A device and method for home health screening based on odor detection

Abstract

The present application relates to the field of home health care, in particular to a device and method for home health preliminary screening based on odor detection. By analyzing the components of the exhaled gas of the subject, the body health condition is analyzed, and the treatment and prescription scheme are given. 101 is an array type gas sensor unit, 102 is the Nth gas sensing unit, 103 is a signal acquisition unit, and 104 is a calculation unit. In the clinical test stage, when collecting various different odor concentration values of the subjects, experienced doctors are invited to fill in the table 4 form, record the corresponding different gas concentration values under different diseases, evaluate and record the corresponding prescriptions for the health conditions of the subjects. After learning the relevant data of multiple subjects, a neural network model can be constructed to enable the model to correspond to the disease description filled in by the doctor according to the concentration values of multiple different gas sensors, realize the process of self-learning, and continuously optimize and improve the model to produce more accurate results.

Description

Technical Field

[0001] This invention relates to the field of home health care, and particularly to a home health screening device and method based on odor detection. By analyzing the composition of the breath of a subject, the device can determine the subject's health status and provide treatment and conditioning prescriptions. Background Technology

[0002] This invention utilizes a humanoid robot for home use combined with corresponding odor detection equipment. Leveraging self-learning and deep learning algorithms employed in AI technology, it accurately analyzes the human body's health status, providing early warning. When abnormal results are detected through exhaled breath analysis, patients should seek medical attention as soon as possible to prevent potential problems and avoid delays in treatment.

[0003] Humanoid robots are developing rapidly, and they have made great progress in performance, factory work, and conversation. However, they have not yet found a breakthrough in home care and have not been widely used.

[0004] This invention collects and analyzes human exhaled gases using an array of gas sensor devices, analyzes the concentration of each gas, and compares it with the health status of the subjects to find the correspondence between different diseases and the concentration of each gas.

[0005] The gas sensor acquisition device corresponding to this invention was placed on the nose of a humanoid robot. Based on AI deep learning algorithms, several subjects were initially diagnosed by experienced doctors, and the concentration values ​​corresponding to the array gas sensor were collected and recorded. Figure 4 The test form (concentration of 12 gases) is used in conjunction with the odor collection device of this invention to analyze the concentration data of the 12 gases for each subject. The analysis results are correlated with the content of the form filled out by an experienced doctor. After learning a certain amount of subject data, a neural network model is constructed through AI deep learning algorithm to find the correspondence between the concentration of the 12 gases and related diseases. Through continuous improvement of the model over a period of time, the optimal neural network model is obtained, thereby enabling the analysis of the concentration of the 12 gases to obtain the subject's condition description and corresponding prescription even without an experienced doctor. Summary of the Invention

[0006] The purpose of this invention is to provide a home health screening device and method based on odor detection. Even without an experienced doctor present, by analyzing the concentrations of 12 gas sensors and using an optimized neural network model, the health status of the subject can be determined, and a corresponding prescription can be provided.

[0007] To achieve the above objectives, the basic units of this invention include: a gas sensor array unit, which collects 12 different odor gases exhaled by the human body; a signal acquisition unit, which collects the exhaled gases of the subject in sequence; and a calculation unit, which analyzes the concentration and composition of the detected different gases, and determines the subject's health status and provides a corresponding prescription based on a model obtained by a deep learning algorithm.

[0008] The aforementioned gas sensor array unit employs an array of gas sensors, with the relevant gas sensors selected to be correlated with the disease. This allows for feedback on the severity of the disease based on changes in concentration.

[0009] The aforementioned signal acquisition unit employs a related circuit design to eliminate the influence of environmental interference factors and acquire the relevant gas concentration values ​​detected by the array gas sensor.

[0010] The aforementioned computing unit analyzes the concentration of relevant sensors detected, and uses a model obtained through deep learning algorithms to determine the different health conditions of the subjects and provide corresponding prescriptions.

[0011] This invention uses an automated device to obtain the composition and concentration of the subject's exhaled gas, enabling preliminary disease screening even without the presence of a professional and experienced doctor. It provides the subject's health status and corresponding prescriptions. When used in conjunction with a home care robot, it is equivalent to having an experienced doctor at home to conduct a health assessment of family members, thus playing a role in preventing health risks in advance.

[0012] This invention utilizes a gas sensor and corresponding acquisition device. During the learning phase, experienced doctors conduct clinical tests, record and analyze the data, construct a large model, and employ deep learning algorithms. By comparing the model's analysis results with those recorded by experienced doctors, the differences between the two are calculated. Then, a backpropagation neural network is used to reverse-correct the weights of each layer of the network. Through continuous accumulation of sample size, the model learns and its output analysis results become closer to those recorded by experienced TCM practitioners. After a period of self-learning and training, the optimal model is obtained. Subsequently, the health status of the subject can be analyzed based on the test results of the exhaled gas acquisition signal, and a corresponding prescription can be given.

[0013] The present invention proposes a home health screening device and method based on odor detection. Once the feasibility of the scheme is verified, it will be effectively promoted and used in home care. It can detect related health abnormalities of family members in advance, allowing them to go to the hospital for further treatment as early as possible and avoid delaying the condition. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the architecture of a home health screening device based on odor detection.

[0015] Figure 2 This is a schematic diagram illustrating the application of an odor sensor in an electronic nose.

[0016] Figure 3 This is a schematic diagram of the electronic nose sensor in operation.

[0017] Figure 4 It is a record form of the concentration test results of 12 gas sensors on the subjects by experienced doctors.

[0018] Figure 5 This is a schematic diagram of the BP neural network algorithm and the reverse correction of network node weights.

[0019] Figure 6 This is a schematic diagram illustrating the combined application of a home care humanoid robot and an odor-detecting home health screening device. Detailed Implementation

[0020] The following description describes embodiments of the present invention with reference to the accompanying drawings. These embodiments are illustrative examples and do not imply that the invention is limited to these specific embodiments. The 12 array sensors in the embodiments are for analytical purposes only; in practical applications, the number is not limited to 12. The present invention can be implemented in various ways as long as it does not depart from its essential principles.

[0021] Embodiment 1 of the invention

[0022] Figure 1 The diagram shows the architecture of a home health screening device based on odor detection. 101 is an array-type gas sensor unit, 102 is the Nth gas sensing unit, 103 is the signal acquisition unit, and 104 is the calculation unit. During the learning phase, different gas concentrations under 12 different conditions will be recorded. During the clinical trial phase, when collecting 12 different odor concentration values ​​from subjects, an experienced physician will simultaneously complete [the necessary procedures / reports]. Figure 4 The table records the concentration values ​​of 12 gases corresponding to different symptoms, assesses the health status of subjects, and records corresponding prescriptions. After learning from the data of 1000 subjects, a neural network model can be constructed. This model can then match the concentration values ​​of the 12 different gas sensors with the descriptions of symptoms filled in by doctors on the table, achieving a self-learning process. During the model refinement phase, a certain amount of clinical data is used to analyze the differences between the output of the large model and the results given by doctors. Then, the backpropagation algorithm is used to continuously optimize the model, and the correct weights of the neural network are obtained through deep learning algorithms. This allows the generated neural network model to correctly output the corresponding health status of subjects based on the concentrations of the multiple gas sensors proposed in this invention, and to provide corresponding treatment prescriptions for those with health risks.

[0023] Figure 2The diagram shows an application scenario of an odor sensor in an electronic nose. 201 shows the gas sensor and measurement system. The main components of human exhaled air are nitrogen, oxygen, and carbon dioxide. When there is an abnormality in health, there will be a rotten apple smell (acetone smell). For example, patients with diabetic ketoacidosis produce a large amount of acetone by breaking down fat in their bodies, and their exhaled air will smell like rotten apples. Patients with kidney failure or uremia cannot excrete metabolic waste through urine, and their exhaled air will have an ammonia smell. Gastrointestinal diseases will produce a putrid and fishy smell, causing bad breath.

[0024] Besides the odors produced by the body's own metabolic decomposition, inducing agents can also be used to assist in detection. C14 detection of Helicobacter pylori has been successfully applied in hospital clinical practice. In April 2026, Professor Jin Han and others from Shanghai Jiao Tong University published their research on "Induced Breath-Based Pan-Cancer Early Screening" in the internationally renowned scientific journal *Cell Biomaterials*. They were the first in the world to discover that a compound called "phenyl-β-D-glucuronide" can be used as an inducing agent to react with β-glucuronidase in the tumor microenvironment of the human body, releasing phenol gas, thereby achieving early warning of various cancers such as gastric cancer, lung cancer, liver cancer, and colon cancer, with a specificity and accuracy rate exceeding 94.3%. Through inducible methods, more disease-related odors can be discovered, and corresponding sensors can be designed for these produced biomarkers. Figure 202 shows an electronic nose system built using AI technology and simulating human anatomy. Based on an AI self-learning model, it can provide accurate diagnoses of diseases through continuous training and improvement. Figure 203 corresponds to the sensor sensitivity of the electronic nose system, and Figure 204 corresponds to the sensing unit, which generates specific signal outputs for specific odors, allowing the electronic nose system to receive the accurate concentration value of that odor. This electronic nose system is used for detecting human body odor in the home, enabling early detection of abnormal health conditions in family members and serving as an early health screening tool. Once an abnormality is detected, timely medical attention can be sought.

[0025] Figure 3 This is a schematic diagram of the electronic nose sensor. 301 is the sensor unit, which can determine the odor of the external environment. Odor conversion can be achieved through various means such as light, heat, humidity, temperature, pressure, and changes in gas-sensitive resistance. Information from the real world is collected and transmitted to the signal processing unit via electrical signals. 302 is the signal processing, calculation and storage unit, which collects the electrical signals of the corresponding channels of the array odor sensor to obtain the corresponding odor concentration value.

[0026] Figure 4This is a record form for experienced doctors to measure the concentration of 12 gases by sensors on subjects. The record form is used to record the concentration values ​​of the 12 gases obtained by the sensors corresponding to the subjects' exhaled breath, as well as the patient's disease status as determined by relevant hospital examination and test reports. The difference description column records the difference between the output results of the AI ​​neural network model and the conclusions drawn by doctors through clinical test results. Once a difference is found, the neural network model needs to be optimized to allow the model to learn and optimize again, so as to ensure that the neural network model is close to or consistent with the actual results obtained by doctors.

[0027] Figure 5 This is a schematic diagram of the BP neural network algorithm and the weights of the backpropagation network nodes. 501 is the input terminal of the neural network. In this invention, the concentration values ​​of different components detected by the array-type odor sensor in the subject's exhaled breath serve as the input variables for the neural network model. 502 is the output terminal of the multi-layer neural network model. In this invention, it is used to generate the subject's health status and corresponding prescription. 503 represents the patient's health status and corresponding prescription recorded by an experienced doctor and analyzed using existing hospital testing and laboratory equipment during model training. 504 represents the difference between the model output and the doctor's recorded results. Using a large AI model and NAS (Neural Architecture Search) technology, a multi-layer neural network model can be automatically generated. Through continuous training with hundreds or even thousands of actual subject data, a neural network model close to the optimal solution will eventually be generated. Subsequently, by inputting the concentration values ​​of different components of the subject in the array odor sensor, this optimal neural network model can output the subject's correct health status and corresponding prescription, thus enabling disease screening and analysis even without a doctor's presence.

[0028] Embodiment 2 of the invention

[0029] Figure 6 This is a schematic diagram illustrating the combined application of a home care humanoid robot and the odor detection device proposed in this invention. The odor detector proposed in this invention can be placed on the nose of the humanoid robot. When testing is required, the robot obtains the concentration values ​​of different components through a row of odor sensors inside the nose. 601 is the odor sensor part of the home care robot. 602 is the hand of the humanoid home care robot, and 603 is the foot of the humanoid home care robot.

[0030] Through continuous self-learning and optimization by the humanoid robot, a neural network model that closely approximates the true solution is eventually obtained. Once the optimized model is developed, the exhaled breath of the subject is detected to obtain relevant odor components and their corresponding concentration values. Then, using this neural network model, the subject's health status and corresponding prescription can be immediately obtained.

Claims

1. A novel odor detection method, characterized in that... It includes a gas sensor array unit, a signal acquisition unit, and a computing unit.

2. The gas sensor array unit according to claim 1, characterized in that... For different disease-related odors, sensors are designed to detect specific changes in light, electricity, heat, pressure, and gas-sensitive resistance as the concentration of the specific gas changes.

3. The signal acquisition unit according to claim 1, characterized in that... The electrical signals generated by the array sensor are filtered, denoised, and amplified to obtain electrical signals related to changes in odor concentration.

4. The computing unit according to claim 1. Its characteristics are... By analyzing and calculating the concentration values ​​of different components detected by the array of odor sensors, the health status of the subject is determined based on the model obtained by deep learning algorithm, and a corresponding prescription is given.

5. A novel odor detection device, characterized in that... It includes a gas sensor array unit, a signal acquisition unit, and a computing unit.

6. The pulse map acquisition unit according to claim 1, characterized in that... For different disease-related odors, sensors are designed to detect specific changes in light, electricity, heat, pressure, and gas-sensitive resistance as the concentration of the specific gas changes.

7. The signal acquisition unit according to claim 1, characterized in that... The electrical signals generated by the array sensor are filtered, denoised, and amplified to obtain electrical signals related to changes in odor concentration.

8. The computing unit according to claim 1. Its characteristics are... By analyzing and calculating the concentration values ​​of different components detected by the array of odor sensors, the health status of the subject is determined based on the model obtained by deep learning algorithm, and a corresponding prescription is given.

Images