WO2024006182A1 - Système et procédé de détection de maladie respiratoire - Google Patents

Système et procédé de détection de maladie respiratoire Download PDF

Info

Publication number
WO2024006182A1
WO2024006182A1 PCT/US2023/026187 US2023026187W WO2024006182A1 WO 2024006182 A1 WO2024006182 A1 WO 2024006182A1 US 2023026187 W US2023026187 W US 2023026187W WO 2024006182 A1 WO2024006182 A1 WO 2024006182A1
Authority
WO
WIPO (PCT)
Prior art keywords
respiratory disease
data set
predicted
symptoms
data
Prior art date
Application number
PCT/US2023/026187
Other languages
English (en)
Inventor
Laura STANLEY
Apostolos KALATZIS
Vishnunarayan Girishan PRABHU
Original Assignee
Immersive Reality Group, Llc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Immersive Reality Group, Llc filed Critical Immersive Reality Group, Llc
Publication of WO2024006182A1 publication Critical patent/WO2024006182A1/fr

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/20—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for electronic clinical trials or questionnaires
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
    • G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
    • G16H40/67—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for remote operation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/0002—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network
    • A61B5/0015—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network characterised by features of the telemetry system
    • A61B5/0022—Monitoring a patient using a global network, e.g. telephone networks, internet
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/0002—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network
    • A61B5/0015—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network characterised by features of the telemetry system
    • A61B5/0024—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network characterised by features of the telemetry system for multiple sensor units attached to the patient, e.g. using a body or personal area network
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7271—Specific aspects of physiological measurement analysis
    • A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • the disclosure provides, in one aspect, a method for predicting respiratory disease comprising: determining a first predicted respiratory disease with a machine learning model based on a physiological data set; determining a first confidence level of the first predicted respiratory disease; if the first confidence level is below a threshold confidence; initiating a symptom inquiry process and receiving a user input on symptoms; determining a second predicted respiratory disease with a rule-base algorithm based on the user input on symptoms; and determining a third predicted respiratory disease with the machine learning model based on the second predicted respiratory disease and the physiological data set; wherein a second confidence level of the second predicted respiratory disease is greater than the first confidence level.
  • the third predicted respiratory disease is bronchitis, acute upper respiratory infection, chronic airway obstructions, chronic obstructive asthma, pneumonia, influenza, or emphysema.
  • the method further comprises displaying the third predicted respiratory disease on a display.
  • the physiological data set and the user input on symptoms are collected at a medical facility.
  • the physiological data set is collected by a wireless sensor worn by a user.
  • the user input on symptoms is received by a user device.
  • the symptom inquiry process presents a subset of questions selected from a full set of questions to display to the user based on a k-means algorithm.
  • the symptom inquiry process is adaptive and determines the subset of questions based on a previous user response.
  • the subset of questions is based on relevance of symptoms to respiratory diseases, the specificity of the symptoms, and likelihood to occur in different conditions.
  • the disclosure provides, in one aspect, a system comprising: a wireless sensor configured to collect and transmit a physiological data set through a wireless communication channel; an input device configured to accept and transmit an observational data set through an input communication channel; a non-transitory data storage device connected through the wireless communication channel to the wireless sensor and connected through the input communication channel to the input device; and a computer including a processor and a non- transitory computer readable memory connected to the non-transitory data storage device.
  • the computer determines a predicted respiratory disease based on the physiological data set and the observational data set stored in the non-transitory data storage device; and transmits the predicted respiratory disease to an output device connected to the computer.
  • the wireless sensor collects the physiological data set in realtime.
  • the wireless sensor is part of a wearable device.
  • the physiological data set includes temperature, heart rate, respiration rate, oxygen saturation, diastolic blood pressure, systolic blood pressure, or any combination thereof.
  • the observational data set is received through the input device, where the input device is a smartphone, a tablet, a laptop, or a desktop computer.
  • the observational data set is entered by a person wearing the wireless sensor, a health care professional, or a third party.
  • the predicted respiratory disease is determined with an artificial intelligence model, wherein the artificial intelligence model includes a neural network, a rule-based system, a Bayesian network, or any combination thereof.
  • the predicted respiratory disease is one bronchitis, acute upper respiratory infection, chronic airway obstructions, chronic obstructive asthma, pneumonia, influenza, or emphysema.
  • FIG. 1 is a schematic of a system for predicting a respiratory disease.
  • FIG. 2 is a flow chart illustrating a method for predicting respiratory disease.
  • a concentration range is stated as 1% to 50%, it is intended that values such as 2% to 40%, 10% to 30%, or 1% to 3%, etc., are expressly enumerated in this specification. These are only examples of what is specifically intended, and all possible combinations of numerical values between and including the lowest value and the highest value enumerated are to be considered to be expressly stated in this disclosure.
  • processors and “central processing unit” or “CPU” are used interchangeably and refer to a device that is able to read a program from a computer memory (e.g., ROM or other computer memory) and perform a set of steps according to the program.
  • processor e.g., a microprocessor, a microcontroller, a processing unit, or other suitable programmable device
  • ALC arithmetic logic unit
  • registers e.g., a modified Harvard architecture, a von Neumann architecture, etc.
  • the processor is a microprocessor that can be configured to communicate in a stand-alone and/or a distributed environment, and can be configured to communicate via wired or wireless communications with other processors, where such one or more processor can be configured to operate on one or more processor-controlled devices that can be similar or different devices.
  • the term “memory” is any memory storage and is a non-transitory computer readable medium.
  • the memory can include, for example, a program storage area and the data storage area.
  • the program storage area and the data storage area can include combinations of different types of memory, such as a ROM, a RAM (e.g., DRAM, SDRAM, etc.), EEPROM, flash memory, a hard disk, a SD card, or other suitable magnetic, optical, physical, or electronic memory devices.
  • the processor can be connected to the memory and execute software instructions that are capable of being stored in a RAM of the memory (e.g., during execution), a ROM of the memory (e.g., on a generally permanent bases), or another non-transitory computer readable medium such as another memory or a disc.
  • the memory includes one or more processor-readable and accessible memory elements and/or components that can be internal to the processor-controlled device, external to the processor-controlled device, and can be accessed via a wired or wireless network.
  • Software included in the implementation of the methods disclosed herein can be stored in the memory.
  • the software includes, for example, firmware, one or more applications, program data, filters, rules, one or more program modules, and other executable instructions.
  • the processor can be configured to retrieve from the memory and execute, among other things, instructions related to the processes and methods described herein.
  • computer readable medium refers to any device or system for storing and providing information (e.g., data and instructions) to a computer processor.
  • Examples of computer readable media include, but are not limited to, DVDs, CDs, hard disk drives, magnetic tape and servers for streaming media over networks, whether local or distant (e.g., cloud-based).
  • Coupled is defined as “connected,” although not necessarily directly, and not necessarily mechanically.
  • the term coupled is to be understood to mean physically, magnetically, chemically, fluidly, electrically, or otherwise coupled, connected or linked and does not exclude the presence of intermediate elements between the coupled elements absent specific contrary language.
  • the term “in electronic communication” refers to electrical devices (e.g., computers, processors, etc.) that are configured to communicate with one another through direct or indirect signaling.
  • a computer configured to transmit (e.g., through cables, wires, infrared signals, telephone lines, airwaves, etc.) information to another computer or device, is in electronic communication with the other computer or device.
  • transmitting refers to the movement of information (e.g., data) from one location to another (e.g., from one device to another) using any suitable means.
  • network generally refers to any suitable electronic network including, but not limited to, a wide area network (“WAN”) (e.g., a TCP/IP based network), a local area network (“LAN”), a neighborhood area network (“NAN”), a home area network (“HAN”), or personal area network (“PAN”) employing any of a variety of communications protocols, such as Wi-Fi, Bluetooth, ZigBee, etc.
  • WAN wide area network
  • LAN local area network
  • NAN neighborhood area network
  • HAN home area network
  • PAN personal area network
  • the network is a cellular network, such as, for example, a Global System for Mobile Communications (“GSM”) network, a General Packet Radio Service (“GPRS”) network, an Evolution-Data Optimized (“EV-DO”) network, an Enhanced Data Rates for GSM Evolution (“EDGE”) network, a 3GSM network, a 4GSM network, a 5G New Radio, a Digital Enhanced Cordless Telecommunications (“DECT”) network, a digital AMPS (“IS- 136/TDMA”) network, or an Integrated Digital Enhanced Network (“iDEN”) network, etc.
  • GSM Global System for Mobile Communications
  • GPRS General Packet Radio Service
  • EV-DO Evolution-Data Optimized
  • EDGE Enhanced Data Rates for GSM Evolution
  • 3GSM Third Generation
  • 4GSM Third Generation
  • 5G New Radio a Digital Enhanced Cordless Telecommunications
  • DECT Digital Enhanced Cordless Telecommunications
  • IS- 136/TDMA digital AMPS
  • iDEN Integrated Digital Enhanced Network
  • a respiratory disease detection system described herein is a system wherein physiological and observational data are collected from one or more wireless sensor, one or more input devices, or some combination thereof. Data is communicated wirelessly to one or more non- transitory data storage devices. The physiological data is then preprocessed, which in some embodiments includes conversion to normalized values, removal of noise, deletion of extraneous information, and reformatting and organizing to facilitate analysis. The preprocessed data is then analyzed by artificial intelligence software executing on a computer, where such software may implement a neural network, rule-based system, Bayesian network, or the like. Results of the analysis may, in some embodiments be shared with a user (e.g., a physician, a patient, a care provider, etc.).
  • a user e.g., a physician, a patient, a care provider, etc.
  • the respiratory disease detection system may, in some embodiments, receive feedback on the accuracy of the analysis from the individual wearing or using the wireless sensor, health professional(s), and from algorithms assessing accuracy.
  • This feedback loop may, in some embodiments, be used to improve analysis for unique individuals, all individuals using the system, or both.
  • the respiratory disease detection system described herein detects the early onset of various respiratory diseases using one or more wireless sensors, wireless communication, and analysis using a novel data pipeline to preprocess both physiological and observed data which facilitates diagnosis using an artificially intelligent algorithm.
  • the integration of the wireless technology, a Remote Health Monitoring System (RHMS) and wireless communication allows automatic and ⁇ or continuous collection of a physiological data set as well as an observation data set (e.g., subjection data regarding aches, pains, cough, restricted breathing, etc.) without requiring an individual to visit a health care facility or meet with a health care professional.
  • the system is also useful when visiting a health care professional who can review all such data from some point in the past to the current time and more effectively diagnose a disease or condition.
  • the respiratory disease detection system includes a wireless physiological data collector, a wireless transfer of such data to a data repository, a pipeline transforming and sanitizing data, and artificial intelligence algorithm(s) used to detect anomalies related to diseases represent a novel system not readily apparent to individuals in the field.
  • the system extracts the physiological signals and symptomatic determinants for the detection of respiratory infectious diseases (acute respiratory infections (bronchitis, acute upper respiratory infection), chronic obstructive pulmonary disease (chronic airway obstruction, chronic obstructive asthma), pneumonia, influenza, emphysema), and then communicates the physiological data wirelessly to a data repository where the pipeline converts the data and the artificial intelligence algorithms analyze the data frequently, or even continuously, for the individual.
  • the system is capable, in some embodiments, to providing continuous monitoring of individuals using human-in-the-loop feedback for the early detection of respiratory diseases.
  • the respiratory disease detection system and example embodiments described herein provides a unique and inexpensive system and method for monitoring individuals.
  • the system enables 24*7 automated patient monitoring at home, freeing up more resources (beds and providers).
  • the system provides consistent or continuous monitoring, thus avoiding false alarms as well as timely detection of the onset of actual respiratory diseases, thereby both reducing unnecessary hospital visits and healthcare bills and avoiding delay in treatment of actual disease.
  • a respiratory disease detection system 10 is illustrated according to one embodiment.
  • the system 10 is shown as a closed loop once the patient or user begins data collection.
  • the sensors capture physiological data which is transferred to the secure cloud database.
  • observed data is also communicated to the secure cloud database through the wireless communication network, direct input, a wired connection, or some combination of one or more of these communication types.
  • the system 10 includes a wireless sensor 14 configured to collect and transmit a physiological data set through a wireless communication channel.
  • the wireless sensor 14 collects the physiological data set in real-time. In other embodiments, the wireless sensor 14 collects the physiological data set periodically.
  • the wireless sensor 14 is part of a wearable device (e.g., an FDA approved device, a smart watch, etc.).
  • the physiological data set includes temperature, heart rate, respiration rate, oxygen saturation, diastolic blood pressure, systolic blood pressure, or any combination thereof.
  • actual weight of the patient, as well as blood pressure, pulse rate, and other physiological data or vitals are captured through any number of wireless sensors (e.g., a blood pressure monitor 14A, a pulse oximeter 14B, a weight scale 14C, a blood glucose monitor 14D, etc.).
  • wireless sensors e.g., a blood pressure monitor 14A, a pulse oximeter 14B, a weight scale 14C, a blood glucose monitor 14D, etc.
  • the system 10 further includes an input device 18 configured to accept and transmit an observational data set (e.g., symptoms) through an input communication channel.
  • the observational data set (e.g., symptoms) is received through the input device 18, where the input device is a smartphone, a tablet, a laptop, or a desktop computer.
  • the observational data set is entered by a person wearing the wireless sensor, a health care professional, or a third party.
  • the integration of observed data improves the efficacy of disease prediction by the system 10.
  • the system 10 further includes a non-transitory data storage device 22 connected through the wireless communication channel to the wireless sensor 14 and connected through the input communication channel to the input device 18.
  • the system 10 further includes a computer 26 including a processor and a non-transitory computer readable memory connected to the non-transitory data storage device 22.
  • the non-transitory data storage 22 is a secure cloud database.
  • the physiological and observed data are stored in a secure and HIPPA compliant cloud database.
  • the system 10 accepts data sets as a stream of data as opposed to single, one off, time stamped values.
  • the data sets are converted, smoothed, and clarified by removal of noise, in the data pipeline and facilitate accurate and timely analysis.
  • the system 10 includes a data analytics engine to converts or otherwise changes the form of the data to make such data more amenable to analysis.
  • the analysis is based on machine learning and other statistical models. This conversion facilitates timely analysis through the machine learning software.
  • the pipeline in this example embodiment may normalize data, reconfigure for more rapid comparison or calculations, or combine data to better facilitate comparison to historical data used in the machine learning software.
  • the analysis provides results which may generate warnings or alerts to the user, patient, or medical professional(s). These results can also be analyzed visually to detect trends or anomalies, in some embodiments.
  • the computer 26 determines a predicted respiratory disease based on the physiological data set and the observational data set stored in the non-transitory data storage device 22.
  • the predicted respiratory disease is determined with an artificial intelligence model, wherein the artificial intelligence model includes a neural network, a rule-based system, a Bayesian network, or any combination thereof.
  • the predicted respiratory disease is one bronchitis, acute upper respiratory infection, chronic airway obstructions, chronic obstructive asthma, pneumonia, influenza, or emphysema.
  • the computer 26 also transmits the predicted respiratory disease to an output device 30 connected to the computer.
  • the predicted respiratory disease is transmitted to the output device 30 with a risk assessment, warning, or alert.
  • the output device 30 includes the feedback to the user of assessments, warnings, alerts, visualizations of the data, or any combination thereof.
  • Some data thresholds as well results may require more immediate attention (i.e overly rapid heart rate or very low blood oxygen level). Such data thresholds prompt this embodiment of the system to issue warnings or alerts 34 to the user, medical professional, or third-party, or more than one of these people.
  • the visualization allows a medical professional to detect anomalies or patterns that may not yet be detectable by the machine learning software or may be unique to the patient.
  • the system 10 includes Interactive Data Acquisition and Communication (ID AC) results evaluation query 38 for data acquisition and communication.
  • ID AC Interactive Data Acquisition and Communication
  • the system 10 has the ability to query data for review and communication to the user or third-parties. Results evaluation are used to review the closeness, mathematically or otherwise, to the actual diagnosis the machine learning software was able to provide. Data may also be extracted by query to review system performance.
  • the system 10 is utilized as a diagnostic tool (e.g., in a medical professional office). In other embodiments, the system 10 is utilized as a disease management tool (e.g., utilized by a user at home).
  • a method 50 for predicting respiratory disease comprises: (STEP 51) determining a first predicted respiratory disease with a machine learning model based on a physiological data set; and (STEP 52) determining a first confidence level of the first predicted respiratory disease. At (STEP 53) if the first confidence level is below a threshold confidence; the method 50 initiates a symptom inquiry process and receiving a user input on symptoms.
  • the method 50 further includes (STEP 54) determining a second predicted respiratory disease with a rule-base algorithm based on the user input on symptoms; and (STEP 55) determining a third predicted respiratory disease with the machine learning model based on the second predicted respiratory disease and the physiological data set.
  • the third predicted respiratory disease is bronchitis, acute upper respiratory infection, chronic airway obstructions, chronic obstructive asthma, pneumonia, influenza, or emphysema.
  • a second confidence level of the second predicted respiratory disease is greater than the first confidence level.
  • the rule-based algorithm prediction is utilized to improve the confidence level of the prediction from the machine learning model.
  • the method 50 is an iterative process that repeats until the confidence level of the predicted respiratory disease is above a threshold confidence.
  • the method further includes displaying the third predicted respiratory disease on a display.
  • the display is a physician display or a patient display.
  • the physiological data set and the user input on symptoms are collected at a medical facility.
  • the physiological data set is collected by a wireless sensor worn by a user.
  • the user input on symptoms is received by a user device.
  • the symptom inquiry process presents a subset of questions selected from a full set of questions to display to the user based on a k-means algorithm.
  • the symptom inquiry process is adaptive and determines the subset of questions based on a previous user response.
  • the subset of questions is based on relevance of symptoms to respiratory diseases, the specificity of the symptoms, and likelihood to occur in different conditions.
  • the method 50 provides a hybrid model of predicting respiratory disease that uses physiological data and observed data (e.g., symptoms) to predict respiratory disease.
  • physiological data e.g., symptoms
  • the combined use of physiological data and symptoms develops a robust model for predicting respiratory diseases.
  • the confidence level of a neural network is used to determine whether to incorporate the rule-based algorithm. If the neural network's confidence level is low, we initiate a symptom inquiry process using a question selection method explained in the question selection method section. The symptom inquiry process aims to gather additional information from the user regarding their symptoms. To accomplish this, a systematic question selection method that takes into account various factors, such as the relevance of symptoms to respiratory diseases, the specificity of the symptoms, and their likelihood to occur in different conditions.
  • the symptoms collected through the inquiry process are then added to the rule-based algorithm.
  • This algorithm utilizes a set of predefined rules and relationships between symptoms and diseases to predict respiratory conditions based on the symptom profile.
  • the output from the rule-based algorithm is then combined with the initial neural network model. This integration allows us to incorporate the knowledge and insights obtained from the rule-based approach back into the neural network. By doing so, a feedback loop is created where the neural network can learn from the rule-based predictions and refine its own predictions accordingly. The prediction process is repeated iteratively until high confidence levels are achieved.
  • a deep neural network leverages the knowledge gained from a pre-trained neural network on predicting respiratory diseases.
  • the pre-trained weights of the earlier network are used as the starting point for training the later network.
  • the new network is fine-tuned by taking a pre-trained network and continuing training on the new one with a smaller learning rate to update the weights for the new respiratory predictions. This can help the new network generalize better and perform better in predicting respiratory diseases.
  • Training an artificial neural network involved adjusting the weights and biases of the neurons in the network based on the input data and the desired output.
  • the neural network model has six inputs (e.g., temperature, heart rate, respiration rate, oxygen saturation, systolic blood pressure, and diastolic blood pressure) and seven outputs (e.g., acute respiratory infections (bronchitis, acute upper respiratory infection), COPD (chronic airway obstruction, chronic obstructive asthma), pneumonia, influenza, emphysema).
  • Training Data A database that includes patient vital signs, symptoms, and respiratory disease diagnosis was used to train the machine learning models. These data was matched with the data of the triage and vital sign tables. Additionally, the patients were matched to the symptomatic determinants (e.g., the self-reported severity level of cough, chills, shivering, muscle aches, etc.) for creating the dataset. Generating this dataset requires significant preprocessing that involves data transformation and cleaning by removing duplicates, nulls, and missing values such that it can be used as input to a machine-learning model.
  • symptomatic determinants e.g., the self-reported severity level of cough, chills, shivering, muscle aches, etc.
  • the training process includes a forward propagation step, calculation of error, and backward propagation.
  • part of the training process includes validation and testing, to determine the generalizability of the developed model to predict the disease on unseen data of a new patient. If the performance was not satisfactory, the training process was repeated with different hyperparameters, such as the learning rate, number of hidden layers, and number of neurons in each layer. 5. Rule-Based Algorithm Using Symptoms
  • a decision tree is utilized to recursively split the symptoms into subsets based on the values of the input features.
  • the algorithm aims to create a tree structure that can accurately predict respiratory disease for new data.
  • the decision tree algorithm uses a metric called information gain to determine which feature to split on at each node.
  • Information metric is defined in the following equation:
  • Entropy is measures through the following equation:
  • the feature with the highest information gain is selected as the best feature to split on.
  • H(S) is Entropy of dataset before splitting
  • T is the set of subsets created after splitting the dataset
  • P(t) is the proportion of the number of elements in each subset t after splitting
  • H(l) is Entropy of the subset t.
  • the data is split into subsets based on the values of that symptom.
  • a new child node is created for each subset, and the process is repeated recursively on each subset until the stopping criteria are met.
  • the stopping criteria is the maximum depth of the tree, a minimum number of samples per node, or a minimum information gain for each split.
  • the algorithm assigns a prediction for the respiratory disease based on the majority class of the training samples that reached that node. This prediction is used for new data points that follow the same path through the decision tree.
  • the rules are implicit in the decision tree structure, as each path from the root to a leaf node represents a set of conditions that need to be met to make a prediction. These conditions are based on the symptoms that were used to split the data at each node. Therefore, the decision tree algorithm decides the rules by selecting the best symptom to split on and creating the tree structure that accurately predicts respiratory disease.
  • the method disclosed herein predicts with a relatively small number of features (questions) data to reduce patient burden and consequently improve adherence to the RHM systems.
  • an adaptively ordering question system is used to improve outcomes while minimizing respondent burden.
  • this approach includes the implementation of a k-means algorithm. The algorithm first will specify k centroids, one for each cluster. Then, all data samples in the dataset will be assigned to the closest centroid. Then, the algorithm re-calculates the positions of the centroids based on the newly assigned samples in each cluster. The algorithm repeats the same process until it reaches a stable state (i.e. no further movement for the centroids).
  • k-means clustering is used to discover the structure of questions and group-related questions based on patients’ responses. Initially, the clustering algorithm will start by randomly selecting the ‘k’ of the questions as initial cluster centroids. Then, the k- means algorithm forms the clusters based on the similarity in responses in the respondent space (patients space). After grouping the questions into k clusters, the best representative of each cluster (each question group) will be selected as the closest question to the cluster centroid. Adaptive questions, where later questions depend on earlier responses, can potentially improve the response rate and compliance. The order of the questions plays a major role in the survey completion, and it is not necessary to collect the answers to all questions to make reasonable predictions.
  • the question selection method disclosed herein helps identify an optimal subset of original questions and reduce the dimensionality of feature space according to Silhouette score.
  • the Silhouette score will be used to determine if the number of clusters are a bad or a good pick for the given Questions.
  • the Silhouette Score is based on the Silhouette width, an indicator for the quality of each item i.
  • Silhouette Score close to +1 indicates that the sample is far away from the neighboring clusters.
  • the Silhouette score of 0 indicates that the sample is on or very close to the decision boundary between two neighboring clusters, and a negative score indicates that those samples might have been assigned to the wrong cluster.
  • the questions used in the questionnaire are mostly traditional questionnaires used by clinicians to diagnose a respiratory disease.
  • a novel part of the questionnaire is the heuristic used to adapt the questionnaire based on the patient feedback and medical history. The questions will aim to understand various factors, including shortness of breath (at rest, physical activity, etc.), cough (frequency, type), phlegm (color), difficulty climbing stairs, wheezing or whistling, smoking (quantity, frequency, etc.), cold, running nose along with several other factors.
  • minimum number of prominent features collected through a smartwatch are used to build a robust and capable model of learning from a limited number of features.
  • the algorithms are adjusted to utilize data from wearables (e.g., smartwatches, fitbit, apple watches).
  • the initial feature set is examined to understand their importance in the artificial neural network.
  • Features with high importance are used to develop another neural network that achieves the same or similar accuracy as the high-dimensional neural network.
  • the dimensionality of the data will increase in each iteration of data acquisition (because a new data element is collected that can be used to generate a set of new features).
  • a flexible predictive model advantageously performs classification on low-dimensional data, and then the dimensionality of the predictive model can be expanded if needed.
  • the lower-dimensional models can be easily built either by projecting a high-dimensional model to low-dimensional space or by performing new training using lowdimensional data.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Pathology (AREA)
  • Databases & Information Systems (AREA)
  • Business, Economics & Management (AREA)
  • General Business, Economics & Management (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

Un procédé de prédiction d'une maladie respiratoire consistant à : déterminer une première maladie respiratoire prédite avec un modèle d'apprentissage automatique sur la base d'un ensemble de données physiologiques ; déterminer un premier niveau de confiance de la première maladie respiratoire prédite ; si le premier niveau de confiance est inférieur à une confiance seuil ; initier un processus d'interrogation de symptôme et recevoir une entrée d'utilisateur sur des symptômes ; déterminer une deuxième maladie respiratoire prédite avec un algorithme à base de règles sur la base de l'entrée d'utilisateur sur des symptômes ; déterminer une troisième maladie respiratoire prédite avec le modèle d'apprentissage automatique sur la base de la deuxième maladie respiratoire prédite et de l'ensemble de données physiologiques ; un deuxième niveau de confiance de la deuxième maladie respiratoire prédite étant supérieur au premier niveau de confiance.
PCT/US2023/026187 2022-06-30 2023-06-26 Système et procédé de détection de maladie respiratoire WO2024006182A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US202263367398P 2022-06-30 2022-06-30
US63/367,398 2022-06-30

Publications (1)

Publication Number Publication Date
WO2024006182A1 true WO2024006182A1 (fr) 2024-01-04

Family

ID=89381230

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2023/026187 WO2024006182A1 (fr) 2022-06-30 2023-06-26 Système et procédé de détection de maladie respiratoire

Country Status (1)

Country Link
WO (1) WO2024006182A1 (fr)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190355473A1 (en) * 2017-01-08 2019-11-21 Henry M. Jackson Foundation For The Advancement Of Military Medicine Systems and methods for using supervised learning to predict subject-specific pneumonia outcomes
US20210319894A1 (en) * 2020-04-08 2021-10-14 CareBand Inc. Wearable electronic device and system using low-power cellular telecommunication protocols
WO2021252768A1 (fr) * 2020-06-10 2021-12-16 Whoop, Inc. Dispositif de surveillance d'infection pouvant être porté sur soi

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190355473A1 (en) * 2017-01-08 2019-11-21 Henry M. Jackson Foundation For The Advancement Of Military Medicine Systems and methods for using supervised learning to predict subject-specific pneumonia outcomes
US20210319894A1 (en) * 2020-04-08 2021-10-14 CareBand Inc. Wearable electronic device and system using low-power cellular telecommunication protocols
WO2021252768A1 (fr) * 2020-06-10 2021-12-16 Whoop, Inc. Dispositif de surveillance d'infection pouvant être porté sur soi

Similar Documents

Publication Publication Date Title
US20230082019A1 (en) Systems and methods for monitoring brain health status
Shishvan et al. Machine intelligence in healthcare and medical cyber physical systems: A survey
JP2022534422A (ja) 生体監視および血糖予測のためのシステムならびに関連付けられる方法
US9754081B2 (en) Context-aware prediction in medical systems
Yu et al. Personalized health monitoring system of elderly wellness at the community level in Hong Kong
US20160302671A1 (en) Prediction of Health Status from Physiological Data
Do et al. Classification of asthma severity and medication using TensorFlow and multilevel databases
WO2022082004A1 (fr) Systèmes et procédés utilisant des techniques d'apprentissage machine ensembliste pour la détection d'évènements futurs
Pourhomayoun et al. Multiple model analytics for adverse event prediction in remote health monitoring systems
Praveen et al. Deep learning based intelligent and sustainable smart healthcare application in cloud-centric IoT
Theerthagiri et al. Cardiovascular disease prediction using recursive feature elimination and gradient boosting classification techniques
Orphanidou et al. Machine learning models for multidimensional clinical data
US20210398683A1 (en) Passive data collection and use of machine-learning models for event prediction
Aghalya et al. Real-time Monitoring and Prediction of Respiratory Diseases Using IoT and Machine Learning
AU2021363110A1 (en) Method and system for personalized prediction of infection and sepsis
WO2016203456A1 (fr) Dispositif, système et procédé de diagnostic amélioré, support de décision, et analyse d'électro-encéphalogrammes
Karimi Moridani An automated method for sleep apnoea detection using HRV
Abineza et al. A machine-learning-based prediction method for easy COPD classification based on pulse oximetry clinical use
WO2024006182A1 (fr) Système et procédé de détection de maladie respiratoire
Madhuri et al. Novel Internet of Things Based Disease Diagnosis Framework for Smart Healthcare Schemes using Combined Optimized Artificial Intelligence Approach
Philip et al. Smart Health Monitoring Using Deep Learning and Artificial Intelligence.
Shirwaikar et al. Design framework for a data mart in the neonatal intensive care unit
Pan et al. Early Prediction of COVID-19 Infection with IoT and Machine Learning
US20240006067A1 (en) System by which patients receiving treatment and at risk for iatrogenic cytokine release syndrome are safely monitored
Basavaraddi et al. Breaking Barriers in Cardiovascular Diagnosis: Cloud-Based Decision Trees Algorithm for Precision Medicine in Heart Disease

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23832189

Country of ref document: EP

Kind code of ref document: A1