AU2020102643A4

AU2020102643A4 - Nlp-artificial intelligence based automatic detection of infection rate of pandemic diseases (covid-19)

Info

Publication number: AU2020102643A4
Application number: AU2020102643A
Authority: AU
Inventors: Puneet H Chamakeri; Veera Chinnadurai; Ranjit H. D.; Jagannath Jadhav; Rubina Jahangir Khan; Arati Patil; Prasad B. Rampure; Surapudi Srinivasa Rao; Amruta P. Sonavale
Original assignee: Chamakeri Puneet H Dr Dr; Jadhav Jagannath Dr; Khan Rubina Jahangir Miss; Patil Arati Dr; Rampure Prasad B Dr; Rao Surapudi Srinivasa Dr; Sonavale Amruta P Mrs
Current assignee: Chamakeri Puneet H Dr Dr; Jadhav Jagannath Dr; Khan Rubina Jahangir Miss; Patil Arati Dr; Rampure Prasad B Dr; Rao Surapudi Srinivasa Dr; Sonavale Amruta P Mrs
Priority date: 2020-10-08
Filing date: 2020-10-08
Publication date: 2020-11-26
Anticipated expiration: 2028-10-08

Abstract

NLP-ARTIFICIAL INTELLIGENCE BASED AUTOMATIC DETECTION OF INFECTION RATE OF PANDEMIC DISEASES (COVID-19) Threat of pandemic diseases is emerging in the current era, specifically Covid-19 acting as infectious disease leading to death of huge population in the universe. Involvement of technology is demanded for innovative solutions in various aspects in perspective of pandemic disease. A hope is offered by Artificial Intelligence (AI) which can efficiently prevent, pre-empt and combat the threats of such epidemic infectious disease. Enormous potential is offered by Al concerning public health for revolutionizing healthcare which can provide expansion for accessing health information along with its services which in turn enhances the responsibility of individual to know about their well being and health. Natural language programming (NLP) is one of the subfield of Al aiming to bridge the gap between human language and computer language. Applications of Al are able to determine significance of text by using NLP by which machine is able to identify phrases and keywords related to pandemic disease such that infection rate is determined automatically from huge amount of datasets. NLP focuses on information processing and managing data to infer relationship between featured words in a prominent way. and preprsigPAy ShData Data YJ VanesTesDofDiseasesRelated Data Figure 1. Proposed System of Artificial Intelligence based Detection of Disease Infection Rate onnesnades data itdeon edpEgy enanEM NER dl (sympu~'ms innags nidd0 in Figure 2. FlowDiagram of Natural Language Programming for SymptomsDetection

Description

NLP-ARTIFICIAL INTELLIGENCE BASED AUTOMATIC DETECTION OF INFECTION RATE OF PANDEMIC DISEASES (COVID-19)

Threat of pandemic diseases is emerging in the current era, specifically Covid-19

acting as infectious disease leading to death of huge population in the universe.

Involvement of technology is demanded for innovative solutions in various aspects

in perspective of pandemic disease. A hope is offered by Artificial Intelligence

(AI) which can efficiently prevent, pre-empt and combat the threats of such

epidemic infectious disease. Enormous potential is offered by Al concerning public

health for revolutionizing healthcare which can provide expansion for accessing

health information along with its services which in turn enhances the responsibility

of individual to know about their well being and health. Natural language

programming (NLP) is one of the subfield of Al aiming to bridge the gap between

human language and computer language. Applications of Al are able to determine

significance of text by using NLP by which machine is able to identify phrases and

keywords related to pandemic disease such that infection rate is determined

automatically from huge amount of datasets. NLP focuses on information

processing and managing data to infer relationship between featured words in a

prominent way.

and preprsigPAy ShData Data

onnesnades

YJ

VanesTesDofDiseasesRelated Data

Figure 1. Proposed System of Artificial Intelligence based Detection of Disease Infection Rate

data itdeon

edpEgy enanEM

NER dl (sympu~'ms innags nidd0 in

Figure 2. FlowDiagram of Natural Language Programming for SymptomsDetection

Editorial Note 2020102643 There is only six pages of the description

COMPLETE SPECIFICATION

(See Section 10; rule 13)

TITLE OF THE INVENTION

APPLICANT

The following specification particularly describes the invention and the manner in which it is to be performed

Field and background of the invention

Substantial threats are posed by infectious diseases such as COVID-19

whose outbreak has created serious repercussions for the security of global public

health. These emerging infectious diseases have spurred global concerns to prepare

themselves for pandemic diseases. Healthcare workforce has deployed at a massive

level to save life from this red alert condition. Research is conducted continuously

since the outbreak of pandemic disease for finding vaccine but yet no fruitful

results hence only way to save life is isolate infectious patients by quarantining

them separately in cells such that spreading of pandemic is reduced.

Technology plays vital role is analyzing clinical data specifically Artificial

Intelligence based detection technique is proposed in this invention. Natural

Language Programming (NLP) is involved in this invention which is able to detect

symptoms of pandemic disease such as fever, cough, head-ache from primary

health records in order to detect the infection rate of pandemic disease. Good

performance is achieved by the proposed natural language processing algorithm in

extraction of symptoms and signs along with which status of symptom and its

duration is detected such as suspected case, negated case and affirmed case.

Artificial neural network plays an important role by which the system becomes

intelligent in detection of pandemic disease.

Summary of Invention

In this invention Artificial Intelligence (AI) based concept is implemented to

automate the process of computation of infection rate of pandemic disease.

Machine learning (ML) is one of the subfield of Al in which system implies to

learn from from its previous experiences and adapt itself based on the learning

experience. ML learning is further categorized into supervised learning and

unsupervised learning where supervised learning uses patterns of training data

while unsupervised learning finds and learns from data patterns such as data

mining identifying patterns from large datasets. In order to track the health

behavior during pandemic period, Deep learning (DL) will suit best which is a

specific subset of machine learning utilizing neural networks.

Specifically it acts as replica of structure and functions of human brain

synthetically where DL is able execute several functions such as Natural Language

Programming (NLP) and image recognition where it is able to handle huge datasets

of information flow. NLP is focused specifically in this invention which is able to

identify phrases and words by using algorithms in natural language which can be

unstructured written text. It is also able to understand the meaning of the text

detected. In NLP, topic modeling approach is involved in this invention where

NLP seeks to identify topics automatically in documents based on inferred

relationship among the featured words prominently.

Brief description of the system

• In this invention, proposed system is able to compute infection rate of

pandemic disease based on Artificial Intelligence based NLP which is a

technique able to process data which is in the form of text.

• First process in NLP is tokenizing which identifies basic units called as

tokens required for the next process. Tokenizing is the process of breaking

down large textual data into smaller textual data in order to facilitate the

analysis process. Hence paragraphs will be tokenized to sentences and

sentences in turn tokenized into words. In our work, we require words to be

identified from huge data set such as cough, fever, head-ache and

tastelessness being the basic symptoms of COVID-19.

• Tokenizing is followed by stemming process which works on transforming

words or tokens obtained from above step into basic forms where the basic

forms may not be same as that of the root word in any language.

• Stemming process utilizes NA algorithm which provides highest accuracy of

95.4 % in converting into basic forms than any other stemming algorithms as

it uses its main reference as basic word dictionary.

• Next process in NLP implementation is stopwords removal in the field of

machine learning where this method is able to remove unnecessary

conjunctions in the text without affecting main words or content.

" System performance is improved by the removal of stopwords which leads

to effective processing of only required data reducing overloading.

• In our invention importance is given to general symptoms of COVID-19 as

aim of this work is detect infection rate based on symptoms.

• List of stopwords is static which implies it is obtained from developer

agency else symptoms are provided by medical experts to be applied for

NLP program for the filtering process to be accurate.

• In NLP, Named Entity Recognition (NER) is a specific part of Information

Extraction (IE) which requires training data from different patient's datasets.

• BIO dataset format is utilized by NER for identifying the symptoms of

COVID-19. If any dataset has words matching the symptoms they are

labeled indicating the closeness of words with symptoms.

• Primary Health centres are involved in data collection of symptoms from

every individual hence have generated large amount of dataset. These

datasets are labeled depending on their closeness with symptoms.

The invention is herein described, with the accompanying block diagrams.

Wherein:

Figure 2. Flow Diagram of Natural Language Programming for Symptoms Detection

Description of the system

" In this system, integration of data translation is involved as the primary

health records can be in any language hence the translation process aims at

changing medical data of any language into English as the ICD reference

data issued by WHO (World Health Organization) is in English.

• Stemming and removing of stopwords changes the structure of medical data

which still undergoes data translation into English followed by Access ICD

data which is done for categorizing symptoms related to the disease

symptoms. Feedback from ICD system is provided in the form of JSON file

containing classification of symptoms of pandemic disease.

• If translation is appropriate, then classification is done accurately to related

symptoms and if the translation result is not appropriate then data parsed to

ICD is also in the form of unclear data resulting in appropriate data.

• Web application is used in web design view for displaying the result on web

page which indicates symptoms found in patients along with computation of

infection rate of the disease.

• Natural language programming (NLP) is one of the subfield of Al which is

able to bridge the gap between human language and computer language

making the detection process accurate.

Claims

Editorial Note 2020102643 There is only one page of the claim CLAIMS We Claim:

1. The proposed system performs detection of infection rate of pandemic

disease based on Artificial Intelligence.

2. Natural Language Programming (NLP) is involved in this system for

bridging the gap between human language and computer language.

3. Symptoms such as cough, sore throat, fever, head ache are given significant

importance for disease detection.

4. If the suspected symptoms occur continuously for a patient, he is counted for

computation of infection rate.

5. Extraction of clinical information from primary health centre records using

natural language processing algorithm.

6. NLP algorithm is also able to capture assertion status such as suspected,

negated or affirmed along with duration of symptom.

Figure 2. Flow Diagram of Natural Language Programming for Symptoms Detection