data into a time-frequency domain representation {X(t,f)}:

[0092]

[0093] Among them, w(τ-t) is the window function, f is the frequency variable, and τ is the integral variable;

[0094] S23. Based on the time crystal theory, construct the Hamiltonian based on the time-frequency domain representation {X(t,f)}

[0095]

[0096] Among them, P i and Q i are conjugate momentum and conjugate coordinates respectively, i and j represent different modes in the frequency domain;

[0097] S24. Hamiltonian Perform quantization processing, through the quantization condition Determine the stability of the periodic pattern, where and is the quantized operator, is the reduced Planck constant, δ ij is the Kronecker delta function;

[0098] S25. Calculating Phase Space in the Frequency Domain The periodic changes in the phase space trajectory are used to identify the hidden periodic patterns {P k (t)}, where P k (t) represents the kth periodic mode at time t, and the topological structure of the phase space trajectory can be expressed by the following topological invariant:

[0099]

[0100] Among them, Γ k represents a closed trajectory in phase space, is the phase gradient, r is the position vector on the path;

[0101] S26, the identified periodic pattern {P k (t)} and its corresponding topological invariant π k Output, as input parameter for subsequent dynamic modeling.

[0102] In this implementation, S3 includes the following steps:

[0103] S31, the identified periodic pattern {Pk (t)} and its corresponding topological invariant π k Input into the Markov chain Monte Carlo model to initialize the model's state parameter X(t), where X(t) represents the patient's multidimensional physiological state vector at time t;

[0104] S32. Model the state transition of the system and define the state transition probability as P(X t+1 |X t ), where X t and X t+1 They represent the physiological state vectors at time t and t+1 respectively, and the transition probability is defined as:

[0105]

[0106] The system refers to a comprehensive computational model used to simulate changes in the patient's physiological state and pharmacokinetic responses. represents the change of Hamiltonian of state vector X at two moments, is the Hamiltonian corresponding to the physiological state vector X, k B is the Boltzmann constant, T is the system temperature;

[0107] S33. In the Markov chain Monte Carlo sampling process, the Hamiltonian Monte Carlo simulation technology is combined to dynamically model the patient's physiological state. The Hamiltonian Monte Carlo simulation simulates the state evolution of the system by solving the following Hamiltonian equation:

[0108]

[0109] Among them, X i is the physiological state component of the i-th dimension, p i is the corresponding conjugate momentum, The Hamiltonian of the system is:

[0110]

[0111] Among them, m i is related to physiological state X i The related virtual mass parameter, V(X), is the potential energy function associated with the physiological state vector X, representing the energy distribution in the patient's physiological state;

[0112] S34, through the Hamiltonian Monte Carlo sampling process, each step result is corrected, and the sampling step length ∈ and the total evolution time T are dynamically adjusted according to the individual differences of the patients:

[0113]

[0114] Among them, α and β are constants adjusted according to the individual characteristics of the patient, Pk (t) is the kth periodic pattern at time t, p i and m i are the conjugate momentum and virtual mass parameter of the physiological state component of the i-th dimension, respectively;

[0115] S35. Combining the results of Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulations, a personalized drug dosage adjustment plan D(t) is generated for the patient, and the calculation formula is:

[0116]

[0117] Where D0 is the initial drug dose, λ is the drug dose adjustment coefficient, π k is the topological invariant of the kth periodic mode, P k (t) is the kth periodic pattern at time t, Indicates the energy distribution of the patient's current physiological state;

[0118] S36, outputting the generated personalized drug dosage adjustment plan D(t), and dynamically adjusting and optimizing it according to the patient's real-time physiological data and feedback during the actual treatment process;

[0119] In this implementation, S4 includes the following steps:

[0120] S41. Based on the generated personalized drug dosage adjustment plan D(t), a multi-dimensional dynamic graph structure representation G(V,E) of the patient's physiological state is constructed, where:

[0121] The node set V represents the components of the patient's physiological state vector X(t), including the conjugate momentum p i , virtual mass parameter m i and drug dosage adjustment parameter D(t);

[0122] The edge set E represents the spatiotemporal associations between different nodes, and the edge weights are calculated by dynamic time warping, specifically:

[0123] w ij (t) = exp(-γ·DTW(X i (t),X j (t)));

[0124] Among them, γ is the control coefficient, DTW(X i (t),X j (t)) is the node v i and v j The dynamic time-warping distance at time t;

[0125] S42. Perform multi-layer spatiotemporal convolution operations on the node features in the graph convolutional network, and analyze the spatiotemporal associations in combination with the temporal convolutional network, focusing on extracting spatiotemporal features. The final representation of the node-level features is:

[0126]

[0127] in, is the adaptive weight of spatiotemporal association, is the weight matrix of the spatiotemporal convolution kernel, and δ is the time delay parameter;

[0128] S43, combined with the Hamiltonian Monte Carlo simulation results, the spatiotemporal characteristics and the periodic pattern of physiological signals P k (t) Fusion to generate graph-level global feature representation h G (t), the specific formula is:

[0129]

[0130] Among them, π k is the topological invariant of the kth periodic mode, P k (t) is the kth periodic pattern at time t, is the final feature representation of the node at time t;

[0131] S44. Representing h based on graph-level global features G (t) and the patient's real-time biofeedback signal {f b (t)}, the drug dosage is further adjusted, and the adjustment formula is:

[0132]

[0133] Where D(t) is the initial drug dosage adjustment plan, β is the feedback control coefficient, is the final feature representation of the i-th node, f b (t) is the biofeedback signal of the patient at time t;

[0134] S45. Output and deploy the optimized drug dosage adjustment plan D opt (t), combined with the patient's real-time physiological feedback for dynamic adaptive adjustment.

[0135] In this implementation, S5 includes the following steps:

[0136] S51. During the treatment process, the patient's core physiological data X(t), including heart rate variability, brain waves and blood oxygen saturation, are collected in real time through intelligent monitoring equipment and constructed into a time series {X(t)};

[0137] S52, performing noise elimination and standardization processing on the time series data {X(t)} to generate a standardized physiological feature vector F(t);

[0138] S53, using the Markov chain Monte Carlo method to dynamically sample the standardized physiological feature vector F(t) and calculate the state transition probability P(X t+1 |X t ), the drug dose D(t) is based on the patient's current physiological state changes and the global feature representation h in claim 4 G (t) is adjusted, and the adjustment formula is:

[0139]

[0140] in, is the Hamiltonian change of the normalized eigenvector, k is a weighted sum that combines periodic patterns with global feature representations. B is the Boltzmann constant, T is the system temperature;

[0141] S54, based on the patient's real-time feedback signal f b (t), further optimize the administration time t g and drug dose D(t), the adjustment formula is:

[0142]

[0143] Among them, Δt is the basic dosing time interval, α is the adjustment coefficient, and h G (t) is the graph-level global feature representation at time t, f b (t) is the patient’s real-time feedback signal;

[0144] S55. Combining the patient's real-time physiological features F(t) and the graph-level global feature representation h G (t) changes in drug dose D opt (t+1) and administration time t g Make further adjustments.

[0145] Embodiment 1:

[0146] Patient Li, 45 years old, male, suffered from major depression for 5 years. After multiple traditional treatments (including the use of SSRIs and tricyclic antidepressants), the symptoms were not significantly relieved, and increased tolerance to the drugs and aggravated side effects occurred. Li's main symptoms included persistent depression, severe sleep disorders, loss of appetite, inability to concentrate, and recurring suicidal thoughts. Due to the severity of his condition, Li was referred to a large psychiatric hospital and was included in a pilot project using the personalized ketamine treatment regimen proposed by the present invention.

[0147] On the first day after Li was admitted to the hospital, the medical team gave him a smart wearable device that can monitor Li's heart rate, blood oxygen saturation, brain waves (EEG) and other physiological data in real time. In addition, the hospital's monitoring system also records Li's behavioral data, such as activity level, sleeping posture, and facial expression changes, through cameras and motion sensors. After collection, these data will be transmitted to the hospital's central data processing system for preprocessing, removing noise and outliers, and standardizing the data.

[0148] On the second day of Li's hospitalization, the hospital's data analysis team began to conduct in-depth analysis of the pre-processed physiological data. The time crystal theory was used to analyze the periodic structure of Li's physiological signals and identify the periodic patterns hidden in the data. Data analysis showed that there was an obvious periodic correlation between Li's heart rate variability and his mood swings, while EEG data revealed the potential physiological mechanism of his sleep disorder. Based on these analysis results, the medical team developed a preliminary ketamine dosing plan for Li.

[0149] During the subsequent treatment, Li received a low-dose intravenous injection of ketamine once a day. After each injection, the smart device and hospital monitoring system will record Li's physiological reactions in real time and feed the data back to the central data processing system. Using Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulation techniques, Li's physiological reactions and drug metabolism are dynamically modeled to generate a personalized drug dosage adjustment plan. In order to further optimize the treatment effect, the system also uses graph convolutional network technology to deeply optimize the drug dosage adjustment plan to ensure that the drug dosage can match Li's physiological state at any time, maximize the treatment effect and reduce side effects.

[0150] Data proves:

[0151] During the one-month treatment, Li's condition improved significantly. The following are the specific changes in Li's physiological and behavioral data before, during and after treatment:

[0152] Table 1

[0153]

[0154]

[0155] During the treatment, Li's heart rate variability gradually increased from 45ms to 65ms, indicating a significant improvement in the function of the autonomic nervous system; sleep time increased from 4 hours to 7.5 hours, indicating a significant improvement in sleep quality; anxiety and depression scores dropped from 9 and 10 points to 2 points, respectively, showing a significant improvement in mental state; the frequency of suicidal thoughts gradually decreased from 5 times a day to completely disappear. These data show that the treatment method of the present invention has a significant effect in improving the physiological and psychological state of patients with major depression.

[0156] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. The use of ketamine in treating major depressive disorder, characterized in that: The steps include: S1. Use smart wearable devices and hospital monitoring systems to continuously collect patients’ physiological and behavioral data and perform data preprocessing. S2. Use time crystal theory to perform periodic structure analysis on preprocessed physiological data to identify hidden periodic patterns in patients’ physiological signals, providing a basis for subsequent dynamic modeling; S3. Based on the identified periodic patterns, combined with Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulation techniques, the patient's physiological response and drug dynamics are dynamically modeled to generate personalized drug dosage adjustment plans; S4. Further optimize the drug dosage adjustment scheme output by the dynamic model through graph convolutional network; S5. During the treatment process, the patient's physiological response is monitored in real time, and the Markov Chain Monte Carlo method is used to dynamically feedback and adjust the drug dosage and administration time; S6. Regularly evaluate and update treatment plans based on patients’ real-time feedback and treatment effects to form a closed-loop personalized treatment optimization process to maximize treatment effects and reduce side effects; S7. Ketamine is administered in the following forms: oral tablets or capsules, nasal spray, sublingual tablets, intravenous injection, subcutaneous injection or intramuscular injection. The form of administration is selected according to the individual needs of the patient.

2. The use of ketamine in the treatment of major depressive disorder according to claim 1, characterized in that: The S2 comprises the following steps: S21, the preprocessed physiological data is represented as time series data {x(t)}, where x(t) represents the physiological signal data collected at time t; S22. Perform a short-time Fourier transform on the time series data {x(t)} by converting the time series data into a time-frequency domain representation {X(t,f)}: Among them, w(τ-t) is the window function, f is the frequency variable, and τ is the integral variable; S23. Based on the time crystal theory, construct the Hamiltonian based on the time-frequency domain representation {X(t,f)} Among them, P i and Q i are conjugate momentum and conjugate coordinates respectively, i and j represent different modes in the frequency domain; S24. Hamiltonian Perform quantization processing, through the quantization condition Determine the stability of the periodic pattern, where and is the quantized operator, is the reduced Planck constant, δ ij is the Kronecker delta function; S25. Calculating Phase Space in the Frequency Domain The periodic changes in the phase space trajectory are used to identify the hidden periodic patterns {P k (t)}, where P k (t) represents the kth periodic mode at time t, and the topological structure of the phase space trajectory can be expressed by the following topological invariant: Among them, Γ k represents a closed trajectory in phase space, is the phase gradient, r is the position vector on the path; S26, the identified periodic pattern {P k (t)} and its corresponding topological invariant π k Output, as input parameter for subsequent dynamic modeling.

3. The use of ketamine in the treatment of major depressive disorder according to claim 1, characterized in that: The S3 comprises the following steps: S31, the identified periodic pattern {P k (t)} and its corresponding topological invariant π k Input into the Markov chain Monte Carlo model to initialize the model's state parameter X(t), where X(t) represents the patient's multidimensional physiological state vector at time t; S32. Model the state transition of the system and define the state transition probability as P(X t+1 |X t ), where X t and X t+1 They represent the physiological state vectors at time t and t+1 respectively, and the transition probability is defined as: The system refers to a comprehensive computational model used to simulate changes in the patient's physiological state and pharmacokinetic responses. represents the change of Hamiltonian of state vector X at two moments, is the Hamiltonian corresponding to the physiological state vector X, k B is the Boltzmann constant, T is the system temperature; S33. In the Markov chain Monte Carlo sampling process, the Hamiltonian Monte Carlo simulation technology is combined to dynamically model the patient's physiological state. The Hamiltonian Monte Carlo simulation simulates the state evolution of the system by solving the following Hamiltonian equation: Among them, X i is the physiological state component of the i-th dimension, p i is the corresponding conjugate momentum, The Hamiltonian of the system is: Among them, m i is related to physiological state X i The related virtual mass parameter, V(X), is the potential energy function associated with the physiological state vector X, representing the energy distribution in the patient's physiological state; S34, through the Hamiltonian Monte Carlo sampling process, each step result is corrected, and the sampling step length ∈ and the total evolution time T are dynamically adjusted according to the individual differences of the patients: Among them, α and β are constants adjusted according to the individual characteristics of the patient, P k (t) is the kth periodic pattern at time t, p i and m i are the conjugate momentum and virtual mass parameter of the physiological state component of the i-th dimension, respectively; S35. Combining the results of Markov Chain Monte Carlo and Hamiltonian Monte Carlo simulations, a personalized drug dosage adjustment plan D(t) is generated for the patient, and the calculation formula is: Where D0 is the initial drug dose, λ is the drug dose adjustment coefficient, π k is the topological invariant of the kth periodic mode, P k (t) is the kth periodic pattern at time t, Indicates the energy distribution of the patient's current physiological state; S36. Output the generated personalized drug dosage adjustment plan D(t), and dynamically adjust and optimize it according to the patient's real-time physiological data and feedback during the actual treatment process.

4. The use of ketamine in the treatment of major depressive disorder according to claim 1, characterized in that: The S4 comprises the following steps: S41. Based on the generated personalized drug dosage adjustment plan D(t), a multi-dimensional dynamic graph structure representation G(V,E) of the patient's physiological state is constructed, where: The node set V represents the components of the patient's physiological state vector X(t), including the conjugate momentum p i , virtual mass parameter m i and drug dosage adjustment parameter D(t); The edge set E represents the spatiotemporal associations between different nodes, and the edge weights are calculated by dynamic time warping, specifically: w ij (t)=exp(-γ·DTW(X i (t),X j (t))); Among them, γ is the control coefficient, DTW(X i (t),X j (t)) is the node v i and v j The dynamic time-warping distance at time t; S42. Perform multi-layer spatiotemporal convolution operations on the node features in the graph convolutional network, and analyze the spatiotemporal associations in combination with the temporal convolutional network, focusing on extracting spatiotemporal features. The final representation of the node-level features is: in, is the adaptive weight of spatiotemporal association, is the weight matrix of the spatiotemporal convolution kernel, and δ is the time delay parameter; S43, combined with the Hamiltonian Monte Carlo simulation results, the spatiotemporal characteristics and the periodic pattern of physiological signals P k (t) Fusion to generate graph-level global feature representation h G (t), the specific formula is: Among them, π k is the topological invariant of the kth periodic mode, P k (t) is the kth periodic pattern at time t, is the final feature representation of the node at time t; S44. Representing h based on graph-level global features G (t) and the patient's real-time biofeedback signal {f b (t)}, the drug dosage is further adjusted, and the adjustment formula is: Where D(t) is the initial drug dosage adjustment plan, β is the feedback control coefficient, is the final feature representation of the i-th node, f b (t) is the biofeedback signal of the patient at time t; S45. Output and deploy the optimized drug dosage adjustment plan D opt (t), combined with the patient's real-time physiological feedback for dynamic adaptive adjustment.

5. The use of ketamine in the treatment of major depressive disorder according to claim 1, characterized in that: The S5 comprises the following steps: S51. During the treatment process, the patient's core physiological data X(t), including heart rate variability, brain waves and blood oxygen saturation, are collected in real time through intelligent monitoring equipment and constructed into a time series {X(t)}; S52, performing noise elimination and standardization processing on the time series data {X(t)} to generate a standardized physiological feature vector F(t); S53, using the Markov chain Monte Carlo method to dynamically sample the standardized physiological feature vector F(t) and calculate the state transition probability P(X t+1 |X t ), the drug dose D(t) is based on the patient's current physiological state changes and the global feature representation h in claim 4 G (t) is adjusted, and the adjustment formula is: in, is the Hamiltonian change of the normalized eigenvector, k is a weighted sum that combines periodic patterns with global feature representations. B is the Boltzmann constant, T is the system temperature; S54, based on the patient's real-time feedback signal f b (t), further optimize the administration time t g and drug dose D(t), the adjustment formula is: Among them, Δt is the basic dosing time interval, α is the adjustment coefficient, and h G (t) is the graph-level global feature representation at time t, f b (t) is the patient’s real-time feedback signal; S55. Combining the patient's real-time physiological features F(t) and the graph-level global feature representation h G (t) changes in drug dose D opt (t+1) and administration time t g Make further adjustments.

Images