Monitoring patient's health
Inventors
Jagirdar, Ammar • Pereira, Adlon • Thummala, Pradeep Kumar • Sharma, Arun Kant • Senapathi, Vijay • Chauhan, Anshul
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11342076-B1
Publication Date
2022-05-24
Expiration Date
Abstract
Disclosed is a system and a method for monitoring patient's health. Data associated with multiple questions nudged to a patient may be received. A plurality of symptoms may be extracted. The plurality of symptoms may be analyzed using to generate a health assessment score. The health assessment score may be compared with a predefined threshold to determine a health condition. A deviation in the health condition may be determined by comparing the health condition and previous health data. A risk level may be identified using an artificial intelligence technique. A course of action may be recommended to the patient in real-time based on the risk level and the health condition. A healthcare practitioner may be identified in real-time. The healthcare practitioner and the patient may be notified in real-time. A follow-up check for the patient may be scheduled based on the risk level and the health condition.
Core Innovation
The invention provides a virtual triage system to monitor patient's health by nudging multiple questions to a patient on a chat platform. The multiple questions are nudged in real-time, presented in multiple languages, and generated based on a trained data model, and the nudges are based on the patients' profile and medical history.
The system receives data from the patient periodically, where the data corresponds to answers associated with the multiple questions. Using natural language processing on the data, it extracts a plurality of symptoms and generates a health assessment score based on analysis of the plurality of symptoms using the trained data model and weightage of each symptom. It then determines a health condition of the patient by comparing the health assessment score with a predefined threshold associated with a set of health conditions, where the health condition corresponds to a disease.
To assess change over time, the system determines a deviation in the health condition by comparing the health condition and previous health data of the patient stored in the trained data model. It identifies a risk level associated with the health condition using an artificial intelligence technique, where the risk level is one of a high risk, a low risk and a moderate risk based on the deviation and the plurality of symptoms. Based on the risk level and the health condition, it recommends a course of action in real-time that comprises a visit to doctor and treatment or medicine for the health condition, identifies a healthcare practitioner in real-time using contextual factors, notifies the healthcare practitioner and the patient in real-time, and schedules a physical follow-up or remote follow-up using a supervised learning approach that learns from past recommendations and actions taken by the patient.
Claims Coverage
The partial content includes two independent claims (system and method). Each independent claim centers on an end-to-end virtual triage workflow with ten inventive features: real-time multilingual question nudging on a chat platform; symptom extraction using natural language processing; weighted health assessment scoring against predefined thresholds; disease determination with deviation from previous health data; AI-based risk level classification; real-time recommendation of a course of action including doctor visit and treatment or medicine; real-time practitioner identification using contextual factors; real-time notification; and follow-up check scheduling using a supervised learning approach based on past recommendations and actions stored in the trained data model.
Real-time multilingual question nudging on a chat platform based on a trained data model and patient profile
Nudge multiple questions to a patient on a chat platform in real-time, present the multiple questions in multiple languages, and generate the nudge based on a trained data model and the patients' profile and medical history.
Periodic answer intake and symptom extraction using natural language processing
Receive data from the patient periodically, where the data corresponds to answers associated with the multiple questions, and extract a plurality of symptoms associated with a health of the patient using a natural language processing technique.
Weighted health assessment scoring from extracted symptoms using a trained data model
Generate a health assessment score of the patient based on an analysis of the plurality of symptoms using the trained data model, wherein the health assessment score is generated based on a weightage of each symptom.
Disease determination by comparing the health assessment score to predefined thresholds
Determine a health condition of the patient based on comparing the health assessment score with a predefined threshold associated with a set of health conditions, wherein the health condition corresponds to a disease.
Deviation determination using previous health data stored in the trained data model
Determine a deviation in the health condition based on comparing the health condition and previous health data of the patient stored in the trained data model.
AI-based risk level identification from deviation and symptoms
Identify a risk level associated with the health condition using an artificial intelligence technique, wherein the risk level is identified based on the deviation and the plurality of symptoms, and wherein the risk level is one of a high risk, a low risk and a moderate risk.
Real-time course of action recommendation including doctor visit and treatment or medicine
Recommend a course of action to the patient in real-time based on the risk level and the health condition, wherein the course of action comprises a visit to doctor and treatment or medicine for the health condition of the patient.
Real-time healthcare practitioner identification using demographic profile, locations, distance, traffic, and feedback
Identify a healthcare practitioner in real-time upon recommending the course of action, wherein the healthcare practitioner is identified based on a demographic profile, the health condition, the risk level, a location of patient, a location of the healthcare practitioner, a distance between the patient and the healthcare practitioner, a traffic condition and feedback associated with the healthcare practitioner received from a set of patients.
Real-time notification to healthcare practitioner and patient based on risk level and deviation
Notify the healthcare practitioner and the patient in real-time based on the risk level and the deviation.
Follow-up scheduling using a supervised learning approach from past recommendations and actions
Schedule a follow-up check for the patient based on the risk level and the health condition, wherein the follow-up check is a physical follow-up or a remote follow-up, and wherein the follow-up check is scheduled using a supervised learning approach that learns from past recommendations and actions taken by the patient, with the past recommendations and the actions taken by the patient stored in the trained data model.
Across the independent claims, claim coverage is centered on a virtual triage workflow that generates real-time multilingual question nudges from a trained data model, extracts symptoms with natural language processing, computes a weighted health assessment score, determines a disease condition via predefined thresholds, derives deviation from previous health data, assigns an AI-based risk level, recommends a real-time course of action including doctor visit and treatment or medicine, identifies an appropriate healthcare practitioner using contextual factors, notifies both parties in real-time, and schedules physical or remote follow-up using supervised learning from past recommendations and actions.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?