Applying Artiﬁcial Intelligence in Healthcare Social Networks to Identity

Critical Issues in Patients’ Posts

Giacomo Fiumara

, Antonio Celesti

2,4

, Antonino Galletta

2,3

Lorenzo Carnevale

2,3

and Massimo Villari

2,3

Department of Mathematics and Computer Science, Physics and Earth Sciences, University of Messina,

Contrada di Dio 1 - 98166, Messina, Italy

Department of Engineering, University of Messina, Contrada di Dio 1 - 98166, Messina, Italy

IRCCS Centro Neurolesi “Bonino Pulejo”, Contrada Casazza, SS113 - 98124, Messina, Italy

Alma Digit S.R.L. Research Labs, Contrada di Dio 1 - 98166, Messina, Italy

Keywords:

Healthcare, Social Network, Artiﬁcial Intelligence, Machine Learning.

Abstract:

Nowadays, the possibility of using social media in the healthcare ﬁeld is attracting the attention of clinical

professionals and of the whole healthcare industry. In this panorama, many Healthcare Social Networking

(HSN) platforms are emerging with the purpose to enhance patient care and education. However, they also

present potential risks for patients due to the possible distribution of poor-quality or wrong information. On

one hand doctors want to promote the exchange of information among patients about a speciﬁc disease, but on

the other hand they do not have the time to read patients’ posts and moderate them when required. In this paper,

we propose an Artiﬁcial Intelligence (AI) approach based on a combination of stemming, lemmatization and

Machine Learnign (ML) algorithms that allows to automatically analyse the patients’ posts of a HSN platform

and identify possible critical issues so as to enable doctors to intervene when required. In particular, after a

discussion of advantages and disadvantages of using a HSN platform, we discuss in detail an architecure that

allows to analyse big data consisting of patients’ posts. In the end, real case studies are discussed highlighting

future challenges.

1 INTRODUCTION

Nowadays, there is an increasing interest of clini-

cal operators in social media, big data analytics and

Cloud computing. All over the world, the num-

ber of investments in Information and Communica-

tion Technology for health and wellbeing (eHealth)

is rapidly increasing. Global eHealth market is ex-

pected to reach USD 308.0 billion by 2022, according

to a new report by Grand View Research Inc. In par-

ticular, the transition of the healthcare industry into

digital healthcare system for management and analy-

sis of patients’ health is expected to be the most vital

driver of the market (www.grandviewresearch.com,

2016). The European Commission’s eHealth Ac-

tion Plan 2012-2020 has already provided a roadmap

to empower patients and healthcare workers, to link

up devices and technologies, and to invest in re-

search towards the personalised medicine of the fu-

ture (ec.europa.eu, 2012). In February 2017, the

European Commission set up an internal task force

bringing together technology and health policy mak-

ers to examine EU policy actions to ensure the trans-

formation of health care into a Digital Single Market

(DSM) bringing beneﬁts for people, health care sys-

tems and the economy (ec.europa.eu, 2014). Guaran-

teeing access to high-quality health care is a key ob-

jective of social protection systems in European coun-

tries and it represents the second largest social expen-

diture item after pensions.

In this panorama, social media represent a tempt-

ing opportunity for healthcare operators for improv-

ing the patients’ well-being. Many social media tools

are available over the Internet such as social network-

ing, professional networking, media sharing, content

production including blogs (e.g., Tumblr) and mi-

croblogs (e.g., Twitter), knowledge/information ag-

gregation (e.g., wikipedia), virtual reality and gaming

environments (e.g., second life). In particular, many

Healthcare Social Networking (HSN) platforms have

Fiumara G., Celesti A., Galletta A., Carnevale L. and Villari M.

Applying Artiﬁcial Intelligence in Healthcare Social Networks to Identity Critical Issues in Patientsâ

Z Posts.

DOI: 10.5220/0006750606800687

In Proceedings of the 11th International Joint Conference on Biomedical Engineering Systems and Technologies (HEALTHINF 2018), pages 680-687

ISBN: 978-989-758-281-3

emerged with the purpose to enhance patient care and

education. Popular HSB platforms include Sermo,

Doximity, Orthomind, QuantiaMD, WeMedUp, Dig-

ital Healthcare and so on. However, these social net-

works require the massive action of medical profes-

sional acting as moderators. In fact, healthcare so-

cial networks present potential risks for patients due

to the possible distribution of poor-quality or wrong

information. On one hand clinical operators want to

promote the exchange of information among patients

about a speciﬁc disease, but on the other hand they do

not have the time to read patients’ posts and moderate

them when required.

The main contribution of this paper is to propose

an Artiﬁcial Intelligence (AI) technique aimed at en-

hancing the well-being of patients participating in a

HSN platform. The basic idea is to allow an automatic

analysis of patients’ posts in order to identify possi-

ble critical issues so as to enable doctors to intervene

when required. In particular, our architecture includes

a Patients’ Posts Analysis System (PaPAS) acting as

a ﬁlter that is connected with a Complex Event Pro-

cessing (CEP) that sends warning and alert messages

with different level of risks to the medical personnel

who can reply to critical patients’ posts. In partic-

ular, in this paper, we focus on the PaPAS specif-

ically analysing the adopted AI approach based on

a combination of stemming, lemmatization and Ma-

chine Learning (ML) algorithms among others.

The remainder of the paper is organized as fol-

lows. Section 2 provides a brief overview of the ma-

jor recent initiatives in the ﬁelds of AI and social me-

dia for eHealth. The advantages and disadvantages

of adopting a HSN platform along with our reference

architecture are discussed in Section 3. Section 4 fo-

cuses on the Patients’ Posts Analysis System speciﬁ-

cally focusing on the adopted AI technique. A discus-

sion of possible application scenarios and future chal-

lenges is provided in Section 5. Section 6 concludes

the paper along with lights to the future.

2 RELATED WORK

Social media and healthcare quality improvement is

an emerging topic (Ranney and Genes, 2016). Re-

cently many scientiﬁc works have been proposed fac-

ing different aspects of social media for healthcare

purposes.

Regarding the social implication of such systems,

the beneﬁts, best practices, risks and ethical issues of

applying social media to healthcare professionals are

discussed in (Lee Ventola, 2014), (Hors-Fraile et al.,

2016), (Pinho-Costa et al., 2016), (Aboelmaged et al.,

2016), (Abbas et al., 2016). Social media can be

used to enhance professional networking and educa-

tion, organizational promotion, patient care, patient

education, and public health programs, but they can

also present several potential risks for patients regard-

ing the distribution of poor-quality information, dam-

age to professional image, breaches of patient privacy,

violation of personal-professional boundaries, and li-

censing or legal issues. However, social media are

also changing the healthcare industry (Opel, 2016)

and marketing (Malvey et al., 2015), (Koumpouros

et al., 2015). Evolution of social media in scientiﬁc

research in the domain of ICT and healthcare pro-

fessionals in Saudi Arabia Universities is discussed

in (Abdullatif et al., 2017), whereas a similar study

performed on young people in Russia is available in

(Bugrezova, 2016). Studies on the effectiveness of so-

cial media data in healthcare communication involv-

ing both medical personnel and patients is proposed in

(Saqib Nawaz et al., 2017a), (Huby and Smith, 2016),

(Smailhodzic et al., 2016), (Benetoli et al., 2017). The

role of social media in menopausal healthcare is dis-

cussed in (Short, 2017). There is also a strong corre-

lation among online data coming from search engines

and social media in the healthcare domain. In this re-

gard, in (Saqib Nawaz et al., 2017b) it is discussed

an approach for collecting twitter data by exploring

contextual information gleaned from Google search

queries logs.

Due to the huge amount of information to be anal-

ysed and processed, many scientiﬁc works have faced

the need of decisional support systems. In this re-

gard, a recommendation approach helping social me-

dia users to identify topics of interests is discussed in

(Li and Zaman, 2014). Such an approach was also

used for the assessment of user similarity in hetero-

geneous network with the purpose to look for people

that can give informational and emotional support in

a more efﬁcient way is discussed in (Jiang and Yang,

2017). Another user similarity study in healthcare so-

cial media using content similarity and structural sim-

ilarity is presented in (Jiang and Yang, 2015). A Study

on healthcare social media aimed at underserved com-

munities based on a mobile decision support system

(MDSS) providing information dissemination is dis-

cussed in (Miah et al., 2017). The need of pervasive

decision support systems in healthcare using intelli-

gent robots in social media is discussed in (Samad-

Soltani et al., 2017).

All aforementioned scientiﬁc works consider the

beneﬁts of using healthcare social media, but in many

cases require a strong interaction of the medical per-

sonnel. For the best of our knowledge a system that

ﬁlter for doctors only the patients’ posts that raises

Figure 1: General healthcare social network scenario including both clinical personnel and patients.

critical issues has not been proposed yet. In this pa-

per, we aim to overcome such a gap.

3 MOTIVATION

Social media applied in an healthcare context repre-

sent a tempting opportunity to improve the patients’

well-being promoting patient care and education. Fig-

ure 1 shows a general healthcare social network sce-

nario including both clinical personnel and patients

who interact by means of a HSN platform available

over the internet. Commonly, the major HSB plat-

forms require the massive action of medical profes-

sional who reply to patient’s queries also acting as

moderators on speciﬁc topics when it is required.

Healthcare social networks present several bene-

ﬁts including:

• promoting networking and information exchange

enabling self-education among patients about par-

ticular diseases.

• sharing patients’ experiences that can be helpful

for other ones;

• supporting the treatment process;

• reducing the patient’s stress when he/she is wait-

ing for a diagnosis or when he/she discovers to be

affected of a particular disease;

• promoting information gathering and prevention

campaign regarding speciﬁc diseases;

• optimizing the work of the clinical personnel who

interact with patients skilled on their diseases;

• promoting knowledge management;

• promoting research and monitoring activities.

All the aforementioned beneﬁts can potentially im-

prove the whole world healthcare education system.

On the other hand healthcare social networks present

several disadvantages including:

• possible distribution of poor-quality or wrong in-

formation among patients;

• the need of qualiﬁed medical personnel who

promptly read patients’ posts and who reply them;

• often the medical personnel do not have the time

to read patients’ posts and to reply them;

• the medical personnel do not want the responsi-

bility of the consequences on patients (worsening,

risk of death or death) when they do not reply in

time.

Figure 2: Patients’ Posts moderation architecture.

• possible legal issues for the medical personnel;

• risks for the reputation of the medical personnel.

A possible solution comes from the automatic

analysis of patients’ posts by means of Artiﬁcial Intel-

ligence (AI) techniques. In fact AI opens towards var-

ious application scenarios of patients’ posts analytics

including the identiﬁcation of critical posts that can

lead toward dangerous situations for patients them-

selves. In order to address such an issue, in this pa-

per, we propose a Patients’ Posts Moderator (PPM)

architecture whose basic ﬂowchart is shown in Figure

2. Both patients and medical personnel interact by

means of a HSN platform. A Patients’ Posts Analy-

sis System (PaPAS) works as batch process that con-

tinuously analyses patients’ posts of a HSN platform.

When a critical issue is detected, it generates an event

that is caught by a Complex Event Processing (CEP)

component that elaborates it and sends an alert mes-

sage to the interested medical personnel who can in-

tervene on the HSN platform, replying to critical pa-

tients’ discussion groups and/or triggering medical in-

terventions (doctors can directly contact the patient

or send ambulance with a medical equipment if re-

quired.).

4 APPLYING ARTIFICIAL

INTELLIGENCE IN HSN

PLATFORMS

In this Section, we speciﬁcally discuss an AI approach

on which PaPAS can be based. The main purpose of

PaPAS is to analyze patients’ posts and the evalua-

tion of possible critical issues that may trigger clin-

icians’ intervention. Figure 3 shows the PaPAS ar-

chitecture. It includes the following components: i)

Extractor, whose role is to extract patients’ content

from the HSN platform; ii) Selector, whose role is to

select relevant keywords; iii) Rank Generator, whose

role is to rank selected keywords; iv) Categorisator,

whose role is to categorise the various levels of seri-

Figure 3: PaPAS architecture.

ousness; v) Classiﬁcator, whose role is to classify pa-

tients’ posts according to different categories; and vi)

Evaluator, whose role is to assess results’ quality. A

detailed description of each components is provided

in the following Subsections.

4.1 Extractor

The ﬁrst component of our system deals with the ex-

traction of patients’ contents from HSN platform. The

accomplishment of this task is not unique because

it greatly depends on the architecture of the consid-

ered social network. In general, social networks pro-

vide APIs that help in the automatic extraction of spe-

ciﬁc information of interest. In some situations these

APIs may not be publicly available, or some infor-

mation (e.g. patients’ posts) cannot be extracted. In

these cases special-purpose pieces of software must

be used, generically referred to as wrappers. Thor-

Table 1: Features measured in a post.

Feature Deﬁnition

PostLength The number of words

Pos NumOfPos/PostLength, where NumOfPos is the number

of positive words/emoticons

Neg NumOfNeg/PostLength, where NumOfNeg is the num-

ber of negative words/emoticons

Name NumOfName/PostLength, where NumOfName is the

number of names mentioned

Slang NumOfSlang/PostLength, where NumOfSlang is the

number of Internet slangs

PosStrength Positive sentiment strength

NegStrength Negative sentiment strength

PosVsNeg (NumOfPos+1)/(NumOfNeg+1)

PosVsNegStrength PosStrength / NegStrength

Sentence The number of sentences

AvgWordLen The average length of words

QuestionMarks The number of question marks

Exclamation The number of exclamations

ough descriptions can be found in (Aggarwal and

Zhai, 2012; Liu, 2006).

4.2 Selector

After the extraction of patients’ posts, the next step

consists in the selection of the relevant keywords that

may trigger clinicians’ intervention. At the beginning,

stop-words are removed. Then, the Natural Language

Toolkit (NLTK)

stemming and lemmatization algo-

rithms are employed in order to reduce inﬂectional

forms and avoid the various syntactical forms a word

may have. Usually, the occurrence of syntactic vari-

ations of the same root form is less frequent in social

media posts with respect to long texts. Nevertheless,

the values of recall (see Subsection 4.6) is negatively

inﬂuenced by the presence of duplicates and/or syn-

tactic variations. Next, the identiﬁcation of relevant

keywords takes place. As usual, a Term Frequency -

Inverse Document Frequency (TF-IDF) approach has

been used (see for example (Liu, 2006)). Here, we

only recall that the importance of a term in a text is

evaluated according to the frequency with which it ap-

pears across multiple texts. Therefore, the Term Fre-

www.nltk.org

quency t f

of the i-th term in a text of n terms is given

max{ f

, f

, . . . , f

}

The Inverse Document Frequency id f

of the i-th term

within N posts in which the i-th term appears at least

once in df

posts is deﬁned as

idf

= log

The resulting TF-IDF term weight is given by:

= t f

· idf

4.3 Rank Generator

Beside the relevant keywords, also a sentiment analy-

sis of the posts is necessary in order to evaluate their

seriousness and therefore trigger the clinicians’s inter-

vention. This is important in order to disambiguate,

namely to discern situations in which a negative term

is used within a positive context (e.g., ‘a friend of

mine suffered from a stroke but after a while has re-

covered’) from situations in which a negative term is

used within a negative context (e.g., ‘a friend of mine

is having a stroke’). To this aim, we resorted to rely on

the work of (Qiu et al., 2011), in which the strength

of emotions in a post are measured as in (Thelwall

et al., 2010). All measures, as in (Qiu et al., 2011),

are summarized in Table 1. Here we want to under-

line the importance of some measures such as, for ex-

ample, PosVsNeg (the ratio of positive over negative

strength) which describes the overall tone of a post.

4.4 Categorizator

A big question is: how serious is the message con-

tained in a user’s post? Another interesting question

is: does it requires a clinician’s immediate interven-

tion? It would be highly desirable that only really

serious posts should trigger an alert that may cause

the intervention of a clinician. To this aim, the num-

ber of the levels of seriousness is a critical feature in

our platform. In fact, a small number of levels (e.g.,

only two) should increase the percentage of interven-

tions while a relatively high number of levels (e.g.,

four or ﬁve) should cause underestimates of the grav-

ity of a situation. Moreover, it has important con-

sequences on the Machine Learning (ML) algorithm

we adopted (see Subsection 4.5), because only su-

pervised algorithms allow to choose in advance the

number of classes in which clusterize patients’ posts.

Therefore, we decided to adopt three levels of seri-

ousness, that are: normal, warning, and critical.

4.5 Classiﬁcator

After having chosen the number of levels of serious-

ness, we classify patients’ posts according to the cate-

gories described in the Subsection 4.4. We devised to

adopt the Na

ıve Bayes Classiﬁer (NBC). It is a simple

yet powerful ML algorithm which embodies some de-

sirable features such as: i) it is extremely fast for both

training and prediction, ii) it produces simple prob-

abilistic prediction, and iii) it performs very well in

similar situations. As to the latter feature, in (Jain

and Kumar, 2015) is reported a similar study in which

posts extracted from Tweeter related to inﬂuenza pan-

demic are classiﬁed using various MLAs. From their

results appear that (NBC) outperforms other MLAs in

terms of precision, recall and F-Measure (see Subsec-

tion 4.6).

4.6 Evaluator

During the training phase, some measures are neces-

sary in order to assess the quality of the results. We

ﬁrst deﬁned the number of true positives (TP) as the

number of posts correctly labeled as belonging to the

positive class, the number of false positives (FP) as

the number of posts incorrectly labeled as belonging

to the postitive class. We also deﬁned the number

of true negatives (TN) as the number of posts cor-

rectly labeled as belonging to the negative class, and

the number of false negatives (FN) as the number of

posts incorrectly labeled as belonging to the negative

class. Having this in mind, as customary, we intro-

duced the precision p, the recall r and the F-Measure

F as

p =

T P

T P + FP

r =

T P

T P + FN

F =

2pr

p + r

5 DISCUSSION AND FUTURE

CHALLENGES

AI-based PaPAS also enables other applications sce-

narios considering a combination with emerging ICT

technologies such as Cloud computing, Edge comput-

ing, Cyber-Physical System, Internet of Things (IoT)

and Big Data analytics technologies.

In fact, AI applied to HSNs can trigger delivery of

various kinds of Cloud-based healthcare services and

applications over telecommunication networks and

the Internet aimed at providing assistance to patients

when warning and critical levels of seriousness occur.

The beneﬁt of adopting AI in telemedicine is twofold:

on one hand it can push down clinical costs and on the

other hand it can improve the quality of life of both

patients and their families. Telemedicine solutions

that can be triggered by AI mechanisms can be aimed

at tele-nursing, tele-rehabilitation, tele-dialog, tele-

monitoring, tele-analysis, tele-pharmacy, tele-trauma

care, tele-psychiatry, tele-pathology, etc.

Feedback provided by AI applied to HSN can

also control physical processes of patients consider-

ing Cyber-Physical System that on one side is con-

nected with an Healthcare Cloud provider (e.g., man-

aged by a Hospital or clinical centre) and on the other

side is connected with the patient by means of a series

of medical IoT devices deployed on him/her by means

of a personal body network including medical sen-

sors and actuators and/or deployed in medical devices

placed in the patient’s home on which the patient is at-

tached or monitoring and controlling the surrounding

environment.

Furthermore, AI applied to HSN opens towards

various scenarios of big data analytics. In fact, it

can allows researchers to study and understand the

origins, causes and diffusion over a wide geograph-

ical area of a particular disease besides understanding

their social implications.

6 CONCLUSION AND FUTURE

WORK

In this paper, we proposed an architecture aimed at

supporting the medical personnel in monitoring and

moderating patients of participating to HSN platform.

In particular, we focused on a PaPAS architecture

that implements the AI logic by means of a combina-

tion of stemming, lemmatization and Machine Learn-

ing (ML) algorithms among others. Speciﬁcally, the

aim of such a system is to enhance the well-being of

patients participating to the HSN platform. Basically,

PaPAS analyses patients’ posts in order to detect pos-

sible critical issues considering three levels of seri-

ousness, that are: normal, warning, and critical. If the

content of a post crosses the threshold of criticality,

the clinical personnel may promptly intervene.

Although our work is in a preliminary state, some

experiments have been carried out that demonstrate

effectiveness of the considered algorithms considered

alone. In particular, it was demonstrated that the

adopted ML algorithm (Na

ıve Bayes Classiﬁer) is fast

and reliable enough to allow real-time applications as

in critical environment.

In future work, we plan to analyse the perfor-

mance of the whole PaPAS considering a concrete

dataset including patients’ posts coming from a HSN

and to make a comparison with existing solutions

based on Deep Learning algorithms such as word-

embeddings and n-grams. Considering the huge

amount of posts that our system must be able to anal-

yse in a real HSN environment, in our implementation

we plan to consider big data analytics solution based

on Apache Hadoop and Spark. Future work will in-

clude the application of PaPAS in different healthcare

contexts.

REFERENCES

Abbas, A., Ali, M., Shahid Khan, M., and Khan, S. (2016).

Personalized healthcare cloud services for disease risk

assessment and wellness management using social

media. Pervasive and Mobile Computing, 28:81–99.

Abdullatif, A., Shahzad, B., and Hussain, A. (2017). Evo-

lution of social media in scientiﬁc research: A case of

technology and healthcare professionals in saudi uni-

versities. Journal of Medical Imaging and Health In-

formatics, 7(6):1461–1468.

Aboelmaged, M., Thomas, S., and Elsheikh, S. (2016).

Trends of social media applications in healthcare: A

managerial perspective.

Aggarwal, C. C. and Zhai, C. (2012). Mining text data.

Springer Science & Business Media.

Benetoli, A., Chen, T., and Aslani, P. (2017). How pa-

tients’ use of social media impacts their interactions

with healthcare professionals. Patient Education and

Counseling.

Bugrezova, E. (2016). The social media contribution

into healthcare practices among russian young people.

Ekonomicheskaya Sotsiologiya, 17(5):103–129.

ec.europa.eu (2012). European commission. ehealth

action plan 2012-2020: Innovative healthcare

for the 21st century [accessed on november

12th, 2017]. https://ec.europa.eu/digital-single-

market/en/news/ehealth-action-plan-2012-2020-

innovative-healthcare-21st-century.

ec.europa.eu (2014). European commission. trans-

formation of health and care in the digi-

tal single market [accessed on november

12th, 2017]. https://ec.europa.eu/digital-single-

market/en/policies/ehealth.

Hors-Fraile, S., Atique, S., Mayer, M., Denecke, K.,

Merolli, M., and Househ, M. (2016). The unin-

tended consequences of social media in healthcare:

New problems and new solutions. Yearbook of medi-

cal informatics, (1):47–52.

Huby, K. and Smith, J. (2016). Relevance of social media

to nurses and healthcare: ’to tweet or not to tweet’.

Evidence-Based Nursing, 19(4):105–106.

Jain, V. K. and Kumar, S. (2015). An effective approach

to track levels of inﬂuenza-a (h1n1) pandemic in india

using twitter. Procedia Computer Science, 70:801–

807.

Jiang, L. and Yang, C. (2015). Determining user similar-

ity in healthcare social media using content similarity

and structural similarity. Lecture Notes in Computer

Science (including subseries Lecture Notes in Artiﬁ-

cial Intelligence and Lecture Notes in Bioinformatics),

9105:216–226.

Jiang, L. and Yang, C. (2017). User recommendation in

healthcare social media by assessing user similarity

in heterogeneous network. Artiﬁcial Intelligence in

Medicine, 81:63–77.

Koumpouros, Y., Toulias, T., and Koumpouros, N. (2015).

The importance of patient engagement and the use of

social media marketing in healthcare. Technology and

Health Care, 23(4):495–507.

Lee Ventola, C. (2014). Social media and health care pro-

fessionals: Beneﬁts, risks, and best practices. P and

T, 39(7):491–499.

Li, J. and Zaman, N. (2014). Personalized healthcare rec-

ommender based on social media. pages 993–1000.

Liu, B. (2006). Web Data Mining: Exploring Hyperlinks,

Contents, and Usage Data (Data-Centric Systems and

Applications). Springer-Verlag New York, Inc., Se-

caucus, NJ, USA.

Malvey, D., Beardsley, A., Howse, P., and Ruff, S. (2015).

Healthcare marketing and social media.

Miah, S., Hasan, N., Hasan, R., and Gammack, J. (2017).

Healthcare support for underserved communities us-

ing a mobile social media platform. Information Sys-

tems, 66:1–12.

Opel, D. (2016). Ethical information ﬂows: Working

with/against the healthcare industry’s fascination with

social media.

Pinho-Costa, L., Yakubu, K., Hoedebecke, K., Laranjo, L.,

Reichel, C., Colon-Gonzalez, M., Neves, A., and Er-

rami, H. (2016). Healthcare hashtag index develop-

ment: Identifying global impact in social media. Jour-

nal of Biomedical Informatics, 63:390–399.

Qiu, B., Zhao, K., Mitra, P., Wu, D., Caragea, C., Yen,

J., Greer, G. E., and Portier, K. (2011). Get online

support, feel better – sentiment analysis and dynamics

in an online cancer survivor community. 2011 IEEE

Third Int’l Conference on Privacy, Security, Risk and

Trust and 2011 IEEE Third Int’l Conference on Social

Computing, pages 274–281.

Ranney, M. and Genes, N. (2016). Social media and health-

care quality improvement: A nascent ﬁeld. BMJ Qual-

ity and Safety, 25(6):389–391.

Samad-Soltani, T., Rezaei-Hachesu, P., and Ghazisaeedi,

M. (2017). Pervasive decision support systems in

healthcare using intelligent robots in social media.

Iranian Journal of Public Health, 46(1):148–149.

Saqib Nawaz, M., Bilal, M., Ikram Ullah Lali, M., Mustafa,

R., Aslam, W., and Jajja, S. (2017a). Effectiveness

of social media data in healthcare communication.

Journal of Medical Imaging and Health Informatics,

7(6):1365–1371.

Saqib Nawaz, M., Mustafa, R., and Lali, M. (2017b). Role

of online data from search engine and social media in

healthcare informatics.

Short, H. (2017). The role of social media in menopausal

healthcare. Post Reproductive Health, 23(1):4–5.

Smailhodzic, E., Hooijsma, W., Boonstra, A., and Langley,

D. (2016). Social media use in healthcare: A sys-

tematic review of effects on patients and on their re-

lationship with healthcare professionals. BMC Health

Services Research, 16(1).

Thelwall, M., Buckley, K., Paltoglou, G., Cai, D., and

Kappas, A. (2010). Sentiment in short strength de-

tection informal text. J. Am. Soc. Inf. Sci. Technol.,

61(12):2544–2558.

www.grandviewresearch.com (2016). Grad view re-

search. ehealth market projected to reach usd

308.0 billion by 2022 [accessed on november 12th,

2017]. http://www.grandviewresearch.com/press-

release/global-e-health-market.