Nasopharynx cancer rehabilitation system and method

Abstract

The embodiment of the invention provides a nasopharyngeal carcinoma rehabilitation system and method, the system comprises an information input port, a training data collection device, a processor and a display screen, the information input port is configured to obtain pre-input information, the training data collection device is configured to collect training data related to a training process of a patient, and the processor is configured to display the training data. The display screen is configured to display a training model so as to guide the patient to perform rehabilitation training of the current round; the processor is configured to determine rehabilitation training parameters based on pre-input information; generating a training model based on the rehabilitation training parameters; based on the training data, determining a training completion degree; and generating training prompt information in response to the condition that the training completion degree does not meet the preset completion degree.

Description

Technical Field

[0001] This specification relates to the field of cancer rehabilitation, and in particular to a nasopharyngeal carcinoma rehabilitation system and method. Background Art

[0002] Nasopharyngeal carcinoma is a common malignant tumor of the head and neck, and its treatment mainly relies on radiotherapy. Due to the large range of treatment exposure and the long treatment period, a large amount of healthy tissue in the body will be damaged during the treatment. Radiotherapy for nasopharyngeal carcinoma usually uses bilateral irradiation technology, and most or even all of the radiation dose is given from the left and right sides of the patient's head, which causes high-dose irradiation of the temporomandibular joint, which can easily lead to degenerative changes in the temporomandibular joint, and then cause atrophy and fibrosis of the masticatory muscles, as well as muscle spasms and pain, thereby inducing complications such as radiation-induced difficulty opening the mouth, dysarthria, and dysphagia.

[0003] Radiation-induced trismus (RIT) is the most common complication, occurring in 10% to 50% of patients. RIT patients experience dysmouth opening due to stiffness in the temporomandibular joint, fibrosis of the masticatory muscles, and pain. Severe dysmouth opening not only affects patients' normal eating and communication but also compromises their physical and mental health and quality of life.

[0004] Currently, rehabilitation training for complications such as difficulty opening the mouth primarily relies on verbal guidance from healthcare professionals, outpatient follow-up, and inpatient review. Studies have shown that patients require long-term adherence to professional mouth-opening exercises to achieve effective rehabilitation. However, patients often lack ongoing medical supervision at home, making it difficult to maintain standardized training. This leads to decreased compliance with rehabilitation and, in turn, compromises its effectiveness.

[0005] Therefore, we aim to propose a nasopharyngeal carcinoma rehabilitation system and method. By combining artificial intelligence with gaming, we can determine personalized rehabilitation training parameters and generate corresponding training models to guide patients through standardized rehabilitation exercises. Furthermore, based on training data, we can determine the degree of training completion, helping to ensure the effectiveness of rehabilitation training for patients. Summary of the Invention

[0006] One of the embodiments of the present specification provides a nasopharyngeal carcinoma rehabilitation system, which includes an information input port, a training data acquisition device, a processor and a display screen; the information input port is configured to obtain pre-input information, and the pre-input information includes patient characteristics and nasopharyngeal carcinoma diagnosis information; the training data acquisition device includes an imaging device and a sound sensor, and the training data acquisition device is configured to collect training data related to the patient's training process; the display screen is configured to display a training model to guide the patient to perform the current round of rehabilitation training; the processor is configured to: determine rehabilitation training parameters based on the pre-input information, and the rehabilitation training parameters include training movements and their training intensities; generate the training model based on the rehabilitation training parameters; determine the training completion based on the training data; and generate training prompt information in response to the training completion not meeting the preset completion.

[0007] One of the embodiments of the present specification provides a nasopharyngeal carcinoma rehabilitation method, which is executed by a processor and includes: determining rehabilitation training parameters based on pre-input information, the rehabilitation training parameters including training movements and their training intensities; generating a training model based on the rehabilitation training parameters; determining training completion based on training data; and generating training prompt information in response to the training completion not meeting a preset completion.

[0008] One of the embodiments of this specification provides a nasopharyngeal carcinoma rehabilitation device, which includes at least one processor and at least one memory; the at least one memory is used to store computer instructions; and the at least one processor is used to execute at least part of the computer instructions to implement the nasopharyngeal carcinoma rehabilitation method.

[0009] One embodiment of this specification provides a computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the nasopharyngeal carcinoma rehabilitation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein:

[0011] Figure 1 is a schematic diagram of a nasopharyngeal carcinoma rehabilitation system according to some embodiments of this specification;

[0012] Figure 2 is an exemplary flow chart of a nasopharyngeal carcinoma rehabilitation method according to some embodiments of this specification;

[0013] Figure 3is a schematic diagram of determining training completion according to some embodiments of this specification;

[0014] Figure 4 is a schematic diagram of determining sub-completion according to some embodiments of this specification;

[0015] Figure 5 This is a flowchart of updating a training model according to some embodiments of this specification. DETAILED DESCRIPTION

[0016] To more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly describes the drawings required for describing the embodiments. Obviously, the drawings described below are merely examples or embodiments of this specification. Those skilled in the art can apply this specification to other similar scenarios based on these drawings without inventive effort. Unless otherwise apparent from the context or otherwise noted, the same reference numerals in the figures represent the same structure or operation.

[0017] It should be understood that the terms "system," "device," "unit," and / or "module" used herein are a method for distinguishing different components, elements, parts, portions, or assemblies at different levels. However, if other terms can achieve the same purpose, the terms may be replaced by other expressions.

[0018] As used in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not refer to the singular but also include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0019] Flowcharts are used throughout this specification to illustrate the operations performed by systems according to embodiments of this specification. It should be understood that preceding or following operations do not necessarily need to be performed in exact order. Instead, the steps may be processed in reverse order or simultaneously. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0020] Nasopharyngeal carcinoma (NPC) is a malignant tumor that develops on the roof and sides of the nasopharynx. The tumor cells are poorly differentiated. Most NPC patients experience varying degrees of difficulty opening their mouths after radiotherapy, primarily related to the radiation dose to the temporomandibular joint, the total dose, and the surface dose.

[0021] Currently, there is no specific treatment for dysmouth opening, and the key lies in long-term prevention. Early functional exercise is one of the most effective and safest methods for preventing dysmouth opening after radiotherapy, reducing its severity and incidence. Long-term functional exercise can improve the flexibility and elasticity of the soft tissues of the temporomandibular joint, promote blood circulation and the absorption of inflammatory substances, and prevent tissue fibrosis or atrophy of fibrous tissue adhesions. It can effectively reduce the incidence of radiation-induced dysmouth opening in patients with nasopharyngeal carcinoma. Functional exercise includes facial exercises (e.g., tooth tapping, cheek puffing, smiling, tongue movements), chewing exercises (e.g., chewing gum), and neck exercises (e.g., lateral bending, rotation, and massage). Functional exercise primarily targets the temporomandibular joint, masticatory muscles, and neck muscles, helping to prevent muscle atrophy and joint sclerosis. Because radiation damage to the temporomandibular joint and masticatory muscles may not be apparent in the early stages of radiotherapy, patients may mistakenly believe they will not develop complications such as dysmouth opening and may forgo functional exercise. However, difficulty opening the mouth is a chronic radiation injury, requiring patients to adhere to regular and standardized functional exercises during and after radiotherapy. Therefore, we hope to propose a nasopharyngeal carcinoma rehabilitation system and method to help patients adhere to regular and standardized rehabilitation exercises and ensure the effectiveness of rehabilitation training.

[0022] Figure 1 1 is a schematic diagram of a nasopharyngeal carcinoma rehabilitation system according to some embodiments of this specification. The following describes in detail the application scenario 100 of the nasopharyngeal carcinoma rehabilitation system according to the embodiments of this specification. It should be noted that the following embodiments are intended only to explain this application and do not constitute a limitation of this application.

[0023] In some embodiments, as Figure 1 As shown, the application scenario 100 of the NPC rehabilitation system may include a NPC rehabilitation system 110 and a patient 120 . The NPC rehabilitation system 110 includes an information input port 111 , a training data acquisition device 112 , a display screen 113 and a processor 114 .

[0024] The nasopharyngeal carcinoma rehabilitation system 110 refers to a comprehensive system that uses intelligent elements to guide patients 120 to perform standardized rehabilitation exercises and reduce complications.

[0025] Patients 120 are individuals who receive rehabilitation training using the nasopharyngeal carcinoma rehabilitation system 110. For example, patients 120 include patients diagnosed with nasopharyngeal carcinoma who have received radiotherapy, patients who have developed complications of radiation damage, patients who have never participated in rehabilitation training, and patients who have participated in rehabilitation training intermittently.

[0026] Information input port 111 is an interface for receiving external information. For example, information input port 111 can be a data import interface at the software level, or an input device or data transmission interface at the hardware level. Hardware input devices can include keyboards, mice, touch screens, etc. Data transmission interfaces can include USB, Type-C data transmission interfaces, etc.

[0027] In some embodiments, the information input port 111 is configured to obtain pre-input information.

[0028] In some embodiments, the training data collection device 112 is configured to collect training data related to the training process of the patient 120 .

[0029] In some embodiments, the training data collection device 112 includes an imaging device, a sound sensor, etc.

[0030] In some embodiments, the training data acquisition device 112 may also include an airflow sensor, a motion sensor, etc. The airflow sensor is configured to obtain the patient's respiratory volume data. The motion sensor is configured to obtain the patient's motion characteristic data. For more information about respiratory volume data and motion characteristic data, please refer to Figure 4 and its related descriptions.

[0031] In some embodiments, the airflow sensor is deployed at the patient's respiratory area. For example, the airflow sensor is mounted on a device similar to a spirometer, and the patient wears it at the patient's respiratory area for training. The patient is instructed to breathe only through the worn area during training. Examples of airflow sensors include thermistor airflow sensors, differential pressure airflow sensors, and ultrasonic airflow sensors. The respiratory area refers to the channel or area in the human body where gas exchange occurs. Examples of respiratory areas include the oral cavity and nasal cavity.

[0032] Motion sensors include, but are not limited to, gyroscopes, accelerometers, and magnetic sensors, and are used to monitor minute movements of the training site, such as shaking. In some embodiments, the motion sensors are deployed at the training site. A training site refers to an area where specific muscle groups are trained and strengthened during rehabilitation training. For example, training sites include the masticatory muscles, neck muscles, facial muscles, pharyngeal constrictor muscles, soft palate, tongue muscles, cricopharyngeal muscles, and expiratory muscles.

[0033] The imaging device refers to a device used to collect motion data of the patient 120. For example, the imaging device includes a mobile phone camera, an infrared camera, a scanner, etc.

[0034] In some embodiments, the imaging device is configured to capture images of the patient 120 while performing training such as neck stretching training, mouth opening training, tongue extension training, mandibular joint movement, and tongue flicking.

[0035] The sound sensor refers to a device used to collect sound data of the patient 120. For example, the sound sensor includes a microphone, a sound pressure sensor, an ultrasonic sensor, and the like.

[0036] In some embodiments, the sound sensor may be worn around the patient's 120 mouth.

[0037] The display screen 113 is a device for displaying a training model to guide the patient 120 to perform rehabilitation training. For example, the display screen 113 includes a mobile phone screen, a tablet screen, etc.

[0038] In some embodiments, the display screen 113 may also be used to display ranking data of the patient 120. For more information about the ranking data, please refer to the relevant description below.

[0039] The processor 114 may process information and / or data related to the NPC rehabilitation system 110. In some embodiments, the processor 114 may include one or more processing engines (e.g., a single-chip processing engine or a multi-chip processing engine). By way of example only, the processor 114 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or any combination thereof.

[0040] In some embodiments, the processor 114 is configured to determine rehabilitation training parameters based on pre-input information; generate a training model based on the rehabilitation training parameters; determine the degree of training completion based on the training data; and generate a training reminder message in response to the degree of training completion not meeting the preset degree of completion. For more information about this embodiment, please refer to Figure 2 and its related descriptions.

[0041] In some embodiments, the processor 114 is further configured to determine, for each training sub-action, the sub-standard degree of the training sub-action based on the action data, the sub-model corresponding to the training sub-action, and the standard evaluation model; determine the sub-completion degree of the training sub-action based on the sub-standard degree; and determine the training completion degree based on the sub-completion degrees of multiple training sub-actions. For more information about this embodiment, please refer to Figure 3 and its related descriptions.

[0042] In some embodiments, the processor 114 is further configured to determine, for each training sub-movement, a sub-standard degree of the training sub-movement based on the movement data, the sub-model corresponding to the training sub-movement, and the standard evaluation model; determine a training load value for the patient 120 completing the training sub-movement based on the respiratory volume data and motion characteristic data of the patient 120 completing the training sub-movement; and determine a sub-completion degree based on the training load value and the sub-standard degree. For more information about this embodiment, please refer to Figure 4 and its related descriptions.

[0043] In some embodiments, the processor 114 is further configured to generate update parameters; based on the update parameters, update the rehabilitation training parameters to determine the updated rehabilitation training parameters; based on the updated rehabilitation training parameters, update the training model. For more information about this embodiment, please refer to Figure 5 and its related descriptions.

[0044] In some embodiments, the processor 114 is further configured to update the rehabilitation training parameters and the training model in response to the training completion degree meeting the preset completion degree, and display the updated training model through the display screen 113. For more information about this embodiment, please refer to Figure 2 and its related descriptions.

[0045] In some embodiments, as Figure 1 As shown, the nasopharyngeal carcinoma rehabilitation system 110 further includes a cloud 115 , and the processor 114 is further configured to upload training data and training completion to the cloud 115 in real time; and obtain ranking data of the patient 120 from the cloud 115 .

[0046] The cloud 115 may be a platform based on a server cluster, a distributed storage system, or other electronic devices capable of data storage, processing, and communication.

[0047] Ranking data refers to a sequential position obtained by comparing the performance of patient 120 using the nasopharyngeal carcinoma rehabilitation system 110 with other patients, indicating the relative position of patient 120 within the population. Ranking data can be a numerical ranking (e.g., 1st, 2nd, etc.), a percentage ranking (e.g., top 10%, top 20%, etc.), etc.

[0048] In some embodiments, the cloud 115 can determine the ranking data of the patient 120 in various ways based on the training data and the training completion degree. For example, the processor 114 uploads the training data and the training completion degree to the cloud 115 in real time. The cloud 115 can calculate the comprehensive score of the patient 120 according to the preset weights and generate the ranking data by sorting in descending order.

[0049] The comprehensive score refers to a comprehensive assessment of the overall training level or rehabilitation progress of the patient 120 based on multiple indicators. A higher comprehensive score indicates a better rehabilitation training effect for the patient 120.

[0050] In some embodiments of the present specification, by uploading training data and training completion to the cloud 115 in real time and combining them with game rankings, the rehabilitation training enthusiasm of patients 120 can be effectively improved, and interaction and competition among patients 120 can be promoted, thereby helping patients 120 to better adhere to rehabilitation training and improve rehabilitation effects.

[0051] For more information about the NPC Rehabilitation System 110 and its components, please refer to Figure 2-Figure 5 and its related descriptions.

[0052] It should be noted that the above description of the nasopharyngeal carcinoma rehabilitation system 110 and its components is for ease of description only and does not limit this specification to the exemplary embodiments described. It is understood that those skilled in the art, after understanding the principles of the system, may arbitrarily combine the modules or construct subsystems that connect to other modules without departing from these principles.

[0053] Figure 2 is an exemplary flow chart of the nasopharyngeal carcinoma rehabilitation method according to some embodiments of this specification. Figure 2 As shown, the process 200 includes the following steps: In some embodiments, the process 200 may be executed by a processor (such as the processor 114).

[0054] In some embodiments, the processor determines rehabilitation training parameters based on pre-input information, and the rehabilitation training parameters include training movements and their training intensity; generates a training model based on the rehabilitation training parameters; determines the training completion degree based on the training data; and generates training prompt information in response to the training completion degree not meeting the preset completion degree.

[0055] Step 210: Determine rehabilitation training parameters based on pre-input information.

[0056] Pre-input information refers to information used to generate rehabilitation training parameters. For example, pre-input information includes patient characteristics, nasopharyngeal carcinoma diagnosis information, etc.

[0057] Patient characteristics refer to parameters that describe individual patient attributes. For example, patient characteristics include body shape, age, gender, history of other concomitant illnesses, and number of rehabilitation training sessions. Body shape information, including body structure and the dimensions of the areas requiring rehabilitation training, can be obtained through 3D scanning.

[0058] Nasopharyngeal carcinoma diagnostic information refers to medical data related to the diagnosis of nasopharyngeal carcinoma. For example, this information includes tumor location, tumor size, tumor surgery information, and information on nasopharyngeal functional limitations. Tumor surgery information includes surgery time, tumor resection location, and incision size. Information on nasopharyngeal functional limitations includes limited mouth opening, difficulty swallowing, difficulty chewing, and neck stiffness.

[0059] The clinical manifestations of limited nasopharyngeal function typically begin with tightness and pain in the temporomandibular joint. As the disease progresses, the joint becomes less mobile and the distance between the incisors gradually decreases, ultimately leading to trismus, slurred speech, and difficulty eating.

[0060] The degree of limited mouth opening can be divided into five levels based on the distance between the incisors. For example, level 0 indicates that the degree of mouth opening remains unchanged compared to before radiotherapy, with the distance between the incisors greater than or equal to 3 cm; level 1 indicates limited mouth opening, with the distance between the incisors being 2.1 to 3.0 cm; level 2 indicates difficulty in eating dry food, with the distance between the incisors being 1.1 to 2.0 cm; level 3 indicates difficulty in eating soft food, with the distance between the incisors being 0.5 to 1.0 cm; and level 4 indicates that the distance between the incisors is less than 0.5 cm, indicating extreme difficulty in opening the mouth and even trismus.

[0061] In some embodiments, the processor can obtain pre-input information in a variety of ways. For example, a doctor can input pre-input information through an information input port (such as information input port 111), and the processor can communicate with the information input port to obtain the pre-input information. For another example, the processor can also import the patient's pre-input information from the hospital's electronic medical record system.

[0062] Rehabilitation training parameters refer to relevant parameters used to guide patients in the rehabilitation training process. In some embodiments, the rehabilitation training parameters include training movements and their training intensity.

[0063] Training movements refer to movements or activities designed to achieve rehabilitation goals. In some embodiments, training movements include muscle group movement training, voice training, etc.

[0064] Muscle group exercise training refers to activities or exercises performed on muscle groups to improve their strength, coordination, and flexibility. Muscle groups trained in muscle group exercise training include the masticatory muscles, neck muscles, facial muscles, pharyngeal constrictor muscles, soft palate, tongue muscles, cricopharyngeus muscles, and expiratory muscles.

[0065] Voice training refers to training methods used to improve the quality of the voice, reduce food regurgitation into the nasal cavity, and enhance swallowing ability.

[0066] In some embodiments, a training action includes multiple training sub-actions.

[0067] Training sub-movements refer to the individual movement units that make up a complete training movement. For example, if a set of training movements includes one swallowing movement and one breathing movement, then both the swallowing movement and the breathing movement are training sub-movements.

[0068] In some embodiments, the processor may perform functional analysis based on the training movements and utilize technical means such as video analysis, motion capture technology, and dynamics assessment to subdivide the overall training movement into several training sub-movements.

[0069] Training intensity refers to the quantitative representation of various physical stimulation parameters during a patient's rehabilitation training. For example, the intensity of muscle group movement training includes the range of motion, number of movements, and duration of the movement. The intensity of vocal training includes the volume, number of vocalizations, and duration of individual vocalizations.

[0070] In some embodiments, the processor can determine the rehabilitation training parameters by querying the training parameter library based on the pre-input information. For example, based on the patient's pre-input information, the processor searches the training parameter library for historical pre-input information whose similarity to the pre-input information meets the similarity condition as historical reference information. The processor filters out the historical rehabilitation training parameters corresponding to the historical reference information with the best effect (such as the highest training completion rate), and uses the historical rehabilitation training parameters as the rehabilitation training parameters for this time. The similarity condition can be set by default by medical technicians. The similarity can be the inverse of the vector distance between the pre-input information and the historical pre-input information. The vector distance can be determined based on cosine distance, etc.

[0071] The training parameter library refers to a database used to store the relationship between historical pre-input information and historical rehabilitation training parameters, and can be determined by medical technicians based on prior experiments.

[0072] In some embodiments, the training parameter library includes rehabilitation training parameters specified by medical technicians for patients corresponding to a variety of typical pre-input information.

[0073] In some embodiments, the training parameter library may further include historical rehabilitation training parameters that are updated based on the training completion degree of historical patients during the rehabilitation training process.

[0074] In some embodiments, each set of rehabilitation training parameters may include multiple sets of training movements and the training intensity of each set of training movements. For example, a set of rehabilitation training parameters may include training movement A (e.g., one swallowing movement, two mouth opening movements, one tongue extension movement, and one breathing movement), the training intensity of training movement A, and training movement B (e.g., two swallowing movements, three mouth opening movements), the training intensity of training movement B, etc.

[0075] Step 220: Generate a training model based on the rehabilitation training parameters.

[0076] A training model refers to a visual simulation model generated based on rehabilitation training parameters. The training model can have a simulation effect. Similar to an animated tutorial, the training model guides patients through the virtual movements presented by the training model, allowing them to perform rehabilitation training. For example, virtual movements can include shoulder shaking, head shaking, and facial movements. As an example, a virtual movement might include one swallow, two mouth openings, one tongue extension, and one exhalation.

[0077] In some embodiments, the processor can construct a training model based on the patient's characteristics and rehabilitation training parameters. To provide personalized training for the patient, the processor can obtain data obtained by scanning the patient (e.g., multi-angle photography, laser point cloud, etc.) and construct a three-dimensional virtual human based on this data using three-dimensional modeling technology. The virtual human performs the training tasks specified by the training model and displays them to the patient on a display screen (such as display screen 113). The patient can imitate the training tasks performed by the virtual human.

[0078] In some embodiments, the processor may also analyze the collected real-time training data of the patient, and drive the virtual human based on the analyzed data to display the patient's real-time training status through the virtual human.

[0079] In this scenario, the processor can interact with the patient through training models and virtual humans to enhance patient motivation.

[0080] Specifically, during rehabilitation training, a game involving catching fruit with the mouth can be used to train the patient's mouth-opening ability. When the patient performs the game, a virtual person also performs the same action. If the patient successfully eats the fruit, the virtual person also does so, providing instant feedback on the patient's training. The gamification of rehabilitation training exercises can make the training more engaging and improve patient engagement.

[0081] Specifically, the training model controls the virtual human to perform corresponding training movements by setting standard parameters, and provides interactive functions to guide patients in real-time feedback training. For example, during the training process, the patient is required to imitate the virtual human to complete specified movements (such as opening the mouth to at least 3cm). When the patient's actual movements do not meet the standards (such as opening the mouth to less than 3cm), visual feedback (such as showing falling fruits on the display screen) is used to remind the patient in real time that there are deviations in the movement. The aforementioned training mode that integrates parameterized standard setting, virtual demonstration and real-time interactive feedback can not only effectively improve the standardization and accuracy of the patient's movements, but also enhance the fun of the training process and the patient's compliance, thereby optimizing the overall rehabilitation training effect.

[0082] Step 230: Determine the degree of training completion based on the training data.

[0083] Training data refers to the data generated by the patient's movements during rehabilitation training. For example, training data includes movement data and vocalization data.

[0084] In some embodiments, the motion data of the training data may be acquired by an imaging device, and the sound data of the training data may be acquired by a sound sensor.

[0085] Training completion refers to the parameter used to assess the patient's completion of the training movements.

[0086] In some embodiments, the processor can determine the degree of training completion based on the training data in a variety of ways. For example, the processor can respectively calculate the ratio of movement amplitude and / or the ratio of sound volume between the training data and the training intensity in the corresponding rehabilitation training parameters. Based on the positive correlation between the degree of training completion and the ratio of movement amplitude and the ratio of sound volume, the degree of training completion of the training data is determined by the following formula (1). s=k1×a+k2×b (1)

[0087] Among them, s is the training completion degree, a is the movement amplitude ratio, b is the sound volume ratio, k1 and k2 are weight coefficients, which can be set by default by medical technicians.

[0088] For more information on how to determine training completion, see Figure 3 and its related descriptions.

[0089] In some embodiments, in response to the training data including motion data, the processor may calculate the ratio of the motion amplitude of the motion data to the motion amplitude of the training intensity in the rehabilitation training parameters, and use the ratio as the motion amplitude ratio. As an example, for a mouth opening movement, the processor may calculate the ratio of the patient's mouth opening size in the motion data to the mouth opening size in the rehabilitation training parameters to obtain the motion amplitude ratio.

[0090] In some embodiments, in response to the training data including vocalization data, the processor may calculate a ratio of an average volume of the vocalization data to a vocalization volume of the training intensity in the rehabilitation training parameters, and use the ratio as a vocalization volume ratio.

[0091] In some embodiments, k2 may also be negatively correlated with the patient's hoarseness.

[0092] In some embodiments of the present specification, when the patient's hoarseness is high, the k2 value decreases accordingly, which means that the evaluation focuses more on the accuracy of the movement amplitude rather than the volume. Therefore, when the patient has difficulty making a sound, the effect of muscle group movement training can still be effectively evaluated and strengthened.

[0093] Hoarseness is assessed as the degree to which the patient's voice is rough, discontinuous, or has abnormal pitch.

[0094] In some embodiments, the processor may determine the patient's level of hoarseness based on a hoarseness assessment model.

[0095] The hoarseness assessment model is a machine learning model, for example, a convolutional neural network (CNN) model.

[0096] In some embodiments, the input of the hoarseness assessment model is the patient's vocal data, and the output is the degree of hoarseness.

[0097] In some embodiments, a hoarseness assessment model can be trained based on a large number of first training samples with a first label. This is accomplished by inputting these first training samples with the first label into an initial hoarseness assessment model, determining a loss function value based on the first label and the results of the initial hoarseness assessment model, and iteratively updating the initial hoarseness assessment model based on the loss function value. Model training is completed when preset conditions are met, resulting in a trained hoarseness assessment model. The preset conditions may include convergence of the loss function or a threshold number of iterations.

[0098] In some embodiments, the first training sample for training the hoarseness assessment model may be a vocal data sample from historical sample data. The historical sample data may be composed of historical vocal data from different patients. The first label may be the actual hoarseness level. The actual hoarseness level may be the average of the hoarseness levels manually labeled by multiple doctors.

[0099] In some embodiments of this specification, a hoarseness assessment model can output an objective hoarseness score by inputting patient voice data. The training data consists of voice samples from a variety of patients, and its labels are based on the average hoarseness scores manually annotated by multiple doctors, ensuring the reliability of the assessment results.

[0100] In some embodiments, in response to the duration and number of exercises corresponding to the patient's training data not meeting the requirements of the rehabilitation training parameters, the processor continuously prompts the patient to continue to complete the actions corresponding to the rehabilitation training parameters.

[0101] Step 240 : In response to the training completion degree not meeting the preset completion degree, generate training prompt information.

[0102] The preset completion level is a numerical value used to assess whether a patient's performance during training meets the established standard. It can be set by medical technicians as a default.

[0103] Training prompts refer to instructive feedback generated by the processor based on the patient's performance during rehabilitation training. For example, in response to a training completion rate falling below a preset completion rate, the processor may generate a training prompt such as "Please repeat this set of exercises."

[0104] In some embodiments, in response to the training completion not meeting the preset completion level, the processor generates relevant training prompt information, and the patient continues training according to the current rehabilitation training parameters until the training completion level meets the preset completion level.

[0105] In some embodiments, in response to the number of times that the training completion degree does not meet a preset completion degree being greater than or equal to a threshold number, the processor determines that the current rehabilitation training parameters are difficult for the patient, and may then request a staff member to manually calibrate the current rehabilitation training parameters. The patient then performs subsequent rehabilitation training based on the manually calibrated rehabilitation training parameters.

[0106] In some embodiments, as Figure 2 As shown, the process 200 may further include step 250 .

[0107] Step 250 : In response to the training completion degree meeting the preset completion degree, the rehabilitation training parameters and the training model are updated, and the updated training model is displayed on the display screen.

[0108] In some embodiments, during rehabilitation training, the processor develops a corresponding rehabilitation plan for the patient, which includes rehabilitation training parameters for each round of the patient's entire rehabilitation training period. In response to the current round of training completing a predetermined degree of completion, the processor proceeds with the next round of rehabilitation training according to the rehabilitation plan, automatically updates the rehabilitation training parameters and training model, and displays the updated training model on the display screen. By way of example only, the processor may increase the difficulty of the current vocal training exercise, such as by increasing pronunciation speed or adding new phoneme exercises.

[0109] In some embodiments of the present specification, when the training completion meets the preset completion level, the processor can update the rehabilitation training parameters and training model to ensure that the training plan matches the patient's current ability, thereby helping the patient achieve better rehabilitation results in the next round of training.

[0110] In some embodiments, in order to adapt to the individual differences and rehabilitation needs of different patients, even if the patient's current training completion degree has met the preset completion degree, it is still possible to continue to use the current rehabilitation training parameters for the next rehabilitation training. For example, for older patients, considering their physical function and adaptability, it may be more appropriate to maintain the current rehabilitation training parameters in the later rehabilitation training process. For another example, when the patient wants to continue practicing the current rehabilitation training parameters again, the current rehabilitation training parameters and training model can also be maintained unchanged. At this time, the display screen will continue to display the current training model so that the patient can repeat the practice and strengthen the training effect of the current stage.

[0111] In some embodiments of this specification, the processor analyzes the patient's pre-entered information to determine rehabilitation training parameters, generate a targeted training model, and assess training completion based on the patient's training data. If training completion does not meet the preset completion threshold, the processor can provide timely training reminders to ensure that the patient receives the latest guidance information. This approach not only improves the targeted nature of rehabilitation training but also accelerates the patient's recovery process.

[0112] It should be noted that the above description of process 200 is for illustration and purpose only and does not limit the scope of application of this specification. Those skilled in the art may make various modifications and variations to process 200 under the guidance of this specification. However, such modifications and variations are still within the scope of this specification.

[0113] Figure 3 This is a schematic diagram of determining training completion according to some embodiments of this specification.

[0114] In some embodiments, the training data includes action data, the training action includes multiple training sub-actions, and the training model includes multiple sub-models. Figure 3 As shown, for each training sub-action, the processor determines a sub-standard degree 340 of the training sub-action based on the action data 310, the sub-model 320 corresponding to the training sub-action, and the standard evaluation model 330. Based on the sub-standard degree 340, the processor determines a sub-completion degree 350 of the training sub-action. The processor then determines a training completion degree 360 ​​based on the sub-completion degrees 350 of the multiple training sub-actions.

[0115] For more information about training data, action data, training actions, training models, training completion, and training sub-actions, see Figure 2 and its related descriptions.

[0116] Submodels are similar to training models, except that they are simulation models corresponding to training sub-movements, while training models are simulation models corresponding to training movements. For example, if the training model corresponds to one swallow and one breath, then submodel A corresponds to one swallow, and submodel B corresponds to one breath.

[0117] Substandard degree refers to the degree of standardization used to evaluate the patient's ability to complete the corresponding training sub-movements of the muscle group movement training.

[0118] In some embodiments, the processor can evaluate the sub-standard degree of the training sub-action while collecting training data. For example, the processor can directly compare the patient's training sub-action in the images and / or videos collected in real time by the imaging device with the corresponding standard sub-action displayed by the training model to determine the sub-standard degree of the training sub-action. During the comparison, the processor can overlap the key frames in the collected images and / or videos with the corresponding actions in the training model and use the overlap ratio as the sub-standard degree. The standard sub-action can be set by default by medical technicians.

[0119] A keyframe is a single frame in a video sequence that represents a specific action characteristic or state. Keyframes can include frames where the movement amplitude of a training sub-action reaches its maximum, and / or frames where the patient's training load value falls below a load threshold.

[0120] For more information about training load values ​​and how to obtain them, see Figure 4 and its related descriptions.

[0121] In some embodiments, the load threshold may be preset based on experience.

[0122] In some embodiments, the processor may further determine the load threshold based on a negative correlation between the load threshold and the patient's tumor surgery information (eg, incision size).

[0123] In some embodiments, the larger the patient's wound size, the less suitable it is for the patient to perform high-load training movements. Therefore, when evaluating the sub-criteria, more emphasis should be placed on analyzing the patient's movement performance in frames with lower training loads.

[0124] In some embodiments, the processor may determine the key frame in a variety of ways.

[0125] In some embodiments, the processor may use as a key frame the frame in which the patient's training sub-movement reaches its maximum amplitude. For example, when a patient is undergoing rehabilitation training, the display screen may require the patient to maintain each training sub-movement for a preset period of time based on rehabilitation training parameters. The processor may select as a key frame the frame in which the patient maintains the training sub-movement with the maximum amplitude.

[0126] In some embodiments, a patient performing a training sub-movement with a high training load value is relatively strenuous and therefore unrepresentative. Therefore, the processor uses frames where the patient's training load value is less than a load threshold as key frames to improve the accuracy of subsequent determination of the sub-standard degree.

[0127] The overlap ratio refers to the degree of conformity between multiple parts of the patient and the training model.

[0128] In some embodiments of this specification, the introduction of sub-standardization means that during rehabilitation training, patients must not only ensure that the range of motion meets established requirements, but also that each sub-movement is standardized. For example, when performing shoulder and neck stretching exercises, patients must maintain stability in other parts of the body to ensure the effectiveness of the rehabilitation training and make the assessment of training completion more accurate.

[0129] Subcompletion refers to the parameter used to assess the patient's completion of the training sub-movements.

[0130] In some embodiments, the processor may use the sub-standard degree as a sub-completion degree.

[0131] In some embodiments, the processor may further determine the sub-completion degree based on the training load value and the sub-standard degree. Figure 4 and its related descriptions.

[0132] In some embodiments, in response to a sub-completion degree of a training sub-action falling below a sub-completion degree threshold, the processor may modify the training model. For example, the processor may reduce the training intensity of the training sub-action, obtain real-time updated rehabilitation training parameters, obtain an updated training model, and perform subsequent rehabilitation training according to the updated training model.

[0133] As an example, in response to the original mouth opening action requiring the patient to open the mouth to 4 cm, if it is recognized that the patient can only open the mouth to 2 cm, the mouth opening amplitude is modified, or the number of training times for the training sub-action is reduced. The sub-completion threshold can be obtained by default.

[0134] In some embodiments, the training model is modified in real time and continuously during the patient's rehabilitation training.

[0135] In some embodiments, in response to a training sub-action being executed multiple times, the processor uses the average of multiple sub-completion degrees corresponding to the training sub-action as the sub-completion degree of the training sub-action. The processor then uses the statistical data of the sub-completion degrees corresponding to each training sub-action in the training action as the training completion degree. The statistical data may include the average, minimum, median, etc. of the sub-completion degrees corresponding to the multiple training sub-actions.

[0136] In some embodiments, the action data also includes multiple sub-collected data. For each training sub-action, the processor determines the sub-standard degree of the training sub-action based on the sub-collected data, sub-model, and standard evaluation model corresponding to the training sub-action.

[0137] Sub-collected data refers to data collected by the imaging device for the training sub-action. For example, when the training sub-data is a swallowing action, the sub-collected data is an image and / or video of the patient's swallowing action captured by the imaging device.

[0138] In some embodiments, the sub-collected data may be key frames corresponding to the training sub-actions.

[0139] In some embodiments, the standard evaluation model is a machine learning model, for example, a convolutional neural network (CNN) model.

[0140] In some embodiments, the input of the standard evaluation model includes sub-collected data and a sub-model, and the output is the sub-standard degree of the training sub-action.

[0141] In some embodiments, the standard evaluation model can be obtained by training based on a large number of second training samples with second labels. The training method of the standard evaluation model is similar to the training method of the hoarseness evaluation model, which can be seen in Figure 2 and its related descriptions.

[0142] In some embodiments, the second training sample for training the standard evaluation model can be a sub-collected data sample and a sub-model sample from the historical sample data. The second label is the actual sub-standard degree corresponding to the sub-collected data sample. The actual sub-standard degree can be obtained by scoring the sample sub-collected data by multiple medical technicians and taking the average value.

[0143] In some embodiments of this specification, compared with the previous use of the overlap ratio as the sub-standard, the average of the scores given by medical technicians to the sub-collected data is used as the sub-standard, and a standard evaluation model is trained. This can eliminate the influence of the sub-standard evaluation of some irrelevant parts and improve the accuracy of determining the sub-standard.

[0144] In some embodiments, the processor determines the training completion degree based on the sub-completion degrees corresponding to the plurality of training sub-actions and the training weights.

[0145] Training weights are numerical values ​​used to adjust or assign importance to different training sub-movements.

[0146] In some embodiments, the training weight is related to the patient's historical completion of multiple training sub-movements in their historical training records. For example, the higher the historical completion of a training sub-movement in the patient's most recent historical training record, the lower the training weight corresponding to the training sub-movement. This allows for more training of training sub-movements with lower completion rates, helping to comprehensively improve the patient's rehabilitation training level.

[0147] In some embodiments, the processor may perform weighted summation of the sub-completion degrees based on the training weights to obtain the training completion degree.

[0148] In some embodiments of the present specification, the processor comprehensively evaluates the training completion based on the sub-completion degrees and training weights of multiple training sub-movements, fully considering the patient's performance on each training sub-movement in historical training records, so that the allocation of training weights is more personalized, thereby being able to more comprehensively and objectively reflect the patient's actual rehabilitation progress and improve the rehabilitation effect.

[0149] In some embodiments of this specification, a processor processes the motion data of training sub-movements to determine sub-standard degrees and sub-completion degrees, and ultimately integrates multiple sub-completion degrees to determine the training completion degree. This not only improves the accuracy of training assessments but also ensures the effectiveness of training, helping patients more comprehensively recover their functions and thus enhancing overall rehabilitation outcomes.

[0150] Figure 4 is a schematic diagram of sub-completion levels according to some embodiments of this specification.

[0151] In some embodiments, the training data includes motion data, respiratory volume data, and motion feature data, the training motion includes multiple training sub-motions, and the training model includes multiple sub-models. Figure 4 As shown, for each training sub-action, the processor determines the sub-standard degree 340 of the training sub-action based on the action data 310, the sub-model 320 corresponding to the training sub-action, and the standard evaluation model 330; determines the training load value 430 of the patient completing the training sub-action based on the respiratory volume data 410 and motion characteristic data 420 of the patient completing the training sub-action; and determines the sub-completion degree 350 based on the training load value 430 and the sub-standard degree 340.

[0152] For more information about training data, action data, training sub-actions, and training models, see Figure 2 and its related descriptions.

[0153] For more information on sub-standards, sub-completions, and sub-models, see Figure 3 and its related descriptions.

[0154] In some embodiments, the training data may also include respiratory volume data and motion characteristic data. For more information about training data, see Figure 2 and its related descriptions.

[0155] Respiratory volume data is information used to quantify the amount of air a patient inhales and exhales during breathing. For example, respiratory volume data includes the volume of air a patient breathes, expressed in milliliters.

[0156] Motion characteristic data refers to information about the quality and characteristics of muscle group movements during rehabilitation training. For example, motion characteristic data includes the movement speed, number of tremors, and amplitude of the training area.

[0157] The training load value is used to characterize the degree of effort required for a patient to complete a corresponding action (e.g., a training sub-action, graded action, etc.). In some embodiments, although a patient completes a training sub-action, different training load values ​​correspond to different training outcomes, i.e., different sub-completion degrees. For example, the more rapid the patient's breathing, the smaller the single respiratory volume, or the more severe the motion sensor jitter, the higher the training load value, i.e., the greater the patient's effort, and the lower the sub-completion degree.

[0158] In some embodiments, the smaller the training load value, the better the training effect. That is, for the same training sub-movement, under the same sub-standard degree, the easier the patient completes (i.e., the smaller the training load value), the faster the patient's recovery progress between the two rehabilitation training sessions.

[0159] In some embodiments, the processor can determine the training load value by the following formula (2) based on the relationship that the training load value is positively correlated with the respiratory volume data, the number of jitters of the motion characteristic data, and the jitter amplitude of the motion characteristic data, and is negatively correlated with the movement speed of the motion characteristic data. p=k3×c+k4×d+k5×e-k6×f (2)

[0160] Among them, p is the training load value, c is the breathing volume data, d is the number of jitters in the motion characteristic data, e is the jitter amplitude in the motion characteristic data, f is the motion speed in the motion characteristic data, k3, k4, k5, and k6 are weight coefficients, which can be preset.

[0161] In some embodiments, when the training site is the respiratory muscles, the patient's respiratory volume data should be used as training data and does not need to be used to calculate the training load value.

[0162] In some embodiments, the processor determines a training load value for the patient to complete the training sub-movement based on a delay value for the patient to complete the training sub-movement.

[0163] In some embodiments, in response to the time length that the patient actually takes to complete the training sub-action for the first time being longer than the time length that the training model defines to complete the standard sub-action, the processor may use the difference between the time length that the training model defines to complete the standard sub-action and the time length that the patient actually takes to complete the training sub-action for the first time as a delay value.

[0164] In some embodiments, the processor may determine the training load value based on the delay value in a variety of ways. For example, the processor may determine the training load value based on a positive correlation between the delay value and the training load value.

[0165] In some embodiments of the present specification, by adopting a method of determining the training load value based on the delay value when the patient completes each training movement, the patient's reaction speed can be monitored and evaluated in real time, so that the training load value can be determined according to the patient's actual performance, ensuring that the training load value matches the patient's immediate ability, and helping the patient to perform rehabilitation training.

[0166] In some embodiments, the processor determines the training load value for the patient to complete the training sub-movement based on the respiratory volume data, motion characteristic data, and the delay value for the patient to complete the training sub-movement when the patient completes the training sub-movement.

[0167] In some embodiments, the processor can determine the training load value of the patient completing the training sub-action by querying a preset relationship table based on the patient's respiratory volume data, motion characteristic data, and delay value of the patient completing the training sub-action.

[0168] In some embodiments, the preset relationship table may include a correspondence between respiratory volume data, motion characteristic data, delay values, and training load values. In some embodiments, the preset relationship table may be determined based on historical data of respiratory volume data, motion characteristic data, and delay values ​​when the patient completes a training sub-movement.

[0169] In some embodiments, the processor may also determine the training load value in any other feasible manner, which is not limited here.

[0170] In some embodiments of the present specification, the processor comprehensively considers the patient's respiratory volume data, motion characteristic data, and the delay value for completing the training sub-movement to determine the training load value, which can more comprehensively evaluate the patient's performance and load in rehabilitation training, thereby effectively promoting the patient's functional recovery.

[0171] In some embodiments, the processor uses the average of the training load values ​​in the historical training records as the reference training load value. The processor determines the sub-completion degree using the following formula (3) based on the relationship that the sub-completion degree is positively correlated with the reference training load value and the sub-standard degree, and negatively correlated with the training load value.

[0172] Among them, r is the sub-completion degree, g is the reference training load value, p is the training load value, and t is the sub-standard degree.

[0173] In some embodiments, in response to the current round of rehabilitation training being the second or subsequent round of rehabilitation training, determining the sub-completion degree based on the training load value and the sub-standard degree includes: the processor divides the training sub-actions based on the grading parameters corresponding to the training sub-actions, and determines multiple graded actions; based on the grades of the multiple graded actions, determines the graded weight of each graded action; and determines the sub-completion degree based on the graded weight of each graded action, the sub-standard degree of the training sub-action, and the training load value.

[0174] Grading parameters refer to the ability to perform a training sub-movement, divided into multiple, gradually increasing, graded training sessions until the patient reaches the target range. For example, during rehabilitation training, if a patient has difficulty opening their mouth 5cm in one go, they can start with a 3cm mouth opening, gradually increase to a 4cm mouth opening, and finally reach a 5cm mouth opening, using grading parameters of 3cm, 4cm, and 5cm.

[0175] In some embodiments, each grading parameter has a corresponding grading action, and grading training is the training process corresponding to each grading action.

[0176] Grading means breaking down a single exercise into multiple movements of increasing difficulty until the desired range of motion is achieved. For example, in the example above, the graded movements would be: a 3cm mouth opening, a 4cm mouth opening, and a 5cm mouth opening.

[0177] In some embodiments, the processor divides the amplitude of the training sub-action according to the grading parameter to obtain multiple graded actions corresponding to the training sub-action.

[0178] In some embodiments, the grading parameters can be determined by medical technicians through a priori experiments.

[0179] In some embodiments, for each retraining sub-action, the processor may further determine the grading parameter corresponding to the retraining sub-action based on the training intensity and sub-standard degree corresponding to the retraining sub-action in the current round of rehabilitation training, the patient's historical review results, and the retraining difficulty through a grading model. For more information on how to determine the grading parameter, please refer to Figure 5 and its related descriptions.

[0180] Graded weights are the weight values ​​assigned to graded movements of different difficulty levels during rehabilitation training or sports training.

[0181] In some embodiments, the processor determines the grading weight of each grading action based on the levels of the multiple grading actions using various methods. For example, the higher the grading action, the higher the grading weight. As an example only, the grading weight can be assigned according to an arithmetic progression, such as grading into four levels with grading weights of 0.1, 0.2, 0.3, and 0.4, respectively.

[0182] Higher-level graded movements refer to those with higher training difficulty and intensity (such as a higher range of motion).

[0183] In some embodiments, the processor can determine the sub-completion degree based on the classification weight of each classified action, the sub-standard degree of the training sub-action, and the training load value in a variety of ways. For example, the processor can use the average value of multiple historical training load values ​​of the classified action in the historical data as the reference training load value of the classified action. The processor then calculates the classification completion degree of each classified action based on the sub-standard degree, the training load value, and the reference training load value using the above formula (3). The processor performs a weighted summation of the classification completion degrees of each classified action based on the classification weight, and then determines the sub-completion degree of the training sub-action.

[0184] Graded completion refers to the parameter that assesses how well the patient completes the graded actions.

[0185] In some embodiments of this specification, the processor subdivides the training sub-movement into multiple graded movements based on the graded parameters of the training sub-movements and determines the graded weights based on the grades of the graded movements. By considering the graded weights and combining the sub-standard degrees and the training load value to determine the sub-completion degrees, the processor can further improve the accuracy of the sub-completion degrees.

[0186] In some embodiments of this specification, the processor uses airflow and motion sensors to acquire the patient's respiratory volume and motion characteristic data. Leveraging the complementarity of these multiple data sources, this allows for a more comprehensive and accurate assessment of the training load values ​​for each training sub-movement. Determining sub-completion levels based on training load values ​​and sub-standard degrees effectively enhances the personalization and efficiency of rehabilitation training and helps more accurately monitor a patient's recovery progress.

[0187] Figure 5 This is a flowchart of updating a training model according to some embodiments of this specification.

[0188] In some embodiments, the processor updates the rehabilitation training parameters based on the updated parameters to determine updated rehabilitation training parameters; and updates the training model based on the updated rehabilitation training parameters.

[0189] For more information about rehabilitation training parameters and training models, please refer to Figure 2 and its related descriptions.

[0190] Step 510 : updating the rehabilitation training parameters based on the updated parameters to determine the updated rehabilitation training parameters.

[0191] Update parameters refer to parameters used to adjust rehabilitation training parameters. For example, update parameters include parameters for adjusting the recurrent training movements and their intensity.

[0192] Retraining exercises are exercises that require retraining. These exercises are exercises for which the patient has failed to meet the pre-set standard. The pre-set standard can be set by the medical technician as a default.

[0193] Intensity adjustment parameters refer to the parameters used to adjust the number of training sessions and duration of repetitive training exercises.

[0194] In some embodiments, the processor may generate updated parameters based on prior experience of medical technicians.

[0195] In some embodiments, the updated parameters include grading parameters corresponding to the next round of rehabilitation training. The processor selects a refresher sub-movement corresponding to the next round of rehabilitation training from the multiple training sub-movements based on the sub-standard degrees of the multiple training sub-movements. For each refresher sub-movement, the processor determines the grading parameter corresponding to the refresher sub-movement in the next round of rehabilitation training using a grading model based on the training intensity and sub-standard degree of the refresher sub-movement in the current round of rehabilitation training, the patient's historical review results, and the refresher training difficulty.

[0196] Recurrent training sub-movements refer to the independent movement units that constitute a complete recurrent training movement.

[0197] Historical review results refer to the results of a postoperative review of the patient's recovery. For example, historical review results include wound recovery and muscle recovery.

[0198] In some embodiments, the processor may obtain historical review results based on a cloud (such as cloud 115 ).

[0199] Recurrent training difficulty refers to the numerical value used to evaluate the difficulty of recurrent training movements.

[0200] In some embodiments, the processor may determine the retraining difficulty using the following formula (4) based on the positive correlation between the retraining difficulty and the amplitude and duration of the retraining action. q=m×n (4)

[0201] Among them, q is the difficulty of retraining, m is the amplitude of the retraining movement, and n is the duration of the movement.

[0202] In some embodiments, the retraining difficulty may also be obtained based on any other feasible method.

[0203] In some embodiments, the hierarchical model is a machine learning model, for example, a deep neural network (DNN) model.

[0204] In some embodiments, the inputs to the grading model include the intensity and substandards of the retraining sub-movements in the current round of rehabilitation training, the patient's historical review results, and the retraining difficulty level. The output is the grading parameters corresponding to the retraining sub-movements in the next round of rehabilitation training.

[0205] In some embodiments, the classification model can be obtained by training a large number of third training samples with third labels. The training method of the classification model is similar to the training method of the hoarseness assessment model, which can be seen in Figure 2 and its related descriptions.

[0206] In some embodiments, the third training sample for training the grading model can be the training intensity sample and substandard degree sample corresponding to the retraining sub-action sample in the historical sample data, or the historical review result sample and retraining difficulty sample of the patient sample. The third label is the grading parameter corresponding to the retraining sub-action sample in the subsequent round, which can be obtained through manual annotation. The subsequent round is the next round of rehabilitation training after the current round.

[0207] In some embodiments, the processor screens historical retraining records in which the training completion degree during rehabilitation retraining is higher than a preset completion degree; and uses the training intensity and sub-standard degree corresponding to the patient's historical retraining sub-movements, the patient's historical review results, and the retraining difficulty in each historical retraining record as historical sample data.

[0208] In some embodiments of this specification, the processor uses the sub-standards of multiple training sub-movements to determine the corresponding retraining sub-movements for the next round of rehabilitation training, ensuring the targeted and efficient rehabilitation training. For each retraining sub-movement, the processor uses a grading model to determine the grading parameters based on the training intensity, sub-standards, the patient's historical review results, and the retraining difficulty. This improves accuracy and facilitates dynamic adjustment of training difficulty based on the patient's actual progress.

[0209] In some embodiments, the update parameters further include training interval parameters corresponding to the retraining sub-action.

[0210] The training interval parameter refers to the time interval between two adjacent graded training sessions.

[0211] In some embodiments, the training interval parameter is positively correlated to the training load value when the patient completes the retraining sub-movement in the current round of rehabilitation training.

[0212] In some embodiments of this specification, a higher training load value indicates a greater physical strain on the patient. The processor will increase the training interval accordingly to ensure the patient has adequate rest and recovery time, thereby avoiding the risk of overtraining. Therefore, by introducing the training interval parameter, the safety and effectiveness of rehabilitation training can be improved.

[0213] In some embodiments, the updated rehabilitation training parameters are similar to the rehabilitation training parameters, which are obtained by adjusting the original rehabilitation training parameters according to the retraining adjustment parameters. Figure 2 and its related descriptions.

[0214] Step 520: Update the training model based on the updated rehabilitation training parameters.

[0215] In some embodiments, the processor will adjust the training model accordingly based on the updated rehabilitation training parameters to ensure that the training model can reflect the latest training requirements and the patient's rehabilitation status. Figure 2 Similarly, see Figure 2 and its related descriptions.

[0216] In some embodiments of the present specification, the processor can update the rehabilitation training program to adapt to the patient's progress and needs at different rehabilitation stages, ensuring the pertinence and effectiveness of the rehabilitation training.

[0217] It should be noted that the above description of process 500 is for illustration and purpose only and does not limit the scope of application of this specification. Those skilled in the art may make various modifications and variations to process 500 under the guidance of this specification. However, such modifications and variations are still within the scope of this specification.

[0218] One or more embodiments of this specification provide a nasopharyngeal carcinoma rehabilitation device. The device includes at least one processor and at least one memory, wherein the at least one memory is configured to store computer instructions; and the at least one processor is configured to execute at least some of the computer instructions to implement a nasopharyngeal carcinoma rehabilitation method.

[0219] One of one or more embodiments of the present specification provides a computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes a nasopharyngeal carcinoma rehabilitation method.

[0220] While the basic concepts have been described above, it will be apparent to those skilled in the art that the detailed disclosure is merely illustrative and does not limit this specification. Although not explicitly stated herein, various modifications, improvements, and revisions to this specification may be made by those skilled in the art. Such modifications, improvements, and revisions are suggested in this specification and remain within the spirit and scope of the exemplary embodiments of this specification.

[0221] This specification also uses specific terms to describe the embodiments of this specification. For example, "one embodiment," "an embodiment," and / or "some embodiments" refer to a feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that references to "one embodiment," "an embodiment," or "an alternative embodiment" two or more times in different locations in this specification do not necessarily refer to the same embodiment. Furthermore, certain features, structures, or characteristics of one or more embodiments of this specification may be appropriately combined.

[0222] In some embodiments, numbers are used to describe the quantity of components and attributes. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise stated, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the description and claims are approximate values, which may change according to the required characteristics of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the settings of such numerical values ​​are as accurate as possible within the feasible range.

[0223] Finally, it should be understood that the embodiments described in this specification are intended only to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly described and illustrated in this specification.

Claims

1. A nasopharyngeal carcinoma rehabilitation system, characterized in that: The system includes an information input port, a training data acquisition device, a processor and a display screen; The information input port is configured to obtain pre-input information, wherein the pre-input information includes patient characteristics and nasopharyngeal carcinoma diagnosis information; The training data acquisition device is configured to acquire training data related to the patient's training process; The display screen is configured to display a training model to guide the patient to perform a current round of rehabilitation training; The processor is configured to: Determining rehabilitation training parameters based on the pre-input information, the rehabilitation training parameters including training movements and training intensities; generating the training model based on the rehabilitation training parameters; Determining a degree of training completion based on the training data; In response to the training completion level not meeting a preset completion level, training prompt information is generated.

2. The nasopharyngeal carcinoma rehabilitation system according to claim 1, characterized in that: The training data includes action data, the training action includes multiple training sub-actions, and the training model includes multiple sub-models; The processor is further configured to: For each of the training sub-actions, Determining a substandard degree of the training sub-movement based on the movement data, the sub-model corresponding to the training sub-movement, and a standard evaluation model; Based on the sub-standard degree, the sub-completion degree of the training sub-action is determined; based on the sub-completion degrees of the multiple training sub-actions, the training completion degree is determined.

3. The nasopharyngeal carcinoma rehabilitation system according to claim 1, characterized in that: The training data includes motion data, respiratory volume data and motion feature data, the training motion includes multiple training sub-motions, and the training model includes multiple sub-models; The processor is further configured to: For each of the training sub-actions, Determining a substandard degree of the training sub-movement based on the movement data, the sub-model corresponding to the training sub-movement, and a standard evaluation model; determining a training load value for the patient to complete the training sub-movement based on the respiratory volume data and the motion characteristic data of the patient completing the training sub-movement; The sub-completion degree is determined based on the training load value and the sub-standard degree.

4. The nasopharyngeal carcinoma rehabilitation system according to claim 1, characterized in that: The processor is further configured to: updating the rehabilitation training parameters based on the updated parameters to determine updated rehabilitation training parameters; The training model is updated based on the updated rehabilitation training parameters.

5. The nasopharyngeal carcinoma rehabilitation system according to claim 1, characterized in that: The processor is further configured to: In response to the training completion degree satisfying the preset completion degree, the rehabilitation training parameters and the training model are updated, and the updated training model is displayed on the display screen.

6. A method for rehabilitation of nasopharyngeal carcinoma, characterized in that: The method is executed by a processor, and includes: Determining rehabilitation training parameters based on the pre-input information, wherein the rehabilitation training parameters include training movements and training intensities; generating a training model based on the rehabilitation training parameters; Determine the degree of training completion based on training data; In response to the training completion level not meeting a preset completion level, training prompt information is generated.

7. The nasopharyngeal carcinoma rehabilitation method according to claim 6, characterized in that: The training data includes action data, the training action includes multiple training sub-actions, the training model includes multiple sub-models, and the method further includes: For each of the training sub-actions, Determining a substandard degree of the training sub-movement based on the movement data, the sub-model corresponding to the training sub-movement, and a standard evaluation model; determining a sub-completion degree of the training sub-action based on the sub-standard degree; The training completion degree is determined based on the sub-completion degrees of the plurality of training sub-movements.

8. The nasopharyngeal carcinoma rehabilitation method according to claim 6, characterized in that: The training data includes motion data, respiratory volume data and motion feature data, the training motion includes multiple training sub-motions, and the training model includes multiple sub-models; The method further comprises: For each of the training sub-actions, Determining a substandard degree of the training sub-movement based on the movement data, the sub-model corresponding to the training sub-movement, and a standard evaluation model; determining a training load value for the patient to complete the training sub-movement based on the respiratory volume data and the motion characteristic data of the patient completing the training sub-movement; The sub-completion degree is determined based on the training load value and the sub-standard degree.

9. A nasopharyngeal carcinoma rehabilitation device, characterized in that: The apparatus comprises at least one processor and at least one memory; The at least one memory is for storing computer instructions; The at least one processor is configured to execute at least part of the computer instructions to implement the nasopharyngeal carcinoma rehabilitation method according to any one of claims 6 to 8.

10. A computer-readable storage medium storing computer instructions, wherein when a computer reads the computer instructions in the storage medium, the computer executes the nasopharyngeal carcinoma rehabilitation method according to any one of claims 6 to 8.

Images