System and method for iterative classification parameter generation based on real-time physiological and audio interview data
A system for generation of diagnosis parameters based on patient-related data, including a processor of a classification server node configured to host a machine learning module and connected to an interview entity node and to at least one medical entity node over a network; and a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to: acquire sensory data from a plurality of biosensors encapsulated into a patient wearable device; receive patient interview data from the interview entity node comprising audio data generated during patient interview; derive a language metadata from the interview data; parse the interview data based on the language metadata to derive a plurality of key features; query a local patients' database to retrieve local historical patients'-related data related to previous patients' engagements associated with previous interview data based on the plurality of features; generate at least one feature vector based on the plurality of features, the sensory data and the local historical patients'-related data; and provide the at least one feature vector to the ML module configured to generate a predictive model for producing at least one diagnosis parameter for generation of the patient-related classification for the at least one medical entity node.
1 . A system comprising:
at least one biosensor;
a microphone;
at least one processor; and
a memory communicatively coupled to the at least one processor, the memory storing non-transitory machine-readable instructions that, when executed by the processor, cause the processor to perform operations comprising:
receiving, from the at least one biosensor, sensory data, the sensory data comprising at least one of heart rate, blood oxygen saturation, or body temperature of a user wearing a wearable device comprising the at least one biosensor;
receiving, from the microphone, audio data based on speech of the user;
transcribing the audio data, resulting in transcribed audio;
retrieving, from at least one database, demographic information of previous patients associated with the user;
generating a first feature vector based on the sensory data, the transcribed audio, and the demographic information;
executing an artificial intelligence (AI) algorithm, wherein:
inputs to the AI algorithm comprise the first feature vector; and
output of the AI algorithm comprises a predictive model;
generating an updated feature vector based on additional sensor data acquired after receiving the sensory data and additional transcribed audio data acquired after receiving the audio data; and
providing the updated feature vector to the AI algorithm, wherein the AI algorithm executes the predictive model to produce a predicted illness and a treatment recommendation for the predicted illness.
2 . The system of claim 1 , wherein the at least one processor is configured to determine whether the sensory data deviates from a prior recorded value by more than a predefined threshold.
3 . The system of claim 2 , wherein the processor is configured to, responsive to detecting a deviation from the predefined threshold, re-execute the AI algorithm using predictive model and updated sensory data.
4 . The system of claim 1 , wherein the predicted illness is stored on a blockchain ledger in association with selected features derived from the transcribed audio.
5 . The system of claim 4 , wherein the at least one a memory has stored additional non-transitory machine-readable instructions that, when executed by the processor, cause the processor to perform operations comprising:
generating the selected features based at least in part on the audio data, wherein the audio data is collected during an interaction between the user and a chat bot; and
deriving a plurality of key features from the transcribed audio prior to the generating of the first feature vector,
wherein the first feature vector is further based on the plurality of key features.
6 . A method comprising:
receiving, at a computer system from a plurality of biosensors of a wearable device, sensory data, the sensory data comprising at least one of heart rate, blood oxygen saturation, or body temperature of a user wearing the wearable device;
receiving, at the computer system, audio data generated based on speech of the user;
transcribing, via at least one processor of the computer system, the audio data, resulting in transcribed audio;
retrieving, via the at least one processor from at least one database, demographic information of previous patients associated with the user;
generating a first feature vector based on the sensory data, the transcribed audio, and the demographic information;
executing, via the at least one processor, an artificial intelligence (AI) algorithm, wherein:
inputs to the AI algorithm comprise the first feature vector; and
output of the AI algorithm comprises a predictive model;
generating an updated feature vector based on additional sensor data acquired after receiving the sensory data and additional transcribed audio data acquired after receiving the audio data; and
providing the updated feature vector to the AI algorithm, wherein the AI algorithm executes the predictive model to produce a predicted illness and a treatment recommendation for the predicted illness.
7 . The method of claim 6 , further comprising: monitoring the sensory data of the user to determine whether the sensory data deviates from a prior recorded value by more than a predefined threshold.
8 . The method of claim 7 , further comprising responsive to detecting a deviation from the predefined threshold, re-executing the AI algorithm using the predictive model and updated sensory data.
9 . The method of claim 6 , further comprising storing the predicted illness on a blockchain ledger in association with selected features derived from the transcribed audio.
10 . The method of claim 9 , wherein generating of the selected features comprises processing the audio data, wherein the audio data is collected during an interaction between the user and a chat bot.
11 . A non-transitory computer-readable medium having instructions stored which, when executed by at least one processor, cause the at least one processor to perform operations comprising:
receiving, from at least one biosensor, sensory data, the sensory data comprising at least one of heart rate, blood oxygen saturation, or body temperature of a user wearing a wearable device comprising the at least one biosensor;
receiving, from a microphone, audio data based on speech of the user;
transcribing the audio data, resulting in transcribed audio;
retrieving, from at least one database, demographic information of previous patients associated with the user;
generating a first feature vector based on the sensory data, the transcribed audio, and the demographic information;
executing an artificial intelligence (AI) algorithm, wherein:
inputs to the AI algorithm comprise the first feature vector; and
output of the AI algorithm comprises a predictive model;
generating an updated feature vector based on additional sensor data acquired after receiving the sensory data and additional transcribed audio data acquired after receiving the audio data; and
providing the updated feature vector to the AI algorithm, wherein the AI algorithm executes the predictive model to produce a predicted illness and a treatment recommendation for the predicted illness.
12 . The non-transitory computer-readable medium of claim 11 , having additional instructions stored which, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
monitoring the sensory data to determine whether the sensory data deviates from a prior recorded value by more than a predefined threshold.
13 . The non-transitory computer-readable medium of claim 12 , having additional instructions stored which, when executed by the at least one processor, cause the at least one processor to perform operations comprising: upon detecting a deviation from the predefined threshold, re-executing the AI algorithm using the predictive model and updated sensory data.
14 . The non-transitory computer-readable medium of claim 11 , having additional instructions stored which, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
storing the predicted illness on a blockchain ledger in association with selected features derived from the transcribed audio.
15 . The system of claim 1 , wherein the AI algorithm comprises a neural network.
16 . The system of claim 1 , wherein the sensor data, the transcribed audio, and the demographic information comprise training data for the predictive model.
17 . The method of claim 6 , wherein the AI algorithm comprises a neural network.
18 . The method of claim 6 , wherein the sensor data, the transcribed audio, and the demographic information comprise training data for the predictive model.
19 . The system of claim 5 , wherein the key features are derived, at least in part, based on a language of the speech of the user; and
wherein the key features are extracted from the transcribed audio based on keyword matching from notes associated with the user.
20 . The system of claim 19 , wherein the predicted illness is sepsis, and wherein the predictive model is configured to perform sepsis risk assessment modeling based at least in part on the key features and the sensory data.