Formatting for Journal

Game Theory Based Healthcare Monitoring and

Personalized Recommendation Provisioning in 5G

Edge Enabled Cognitive IoT

1. Introduction

Cognitive computing is the use of electronic modules for stimulating the thought

process of the humans in complicated situations using self-learning algorithms which are

used for data mining, natural language processing and pattern mining. It process huge amount

of data at specific time and provide intelligent recommendations. Rapid increase of medical

data cognitive computing is used for provide better solutions in healthcare system. The

individual medical data of the patients are combined and stored in a large database which

helps to diagnose the disease and detect the treatment from the available medical records

which is similar to the current disease. It integrates the patient information to the medical

experts anywhere and anytime through internet. It makes the medical experts to know the

several kinds of medical information of the patient. The patient medical data are collected by

smart healthcare monitoring devices. The benefits of cognitive computing in healthcare is

defined as follows,

 To enhance the communication between patient and doctor

 Efficient data management and processing

 Incorporating unstructured and structured data

Cognitive computing learns people behavior and activities such as voice, environment,

and psychology using IoT sensors and wearable IoT devices. The IoT devices collect more

unstructured data when compared to structured data. Cognitive computing able to learn and

trained both unstructured and structured data by itself. Cognitive edge computing provides

high quality services to the users which reduce latency during data transmission from user to

cloud computing in cognitive based healthcare system because edge servers are located

nearest to the user. It helps to transmit the data at emergency situation with minimum delay.

Healthcare system not only includes critical data it also includes large amount of health

information which can easily transmit to the medical expert with minimum delay and cost.

Based on the patient medical report the cognitive computing makes a decision.

The smart healthcare system monitors the patient health condition however the patients’

needs to get suggestions from the medical expert or doctors. The doctors are providing their

suggestion based on the patient medical history and current health status through cloud

server. An ontology is constructed for provide the knowledge of the particular domain. The

patient medical informations are presented in the ontology which provides the relationship

between the doctors and patients. It also includes the information of patients like drug

information, disease type, medical history, personal information and etc. which are used to

identify the semantic and syntax representation of the medical information. This type of

knowledge is expressed by SNOMED-CT ontology which provides high quality medical

reports. Open Biology Oriented (OBO) language is used for constructing ontology in

healthcare system which is based on the general concept of Web Ontology Language (OWL).

1.1 Research Aim & Scope

The aim and scope of this research is to monitor the health of patients and provide

recommendations using cognitive computing with minimum delay in 5G-edge assisted Cloud

IoT environment.

1.2 Research Objectives

The main objective of this research is to make a decision of patient health status and

provide recommendations for taking action at emergency situation. The other objectives of

this research is listed as follows,

 To increase accuracy and reduce complexity by presenting pre-processing during

data transmission from IoT layer to edge layer.

 To reduce congestion and latency during medical data transmission in emergency

situation an optimal gateway selection is presented.

 To send accurate recommendations to the patients are effectively handled which

increase high accuracy and prevention of emergency situation.

 To make accurate decisions of the patient health condition (i.e normal, mild,

moderate and severe) for accurately providing recommendations for the

corresponding patient.

 To efficiently manage the data by constructing ontology that can access and

verifies both syntax and semantic manner of the medical data.

2.1 Overall Problem Statement

The cognitive healthcare system has many challenges due to handling numerous

amounts of data in 5G edge assisted IoT cloud environment. However, the existing works

lack in data management and heterogeneity which are listed as follows,

 Improper data management: IoT environment process and handled large amount of

data for monitoring patient health status, however the improper data management

degrades the performance of the health monitoring. It increases high complexity

during searching and collecting the health information for decision making from the

database thus leads to high execution time.

 Increase latency: The existing works perform inefficient preprocessing, clustering

which increase high complexity and increase latency during data transmission. Some

researchers directly send the patient information to the cloud that increases high

latency due to long distance and congestion. At the same time the cloud collects the

data from individual devices which increase high latency.

 Lack of Heterogeneity: The existing works process only homogeneous data it

degrades the performance, however IoT environment receives heterogeneous data

hence it does not effectively handle and process other types of data which reduces the

performance of decision making in cognitive based healthcare system.

 Lack of personal recommendations: Most of the researchers only concentrating on

decision making for monitoring patient health condition however recommendation

provision is also one of the significant process in healthcare system. Hence, it will not

be suitable for patients under different demographic groups

2.2 Specific Problem Statement

Reference 1

Title: A cognitive technology based healthcare monitoring system and medical data

transmission

Concept

The cognitive computing based management and processing of healthcare data was

proposed in this paper. The processing of raw medical data was performed in order to extract

useful information for diagnosis of patients at a right time. The limitations of managing

massive healthcare records were overcome implementing clustering based management of

data. Initially, the IoT devices were incorporated in order to record the patient’s real time

health factors. The acquired data were stored in the local database for further processing of

the raw data. The pre-processing of data was carried out by performing elimination of

redundant information. The formation of both intra clusters and inter clusters was carried out

based on the nature of disease and hospital which facilitated effective management of data.

The data packets were modeled with specific packet header from which the analysis of

packets was performed by utilizing simulation annealing in order to compute the high priority

data.

Problems

 The management of massive data was carried out by construction of inter and intra

clusters based on the several features but the ineffective formation of clusters resulted

in overlapping issues and formation of several empty clusters.

 The processing of data was performed to provide immediate services to the high

priority emergency patients. But it was carried out in the cloud server which is located

in a remote location thereby resulting in increased latency.

 The verification of critical data was performed by implementing simulation annealing,

but this algorithm is inefficient and does not guarantee optimal solutions. This leads to

increase in false alarm rate.

Solutions

 The clustering of input data is carried out by implementing DBSCAN++ algorithm in

order to categorize the outliers and the management of massive data is performed by

construction of semantic based ontology.

 The cognitive decision making is carried out in the edge layer thereby achieving the

low latency further the integration of 5G communication achieves ultra-low latency in

providing recommendations in emergency situations.

 The cognitive decision making is executed based on the knowledge provided by the

ontology in which both the global and local agents are used to produce accurate

results without any false alarms.

Reference 2

Title: Edge cognitive computing based smart healthcare system

Concept

The allocation of healthcare services based on edge cognitive computing was

proposed in this paper. The limitations of cloud based cognitive computing approaches were

overcome by introducing edge based approach. Initially, the user health readings were

recorded by the smart wearable devices which were then transmitted to the edge node for

processing of data. The processing unit comprised of two engines namely data cognitive

engine and resource cognitive engine each having own purposes. The risk level of the patient

was computed from the acquired reading in the data cognitive engine from which the report

was generated to the resource cognitive engine to allocate resources based on the priority.

The processes involved in this approach are, initialization of users, uploading of data,

allocation of resources based on risk evaluation, result generation, handover of mobile users

and alarm generation during emergency condition.

Problems defined

 The risk level of the patient was determined from the features provided by the input

data but the presence of noise and artifacts in the input data degrades the accuracy of

risk evaluation which resulted in increased false alarms.

 The large amount of user data was stored in the cloud layer which will be retrieved by

the edge node whenever necessary. The inefficient management of those massive data

increases complexity in retrieval of information.

 The risk state of the patient was evaluated based on the health parameters of the

patient but several other factors such as psychological factors, and environmental

factors are required to provide accurate determination of risk state.

Solutions

 The data processor in the cognitive engine performs mapping of non-linearity into

high dimensional space using IMQSVR and further the clustering is carried out to

categorize the outliers which are then removed by performing dimensionality

reduction.

 The massive amount of data used for the determination of cognitive decisions are

managed efficiently by construction of semantic based ontology in which both the

SNOMED-CT and OBO ontologies are integrated and the data are managed by three

significant classes.

 The cognitive decision making is carried out based on all the factors such as

psychological factors, physiological factors, environmental factors and past records of

the client thereby providing personalized recommendations.

Reference 3

Title: CIoT‑Net: a scalable cognitive IoT based smart city network architecture

Concept

The cognitive computing based processing of massive data from the smart city

environment was proposed in this paper. The limitations such as scalability and security in

managing a large set of data were addressed. The proposed approach composed of four layers

namely IoT layer, data layer, cognitive computing layer and application layer respectively.

The reading provided by the sensors in the smart city were acquired and interpreted by

performing semantic based modeling. The categorization of data was carried out based on the

purpose of the data through which the processing of respective data was performed. The

extraction of cognitive traits from the data was performed in order analyze the data. The

machine learning algorithms were utilized to provide effective services based on the extracted

features. The services provided by this approach facilitated many fields such as police

department, medical care, law, media industry.

Problems defined

 The CIoT-Net approach was utilized to provide cognitive services in the field of smart

city environment such as traffic management and health care but the latency involved

in providing the services was high which affects the proper execution of the services.

 The data collected from the sensors were preprocessed in order to extract the

necessary information but the inefficiency in preprocessing resulted in presence of

noise and increased dimensionality which affects the accuracy of making decisions.

 The CIoT-Net approach provided precise services in the smart city environment from

the information available in the massive dataset but the management of these massive

dataset was not properly carried out which increased the complexity.

 Here, services are provided to the emergency situation based on the risk level of the

events, however the risk analysis is not properly investigated in this research thus

increases high false alarm rate.

Solutions

 The latency involved in providing the decisions and recommendations to the

respective individual is found to be very low due to the computation of decision

making in edge layer and integration of 5G communication.

 The data processor in the cognitive engine performs mapping of non-linearity into

high dimensional space using IMQSVR and further the clustering is carried out to

categorize the outliers which are then removed by performing dimensionality

reduction.

 The massive amount of data used for the determination of cognitive decisions are

managed efficiently by construction of semantic based ontology in which both the

SNOMED-CT and OBO ontologies are integrated and the data are managed by three

significant classes.

 The reduced false alarm rate is achieved by the proposed approach by precisely

determining the severity of the patient which is performed by means of stackelberg

game between the global and local agents.

Reference 4

Title: Cognitive multi-agent empowering mobile edge computing for resource caching and

collaboration

Concept

The edge computing based execution of cognitive tasks by the cognitive agents was

proposed in this paper. The cognitive model comprised of several cognitive agents which

performed caching and collaboration of resources. The user’s behavioral data was utilized to

build a personalized model. The system model comprised of three layers namely end users,

mobile edge layer and cloud layer respectively. The cognitive agents such as knowledge base,

decision making and execution of actions, analyzing data and processing the data, and

communication with the agents were carried out in order to predict the activities of the user.

Based on the personalized model of the user, the caching of user related application in the

edge layer and collaboration of resources between the agents was carried out. The Long Short

Term Memory (LSTM) based model was deployed for the purpose of prediction of behaviors

and Deep Q Network (DQN) based provisioning of services was carried out.

Problems defined:

 This approach utilized the data provided from the end users but the lack of

preprocessing in the cognitive engine affects the efficiency of the cognitive agents in

predicting the user behavior.

 The prediction of user behavior was facilitated by implementing LSTM model but the

prediction result provided by this model increases time complexity which thereby

affected the performance of this approach.

 The caching of resources related to the user behavior was executed in order to provide

efficient provisioning of services but the resource caching was performed only based

on predicted behavior. The lack of consideration of several environmental and social

parameters affected the efficiency of this approach.

Solutions

 The data processor in the cognitive engine performs mapping of non-linearity into

high dimensional space using IMQSVR and further the clustering is carried out to

categorize the outliers which are then removed by performing dimensionality

reduction.

 The cognitive decision making was performed by the decision maker by

implementing a stackelberg game based approach between the global agent and local

agents’ thereby achieving improved performance.

 The cognitive decision making is carried out based on all the factors such as

psychological factors, physiological factors, environmental factors and past records of

the client thereby providing personalized recommendations.

Reference 5

Title: A Cloud-Edge Collaboration Framework for Cognitive Service

Concept

The integration of edge computing and cloud computing was utilized to provide

cognitive services. The limitations of several conventional edge assisted cloud based

approaches such as inappropriate usage of edge node and static nature of deep learning

models were mitigated by incorporating the edge node based computation of cognitive

services and automatic updation of deep learning models to achieve better performance. The

system model comprised of three layers namely cloud server, edge server, and mobile

devices. The cognitive services considered in this approach are vision related services which

produces real time vision data. The vision data provided by the smart devices were processed

in the edge layer. The edge layer comprised of edge CNN which provides provisioning of

services with reduced latency. The cloud CNN was deployed to train the edge CNN with the

massive data stored in the cloud layer.

Problems defined

 The computation of cognitive services was performed by the edge CNN model which

was trained by cloud CNN model using its massive dataset but the ineffective

management of massive data resulted in increased computational and time

complexity.

 The real time updation of edge CNN was performed with the help of data provided by

the end user. The lack of preprocessing of input data affects the efficiency of the edge

CNN model which thereby affects the performance of this approach.

 The retraining of edge CNN was carried out in a random manner which requires high

resources. The simultaneous action of computation of cognitive services along with

retraining of edge CNN increases the resource complexity.

 Here, optimal services are provided by the edge layers but it does not take any actions

and given priority for emergency situation thus reduce the efficiency of the proposed

work.

Solutions

 The clustering of input data is carried out by implementing DBSCAN++ algorithm in

order to categorize the outliers and the management of massive data is performed by

construction of semantic based ontology.

 The data processor in the cognitive engine performs mapping of non-linearity into

high dimensional space using IMQSVR and further the clustering is carried out to

categorize the outliers which are then removed by performing dimensionality

reduction

 The updation of model utilized for the purpose of cognitive decision making is carried

out in a proper manner and the complexity is reduced by means of properly managing

the data by means of construction of ontology.

 The cognitive decision making is performed to determine the heath condition of the

patient into four classes and through which the severity of the emergency situations is

computed.

3. Proposed work

The proposed work concentrates on providing cognitive decisions in the field of

healthcare based on the information from the well managed knowledge base. The proposed

approach comprises of four major elements namely IoT devices, 5G access points, Edge

layer, and cloud layer respectively. The computations are carried out in the edge layer in

which the cognitive engine is placed. The cognitive engine comprises of three entities such as

data processor, knowledge base, and decision maker. The 5G communication is utilized in

order to achieve the required ultra-low latency in providing decisions to the healthcare users.

The major processes involved in this approach are as follows,

 Data Sensing and Transmission

 Optimal Gateway Selection

 Data Processing

 Semantic Based Ontology Construction

 Cognitive Context Based Decision Making

A. Data Sensing and Transmission

The IoT sensors which are attached to the patient’s body are responsible for the

purpose of acquisition of real time sensing of patient’s health conditions. The sensor inputs

comprising of factors such as heart rate, EEG, ECG, blood sugar, Blood pressure, Body

movement, rate of cholesterol, etc. along with the information such as sex, age, height and

weight, previous medical records of the patients are considered as the input parameters which

are provided in order to effectively formulate the cognitive decisions. These heterogeneous

data are transmitted to the above layers for further processing.

B. Optimal Gateway Selection

In 5G communication, the AP acts as the gateway for transmitting the sensed data to

the edge layer for computation. The availability of large number of IoT devices in the

environment which are utilized for various purposes results in congestion which thereby

increases the latency in the network. In order to achieve the ultra-low latency, the selection of

gateway is done in an optimal manner. The Nomadic People Optimizer (NPO) is

implemented in order to select the optimal gateway based on the significant uplink and

downlink characteristics such as distance, load, link quality, and congestion rate. By doing

so, the proposed approach is able to achieve increased efficiency.

C. Data processing

Once the data reaches the edge layer, the cognitive computing of data is executed. The

processing of input data is performed by the data processor in which the extraction of

valuable information is performed. The processing of data comprises of three stages namely,

non-linear data mapping, clustering and dimensionality reduction. Initially, the fusion of

heterogeneous input is carried out and the mapping of nonlinearity into the linear

counterparts is performed by Implementing Inverse Multi Quadratic Support Vector

Regression (IMQSVR). The mapped linear data are then clustered in order to differentiate

the data based on distance, type and similarity. The clustering is performed by utilizing the

Density Based Spatial Clustering of applications with Noise ++ (DBSCAN++) algorithm

which possess low time complexity in determining the outlier data with noise. Further, the

removal of outlier data is carried out by performing reduction of dimensionality. Through

this, the effective processing of data is executed from which the analysis of data is performed.

D. Semantic Based Ontology Construction

The analysis of data for the purpose of providing cognitive decisions is facilitated by

the construction of Ontology based on both syntax and semantic factors. The main motive of

ontology construction is to achieve effective management of massive data in the knowledge

base thereby obtaining the decisions with low complexity. The Open Biomedical Ontology

(OBO) is integrated with the SNOMED-CT for the purpose of ontology construction. The

major classes of the ontology are physical factors, psychological factors, and personal

information respectively. The clinical status of the individual will also be considered in the

ontology construction. The semantic rules for the purpose of retrieval of the information are

generated based on which the decision making will be carried out.

E. Agent Based Cognitive Decision Making

The cognitive decision making provides the current health condition of the patient based on

the knowledge learned from the knowledge base. The decision maker comprises of a global

agent and several local agents in order to effective determine the decision. The Stackelberg

Game is formulated between the global and local agents in order to achieve the effective

cognitive decision. The utility function of the global agent is to provide the accurate decision

based on the knowledge acquired from the knowledge base. The game theory is adopted to

maximize the utility function of the global agent. Value transfer approach is utilized to in

order to facilitate the fast learning of knowledge. The actions of all the agents were

considered and the action with largest probable reward function is selected as the global

decision. By doing so, the current health condition of the patient is categorized into normal,

mild, moderate, and severe states. These decisions are stored as history decisions in the cloud

layer which are further utilized to achieve better decision results. The personalized

recommendations are generated based on the decisions taken. For instance, if the patient’s

health condition is found to be severe then the respective doctor/physician is informed along

with the generation of alarm to the emergency vehicles nearer to the location of patient.

Several other recommendations such as drug, hospital, expenditure etc. are also produced

based on the situations.

The proposed approach is validated by means of several performance metrics such as,

 Accuracy

 Execution time

 Response time

 Throughput

 CPU utilization

 No of computations vs. Latency

 No of edge nodes vs. Latency

 No of computations vs. Bandwidth utilization

 No of edge nodes vs. Bandwidth utilization

EEG

Ear

sensors

ECG

Motion

detection

sensor

SpO2, Blood

pressure sensor

EMG

Sensing layer 5G

gateway

Optimal

Gateway

Selection

using NPO

Data processor

Non-linear

mapping using

IMQSVR

DBSCAN++

Clustering

Dimensionality

reduction

Knowledge base

Semantic based ontology

construction

Decision maker

Global agent

Local agents (1,2,3..N)

Edge layer

Cloud layer Store

decisions

Normal

Mild

Moderate

Severe

Personalized

Recommendations

Research Highlights

 The preprocessing of medical data performs in order to perform nonlinear

mapping and clustering the data which removes the noise and complexity by

inverse multi quadratic Support vector regression and DBSCAN++ respectively.

Outlier based dimensionality reduction is used for removing outlier from the data

which increase accuracy in patient health monitoring.

 Optimal gateway is selected by Nomadic People Optimizer (NPO) for reducing

latency and congestion during medical data transmission which increase QoS of

this research

 Large amount of data are perfectly managed by proposing OBO ontology which

construct the ontology for medical data which reduces the complexity of data

management in cognitive computing based healthcare system.

 Patient health condition monitoring and decision making is performed by

Stackelberg game theory which provides the optimal decisions based on the

current health status of the patient.

 Recommendations are provided to the patients based on the current health status

which improves the efficiency of the healthcare system and helps to take

corresponding actions at any situation (ex. normal, emergency).

Reference 1

Title: Cognitive Smart Healthcare for Pathology Detection and Monitoring

Concept

This paper proposed cognitive healthcare approach for EEG pathology detection and

classification using deep learning methods. The proposed smart city environment includes

doctor, stakeholders and other help assistants. The main goal of the proposed work is to

diagnose the disease in accurate with low cost and easy access. Patient facial emotions, voice,

movements are recorded for diagnosing pathology. The EEG signals are pre-processed by

converting the EEG single into time series images. Then spatial and temporal features are

extracted from the pre-processed signal using VGG-16 and AlexNet. Based on the features

SVM classifies the EEG singles into normal and abnormal. The simulation result shows that

the proposed work achieves better accuracy compared to other state of the art methods.

Limitations

 Here, SVM is used for signal classification for detecting patient’s conditions; however

it is suitable for small scale environment because SVM takes much time to train the

data for large scale dataset.

 The proposed work achieves less accuracy due to lack of preprocessing like noise

removal, thus reduces feature extraction and classification accuracy and increase false

positive rate.

Reference 2

Title- Brain-Inspired Intelligence for Real-Time Health Situation Understanding in Smart

e- Health Home Applications

Concept

This paper proposed smart e-health home based application based on cognitive

computing for enhancing the health situations and reducing cost for healthcare. The proposed

work has two sections such as preceptor and feedback executive. These two sections include

training, prediction and steady state. Here, fuzzy logic is used for generating decision rules of

patients or users for Non-Gaussian and nonlinear health environment. Bayesian algorithm is

used to construct decision tree in the proposed healthcare environment. Based on the decision

the proposed classifies the patients into healthy and unhealthy. The simulation result shows

that the proposed system achieves high accuracy.

Limitations

 Here, Bayesian analysis is used for decision tree construction which has high

computational cost with large amount of parameters thus reduces the goal

achievements of this research.

 This research only concentrates on disease prediction but providing solutions are also

important in healthcare which helps to face the emergency situation of the patients.

Reference 3

Title- Activity Recognition for Cognitive Assistance Using Body Sensors Data and Deep

Convolutional Neural Network

Concept

This paper proposed to recognize the patient’s activity using deep convolutional

neural network (CNN) for smart healthcare. The proposed work consists of data acquisition,

feature extraction, normalization and training testing process. In first, collects the patients

EEG data using various sensors. In second, PCA algorithm is used to extracting the features

from the collecting data. In third, the features are normalized by Z-score based normalization.

Based on the normalized features the proposed CNN analyses the behavior of the user or

patients. The simulation result shows that the robustness of this research for smart healthcare

system.

Limitations

 Here, PCA algorithm is used for feature extraction which reduces the performance of

the feature extraction because every time numbers of principle components are need

to selected otherwise which may miss the features thus reduces the recognition

accuracy

 This research obtains less accuracy because of using the raw signals for feature

extraction which reduces feature extraction and recognition accuracy.

Reference 4

Title- Big Data-Driven Cognitive Computing System for Optimization of Social Media

Analytics

Concept

This paper proposed big data based cognitive computing for optimizing problem of E-

projects portfolio selection (EPPS) problem in social media. The proposed work used hybrid

fuzzy multi objective optimization algorithm namely NSGA-III and MOIWO algorithms for

solving optimization problems. It minimizes the skewness, variance, and kurtosis and risk

measurement and maximizes the total expected values. For evaluating the performance of the

proposed work this research used 125 active E-projects over a period 2014-2018. The

experimental results show that the proposed model achieves high performance compared to

existing works.

Limitations

 Here, hybrid fuzzy multi objective optimization is proposed for solving EEPS

problems which reduce the accuracy of optimization because the rules are updated for

every time, otherwise it does not provide the optimal solution to the EEPS problem

Reference 5

Title- ITEMa: A methodological approach for cognitive edge computing IoT ecosystems

Concept

This paper proposed IoT based smart ecosystem modeling (ITEM) approach for

cognitive edge computing IoT environment. The proposed work includes three layers such as

bottom layer, intermediate layer and awareness layer. The bottom layer consist sensors,

actuators, data sources, and data consumers. Each user is accessed and manger the data

through smart object. The intermediate layer includes set of logical units which collects the

information with its location. The awareness layer has the responsible for taking decisions

based on the behaviors. This research work is applied and evaluated for smart office

environment.

Limitations

 Here, data are collected from sensors and actuators which are suitable for small scale

environment when large environment the data are collected from many sensors and

actuators thus leads to high complex during individual data can be accessed and

processed.

 This research takes high much of time for providing solutions because each and every

time it will perform all the three layers process thus increase high latency that reduces

the performance of the proposed work.

Reference 6

Title- A 5G Cognitive System for Healthcare

Concept

This paper proposed a novel technique for healthcare in 5G cognitive environment.

The proposed work consist three-layer architecture such as infrastructure layer, resource

cognitive engine layer and data cognitive engine layer. The first layer includes user terminal,

edge cloud, remote cloud and RAN. And the RAN network is used to provide a

communication to the infrastructure layer. The second layer includes resource cognitive

engine which provides optimized solution by perception, user information and network

context learning. It can achieve high flexibility, scalability, reliability, and low latency for

the proposed 5G cognitive environment. In third layer includes data cognition engine which

can achieve environmental perception, human cognition using intelligent algorithms.

Limitations

 The proposed work has less accuracy and high time complexity because of using

intelligent algorithms in 5G cognitive healthcare system.

 The edge cloud collect the information from the user terminal such as smart phone,

sensors and etc. which increase high complexity because of collect the information

directly from the individual users that leads to high latency.

Reference 7

Title- Complementing Agents with Cognitive Services: A Case Study in Healthcare

Concept

This paper proposed novel methods with complementing agents and cognitive

services for healthcare applications. Author proposed a model for integration between Belief

Desire Intention (BDI) agents using machine learning algorithms which manage the activities

of trauma resuscitation inside emergency situation. The main goal of this research is to

provide the flexible cognitive services to the individual patient context and history. In trauma

management, the trauma agent collects and tract the data which are relevant to the particular

event. The location services are provides by beacon based infrastructure. Based on these

collected information recommendations or suggestions are provided to the individual patients

in the environment.

Limitations

 Here, Trauma agent collects the relevant information of the events; however, it cannot

collect the semantic information of the events thus reduces the performance of

providing recommendations and suggestions.

 Here, machine learning algorithms are used for managing the activities of trauma

resuscitation which is suitable for small scale environment when large scale

environment machine learning takes high much of time for training thus leads to high

latency.

Reference 8

Title- Experimentation of a Smart Learning System for Law Based on Knowledge Discovery

and Cognitive Computing

Concept

This paper proposed smart learning system based on cognitive computing and

knowledge discovery techniques for law students. The proposed work has many components

such as knowledge discovery, learning, and management component. This work is used for

analyzing the process of legal case and relationship with its domain knowledge. It can able to

match the relative topic using domain knowledge. For that, it includes three processes which

are concept extraction, classified the extract concept based on legal ontologies and semantic

search. The simulation result shows that the proposed work achieves high performance in

terms of efficiency, usability and effectiveness compared to existing works.

Limitation

 Here, ontology is constructed for discover the knowledge for law students however it

does not allow new request enter into the ontology which reduces the performance of

the proposed work.

Reference 9

Title- Research and Design on Cognitive Computing Framework for Predicting Judicial

Decisions

Concept

This paper proposed cognitive computing for forecasting the decisions of judicial. The

proposed work includes three layers such as legal semantics understanding layer, legal

knowledge reasoning layer and legal knowledge learning layer. First layer includes three

processes such as data preparation, extraction rules and establishment of first order logic

base. The data are collect from china judgments online and the factors are extracted from the

decision of judicial. The rules and logical statements are collected by first order logic base

and TML language is used for generate rules by extracting the legal factors. In second layer,

legal factors values are collected and identify the concept and relationship using deep

learning. The past and future rules are provided by Bi-LSTM and CRF. In third layer,

provides the predicted judicial decisions.

Limitations

 Here, semantic matching and rules predictions are performed by deep learning

algorithms which are not provide the accurate semantic matching thus reduces

prediction accuracy.

Reference 10

Title- A Medical Decision Support System to Assess Risk Factors for Gastric Cancer Based

on Fuzzy Cognitive Map

Concept

This paper proposed fuzzy cognitive map technique for making medical decision for

accessing risk factor of cancer. In first the risk value of the cancer is evaluated by considering

risk factors, symptoms. Based on these factors decisions are made by FCM which includes

three main processes such as concept identification, evaluate relationship among initial

weights and concepts and weighting. Then Nonlinear Hebbian learning (NHL) algorithm is

proposed is used to calculate the weight values. The proposed FCM method effectively

calculates the risk value of the cancer. The simulation result shows that the proposed model

achieves high performance compared to existing works.

Limitations

 Here, the risk value of cancer is predicted and analyzed however personalized

recommendation provision is also important for taking corresponding action based on

risk level.

 In fuzzy logic the weight values are updated at every time otherwise it does not

provide optimal risk value which reduces the accuracy of the process.

Reference 11

Title: COVID-SAFE: An IoT-based System for Auto-mated Health Monitoring and

Surveillance in Post-Pandemic Life

Concept

Fog computing based monitoring of patient’s health was performed in this paper. The

maintenance of physical distance between the people in order to mitigate the COVID-19

pandemic was suggested by the application created in the smartphone. The risk of spread of

pandemic was predicted by means of Fuzzy approach based on the health conditions of the

individual and environmental conditions around the individuals. The health factors

considered in this approach are severity of cough, body temperature, oxygen level, presence

of respiratory problems, age, and gender etc. The prediction of risk factor was categorized

into three factors namely low, moderate and high.

Limitations

 Only generalized assistances were provided based on the severity and the assistance

provided cannot suit for individuals in all demographic groups.

 The management of data for the purpose of prediction of risk factor was not carried

out in an effective manner which resulted in increased complexity and the lack of

preprocessing of input sensor data reduces the accuracy of prediction.

Reference 12

Title: A Flexible and Pervasive IoT Based Healthcare Platform for Physiological and

Environmental Parameters Monitoring

Concept

The smart wearable based monitoring of healthcare of the patients was proposed in

this paper. The IoT gateway was incorporated in order to achieve the communication between

the patient and doctor in a simultaneous manner. The system model comprised of four levels

such as sensor level, sensor node level, data collection level, data management level

respectively. Both the health characteristics and environmental features were acquired for the

purpose of monitoring. Once the acquisition of data was completed the processing of data

was carried out. The processes such as creation of pattern, conversion of data, calibration and

re-calibration were carried out.

Limitations

 The clinical investigation of healthcare patient was performed by means of IoT

devices but the provision of solutions took more time due to increased congestion in

the network.

 The preprocessing of input data acquired from the smart wearable was not executed

which affects the efficiency of the investigation process.

Reference 13

Title: HealthFog: An Ensemble Deep Learning based Smart Healthcare System for

Automatic Diagnosis of Heart Diseases in Integrated IoT and Fog Computing Environments

Concept

The automatic diagnosis of heart disease based on fog-IoT architecture was proposed

in this paper. The limitations of cloud computing in healthcare monitoring such as increased

latency was addressed by utilizing fog based approach. The IoT devices were used to acquire

the input sensor data from which the diagnosis of heart disease was carried out in an

automatic manner. The ensemble based lightweight deep learning model was implemented in

order to perform the real time diagnosis. The fog broker comprised of four entities such as

cloud integrator, resource manager, data manager, and security manager through which the

bidirectional communication between the patients and doctors was achieved.

Limitations

 The prediction of heart disease was performed by using the ensemble learning model

but the accuracy of the predicted results was not sufficient as it produced more

number of false alarms.

Reference 14

Title: IoT Based Smart Edge for Global Health: Remote Monitoring with Severity Detection

and Alerts Transmission

Concept

The edge computing based monitoring of health conditions and transmission of alert

messages was proposed in this paper. The limitations of conventional cloud computing based

healthcare monitoring approaches were considered and an effective edge based approach was

proposed. The health parameters considered in this approach were namely pulse rate, oxygen

level, blood pressure, etc. Initially, the conversion of sensor data into information was carried

out in order to compute the criticality of the patient. The transmission of alert messages was

based on the measure of criticality of the data.

Limitations

 The real time monitoring and determination of critical condition was carried out but

the lack of cognitive based decision making affects the accuracy of the criticality of

the patients.

Reference 15

Title: An IoT-Based Healthcare Platform for Patients in ICU Beds During the COVID-19

Outbreak

Concept:

The remote monitoring of COVID-19 patients by utilizing the smart sensors was

proposed in this paper. The major causes for the increasing number of COVID-19 cases were

examined and a remote monitoring approach was proposed. The cloud computing based

architecture was incorporated by this approach. The system module comprised of three

modules namely hospital module, intelligence module, and IoT data collection module. The

rule based approach was incorporated in order to perform effective monitoring of the remote

patients. The proposed approach was validated by comparing with the existing approaches in

terms of several significant metrics.

Limitations

 The remote monitoring of COVID-19 patients in order to provide effective assistance

was performed in a cloud based architecture which increased the latency in

monitoring thereby resulting in severe problem to the patients.

 The big data was incorporated in the proposed approach but the lack of management

of massive data in the big data environment resulted in increased complexity in

performing the monitoring process.

References:

1. Kumar, M.A., Vimala, R., & Britto, K.R. (2019). A cognitive technology based

healthcare monitoring system and medical data transmission. Measurement, 146, 322-

332.

2. M. Chen, W. Li, Y. Hao, Y. Qian, I. Humar, Edge cognitive computing based smart

healthcare system, Future Generation Computer Systems (2018).

3. Park, J., Salim, M.M., Jo, J., Sicato, J.C., Rathore, S., & Park, J.H. (2019). CIoT-Net:

a scalable cognitive IoT based smart city network architecture. Human-centric

Computing and Information Sciences, 9, 1-20.

4. Wang, R., Li, M., Peng, L., Hu, Y., Hassan, M., & Alelaiwi, A. (2020). Cognitive

multi-agent empowering mobile edge computing for resource caching and

collaboration. Future Gener. Comput. Syst., 102, 66-74.

5. Ding, C., Zhou, A., Liu, Y., Chang, R.N., Hsu, C., & Wang, S. (2020). A Cloud-Edge

Collaboration Framework for Cognitive Service. IEEE Transactions on Cloud

Computing, 1-1.

6. Chen, M., Yang, J., Hao, Y., Mao, S., & Hwang, K. (2017). A 5G Cognitive System

for Healthcare. Big Data Cogn. Comput., 1, 2.

7. Montagna, S., Mariani, S., Gamberini, E., Ricci, A., & Zambonelli, F. (2020).

Complementing Agents with Cognitive Services: A Case Study in Healthcare. Journal

of Medical Systems, 44.

8. Capuano, N., & Toti, D. (2019). Experimentation of a smart learning system for law

based on knowledge discovery and cognitive computing. Comput. Hum. Behav., 92,

459-467.

9. Li, J., Zhang, G., Yu, L., & Meng, T. (2019). Research and Design on Cognitive

Computing Framework for Predicting Judicial Decisions. Journal of Signal Processing

Systems, 1-9.

10. Mahmoodi, S.A., Mirzaie, K., Mahmoodi, M., Mahmoudi, S.M., & Huang, T. (2020).

A Medical Decision Support System to Assess Risk Factors for Gastric Cancer Based

on Fuzzy Cognitive Map. Computational and Mathematical Methods in Medicine,

2020.

11. Vedaei, S., Fotovvat, A., Mohebbian, M., Rahman, G.M., Wahid, K., Babyn, P.,

Marateb, H., Mansourian, M., & Sami, R. (2020). COVID-SAFE: An IoT-Based

System for Automated Health Monitoring and Surveillance in Post-Pandemic Life.

IEEE Access, 8, 188538-188551.

12. Haghi, M., Neubert, S., Geißler, A., Fleischer, H., Stoll, N., Stoll, R., & Thurow, K.

(2020). A Flexible and Pervasive IoT-Based Healthcare Platform for Physiological

and Environmental Parameters Monitoring. IEEE Internet of Things Journal, 7, 5628-

5647.

13. Tuli, S., Basumatary, N., Gill, S.S., Kahani, M., Arya, R., Wander, G., & Buyya, R.

(2020). HealthFog: An Ensemble Deep Learning based Smart Healthcare System for

Automatic Diagnosis of Heart Diseases in Integrated IoT and Fog Computing

Environments. Future Gener. Comput. Syst., 104, 187-200.

14. Pathinarupothi, R.K., Durga, P., & Rangan, E. (2019). IoT-Based Smart Edge for

Global Health: Remote Monitoring With Severity Detection and Alerts Transmission.

IEEE Internet of Things Journal, 6, 2449-2462.

15. Filho, I., Aquino, G.S., Malaquias, R.S., Girão, G., & Melo, S. (2021). An IoT-Based

Healthcare Platform for Patients in ICU Beds During the COVID-19 Outbreak. IEEE

Access, 9, 27262-27277.

16. Vedaei, S., Fotovvat, A., Mohebbian, M., Rahman, G.M., Wahid, K., Babyn, P.,

Marateb, H., Mansourian, M., & Sami, R. (2020). COVID-SAFE: An IoT-Based

System for Automated Health Monitoring and Surveillance in Post-Pandemic Life.

IEEE Access, 8, 188538-188551.

17. Haghi, M., Neubert, S., Geißler, A., Fleischer, H., Stoll, N., Stoll, R., & Thurow, K.

(2020). A Flexible and Pervasive IoT-Based Healthcare Platform for Physiological

and Environmental Parameters Monitoring. IEEE Internet of Things Journal, 7, 5628-

5647.

18. Tuli, S., Basumatary, N., Gill, S.S., Kahani, M., Arya, R., Wander, G., & Buyya, R.

(2020). HealthFog: An Ensemble Deep Learning based Smart Healthcare System for

Automatic Diagnosis of Heart Diseases in Integrated IoT and Fog Computing

Environments. Future Gener. Comput. Syst., 104, 187-200.

19. Pathinarupothi, R.K., Durga, P., & Rangan, E. (2019). IoT-Based Smart Edge for

Global Health: Remote Monitoring With Severity Detection and Alerts Transmission.

IEEE Internet of Things Journal, 6, 2449-2462.

20. Filho, I., Aquino, G.S., Malaquias, R.S., Girão, G., & Melo, S. (2021). An IoT-Based

Healthcare Platform for Patients in ICU Beds During the COVID-19 Outbreak. IEEE

Access, 9, 27262-27277.