The Need for Trustworthiness Models in Healthcare Software

Solutions

Raja Manzar Abbas

, Noel Carroll

1,2

, Ita Richardson

1,2

and Sarah Beecham

Lero- The Irish Software Research Centre, University of Limerick, Limerick, Ireland

ARCH- Applied Research for Connected Health Technology Centre, University of Limerick, Limerick, Ireland

Keywords: Trustworthiness, Healthcare Software, Trustworthiness Attributes.

Abstract: Trustworthiness in software is of vital importance to technology users, especially in health, where lives may

depend on the correct application of software that is fit for purpose. Despite the risk posed by improper use

of technology in the health domain, there is evidence to suggest that stakeholders often trust the software

without fully appreciating the possible consequences. In this paper, we explore what determines

trustworthiness in healthcare software solutions. While there are often claims of improved quality of care,

increased safety and improved patient outcomes using healthcare technology – the scientific basis for such

claims appear to be uncritically accepted. Ultimately, this can lead to a surge in healthcare software solutions,

some of which may be misaligned with healthcare needs and potentially lead to fatal outcomes. To support

health technology stakeholders, we propose a ‘trustworthiness healthcare software model’ that can be

employed to assess the level of trustworthiness associated with healthcare software solutions.

1 INTRODUCTION

The United Nation’s International Standard Industrial

Classification, (2016) categorizes healthcare as

generally consisting of hospital activities, medical and

dental practice activities. Implementations of

potentially transformative healthcare technologies are

currently underway internationally, often with

significant impact on national expenditure. For

example, Ireland has invested approximately €900

million in its e-health while the UK has invested at

least £12.8 billion in a National Programme for

Information Technology (NPfIT) for the National

Health Service. Similarly, the Obama administration

in the United States has committed to a US$38 billion

Healthcare investment (Catwel et al., 2009).

Such large-scale expenditure has been justified on

the grounds that electronic health records (EHRs),

picture archiving and communication systems

(PACS), electronic prescribing (ePrescribing) and

associated computerised provider (or physician) order

entry systems (CPOE), and computerised decision

support systems (CDSSs) will help address the

problems of variable quality, safety and trust in the

modern health care. However, the scientific basis of

achieved quality and trust – which are repeatedly

made and are seemingly uncritically accepted –

remains to be established (Huckvale et al., 2012;

Institute of Medicine, 2007).

1.1 Problem Statement

The ultimate goal of software is to help end-users to

accomplish their tasks in a convenient and efficient

manner. However, the literature suggests that

software technology advancements in healthcare

often failed to ease the lives of the healthcare

professionals. Instead, healthcare professionals often

report a loss of productivity while using healthcare

software. This leads to a lack of trust in the healthcare

software (Velsen et al., 2016).

1.2 Research Question

In this paper we examine the literature on

trustworthiness in healthcare and look particularly at

the associated attributes. We also explore the need for

a healthcare software model of trustworthiness.

Considering the broad and vast nature of software

technology use in healthcare, we argue that

stakeholders need to have a set of criteria by which

they can assess the level of trustworthiness of a given

technology.

Abbas R., Carroll N., Richardson I. and Beecham S.

The Need for Trustworthiness Models in Healthcare Software Solutions.

DOI: 10.5220/0006249904510456

In Proceedings of the 10th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2017), pages 451-456

ISBN: 978-989-758-213-4

451

There is an apparent lack of insight into what a

trustworthiness healthcare software model should

capture, and how it should be applied. To address

these gaps, we formulate the following research

questions:

 RQ1. What are the key attributes that define

trustworthiness of healthcare software?

 RQ2. What are the current models or

frameworks that capture trustworthiness of

healthcare technology?

1.3 Methodology

To explore these questions, we undertook a structured

literature review. A structured literature review may

be described as appraisals of past studies conducted

systematically, purposefully and methodologically

(Armitage and Keeble-Allen, 2008; Petticrew, 2001).

In the research discussed in this article, a literature

search was completed in the bibliographic databases

ACM Digital Library, IEEE, Springer LINK

and Google Scholar, using the keyword search

phrases ‘trustworthiness’, ‘healthcare software’,

‘healthcare trust’, ‘trustworthiness models’,

trustworthiness frameworks’, ‘trust attributes’,

‘trustworthy attributes’, ‘software trustworthiness’

and ‘healthcare software trustworthiness’. 2894

initial reference sources were found. From these, after

screening titles and abstracts , 2224 were deemed not

eligible. Out of remaining 670 research articles, 536

articles were screened out after applying the exclusion

criteria on the titles and abstracts - 238 were not

relevant to software engineering, 107 research articles

had no specific intervention about software

trustworthiness (trustworthy, trust), 187 articles did

not mention software attributes and/or models and 4

research articles were not written in English. After

reviewing the full text of the remaining 134 studies,

83 more studies were excluded due to lack of

relevance to the topic and 51 studies were selected as

primary studies.

2 IMPACT OF HEALTHCARE

TECHNOLOGY

Due to the growth in population and shift in

demographics, there is a significant pressure on the

global healthcare system. Shojania et al. (2016),

attribute a toll to medical error of 251,454 deaths in

US hospitals per year, making, they say, medical error

the third-leading cause of death in the USA. The

Institute of Medicine study estimated the cost of

nonfatal medical errors is between $17 billion and $19

billion each year, and that between 2.9% and 3.7% of

all patients admitted suffer some type of injury

because of medical mismanagement. As a result, there

is a growing focus on healthcare technology to offer

greater service efficiency and it has given rise to a

comprehensive sociotechnical model for managing

healthcare through software solutions.

Technological advances have encouraged the

development of new technologies that drive

connectivity across the healthcare sector—apps,

gadgets, and systems that personalize, track, and

manage care using just-in- time information

exchanged through various patient and community

connections (Leroy et al., 2014).

This paradigm shift heavily emphasizes the

process of software development in healthcare

systems. It has also contributed to a shift in healthcare

practice, highlighting our growing reliance and

trustworthiness of software to support healthcare

decisions. However, trusting the healthcare software

solution without validating can have serious and

potentially fatal consequences (Carroll, 2016).

3 TRUSTWORTHINESS – WHO

CARES?

Trustworthiness in healthcare software is the sum of

trust in different factors. The composition these

factors can differ for different healthcare users. For

example, for patients, trustworthiness in software

consists of, mostly, a perceived level of control and

privacy, while for healthcare professionals, a larger

and different set of issues play a role, including

reliability and a transparent data storage policy. The

set of factors that affect trustworthiness in a healthcare

portal are different from the sets that have exist for

general software domain. There is a need to study

trustworthiness in healthcare software as a separate

subject to inform the design of reliable interventions.

3.1 Need for Trustworthiness

Healthcare Software Model

With significant growth in healthcare software

solutions, software is having an increasing impact on

clinical decisions and diagnosis. However, there is

little evidence as to the trustworthiness of software.

For example, a glitch in St. Mary’s Mercy Medical

Centre’s (Cork, Ireland) patient management system

“killed” 8,500 patients on paper (National Computer

Security Center, 1985). When St. Mary’s upgraded its

HEALTHINF 2017 - 10th International Conference on Health Informatics

452

patient management software, there was a mapping

error in the alteration process that triggered the

program to notify 8,500 patients of their incorrect

death.

While unregulated medical devices rarely find

their way to patients, the same cannot be said about

the largely unregulated market for health applications

and software. As such, in reality, there exists a

considerable gap between the potential benefits that

software’s could provide, and what healthcare

professionals are currently likely delivering in

practice. Recent reviews in the therapeutic areas of

bulimia (Nicholas et al., 2015), asthma (Huckvale et

al., 2012), Post-traumatic stress disorder (PTSD)

(Olff, 2015), insulin dosing (Huckvale, 2015) and

suicide prevention (Larsen et al., 2016) have yielded

worrying conclusions regarding the quality, scientific

basis and often blatant disregard for safety (The Daily

Mail, 2014),

These errors go some way to illustrating the need

for trustworthy healthcare software. Although

different researchers have tried to address some

attributes of trustworthiness not all attributes have

been identified. We discuss some of the models and

standards and the attributes they cover in an effort to

understand what areas they are lacking in.

3.2 Impact of Trustworthiness on

Quality

The number of medical errors caused by devices

(which have embedded software) and software

applications naturally leads to the questions:

 How can healthcare software be made

trustworthy?

 What process/mechanism would achieve this?

 How do we inform users and healthcare

providers which healthcare software solution

can be trusted and why?

We first identified the attributes of trustworthy

healthcare software and why there is a need for a

trustworthy healthcare software model.

4 DEFINING

TRUSTWORTHINESS

According to Merriam–Webster Dictionary

(2004), trustworthy means ‘worthy of confidence’.

For software products, researchers and practitioners

have a slightly different understanding of ‘trustworthy

software systems’, since we need to view

trustworthiness over time.

Trustworthiness is defined by Amoroso et al.

(1994) as a “level of confidence or degree of

confidence” and software trustworthiness is defined as

a “degree of confidence that the software satisfies its

requirements”. Since the definition is expressed as a

“degree of confidence”, Amoroso and Taylor

illustrates trustworthiness is dependent upon

management and technical decisions made by

individuals or groups of individuals evaluating the

software. Software trustworthiness is expressed in

terms of a set of requirements, where the ‘set’ is

variable. For example, trustworthiness may be

dependent on the set of functional requirements, or

may be a critical subset of functional requirements, or

it may be some set of requirements that include non-

functional assurance requirements like safety or

security (Amoroso et al., 1994).

In his ICSE 2006 Keynote speech, Boehm (2006)

pointed out the increasing trend of software criticality

and dependability as one of the key software trends.

Over the past 50 years, different strategies such as

formal methods, security assurance techniques, defect

prediction, failure mode and effects analysis, testing

methods, and software assurance techniques have

been proposed to address different aspects of software

trustworthy challenges. Based on these studies,

numerous quality categories and attributes have been

studied as major factors influencing on software

trustworthiness. Among them are included

functionality, reliability, safety, usability, security,

portability, and maintainability, etc.

In addition, Zhang et al. (2012) reviews the

appropriateness of the software attributes summarized

by Yang et al. (2009), and suggests that the

trustworthiness of software is related to the following

set of properties which they redefined to address the

trustworthiness context as:

 Safety

 Validity

 Reliability

 Reusability

 Scalability

 Maintainability

 Performance

Carbone et al. (2013) defined trustworthiness from

information and communication technology (ICT)

systems where security challenges include both

confidentiality (or privacy) and in integrity (or trust)

of the data. In particular, the notion of trustworthiness

seems relevant for tagging databases and electronic

patient records with information about the extent to

which test results, diagnoses and treatments can be

trusted.

The Need for Trustworthiness Models in Healthcare Software Solutions

453

Initial findings from literature suggest a lack of

understanding of trustworthiness in the healthcare

software domain and a need for a trustworthiness

process model that define standards or best practices

about the trustworthiness of healthcare software.

In this position paper, we propose the key

attributes that define trustworthiness of healthcare

software and current models that capture

trustworthiness of healthcare technology.

5 THEORETICAL

FOUNDATIONS

Though there is a growing consensus that

trustworthiness is characterized as one that satisfies a

collection of critical quality attributes, yet, there is a

lack of common understanding of healthcare software

trustworthiness – particularly in a healthcare context.

Based on the literature and the sample of definitions

introduced here, we identify that the key factors of

trustworthiness for healthcare software should be

regulation, confidence of the users and meet its

requirements and objectives in a satisfactorily

manner.

5.1 Theoretical Influences on

Developing a Trustworthiness

Healthcare Software Model

The Capability Maturity Model Integration (CMMI)

guides organisations with a view to ensuring that the

correct software is being developed throughout each

stage in the development cycle and conforms to

specification. However, it is important to realise that

software process models, such as CMMI, do not cover

healthcare regulations, and that they need to be used

in conjunction with the regulations (Burton et al.,

2006).

There have been some new developments in this

area, for example the development of MDevSPICE

(formally known as MediSPICE) (Clarke et al., 2014;

McCaffery et al., 2010). The MDevSPICE framework

is one of the first attempts to address the safety

concerns faced by healthcare software producers and

presents a software safety assessment process.

Verification and validation activities are very

important in software development and can consume

much of a project’s costs and effort. While

verification and validation are addressed by process

models and standards for both generic and safety

critical software development, there are still

challenges in undertaking its successful

implementation as part of the software development

process. The process of verification and validation

requires a clear understanding of how each activity is

undertaken and related to each other, which is

important in a healthcare environment (Carroll and

Richardson, 2016).

For example, the development of an international

software process improvement (SPI) framework for

the medical device industry acts as a key enabler of

best practice for the healthcare sector. SPI techniques

offer a continuous cycle of performing an assessment

and restarting the cycle (McHugh et al., 2012) with

the aim of reducing defective software. Software may

also be vulnerable to outside attack. Many hospitals

and healthcare facilities use various threat

management software and firewalls to monitor their

mobile device applications to ensure that they are

secure and safe. In most cases, within the USA, this is

a requirement of Health Insurance Portability and

Accountability Act (HIPPA).

HIPAA is a framework which is followed by

number of organisations for maintaining the security

and privacy of the health information. HIPAA came

into force in 1996 to address a number of concerns,

most notably the need for increased protection of the

medical records of the patients against unauthorised

access (Wu et al., 2012). HIPAA provides a national

standard for electronic healthcare transactions. It also

provides regulations regarding healthcare information

security and privacy (Jepsen, T, 2003). HIPAA covers

entities such as healthcare providers, insurers and

providers of health plan. Healthcare organisations are

now required to individually assess their security and

privacy requirements using various auditing tools.

Healthcare technology systems have access to

personal identifiable information.

Our traditional view of privacy protection

methods through various anonymization techniques

does not provide an efficient way to deal with the

privacy of technological healthcare software

solutions. For example, in response to growing

concerns on privacy and data security, in 2014, the

European Commission published a Green Paper on

mHealth (European Commission, 2014). Through

wide stakeholder consultation, the paper discusses the

main barriers and issues related to mHealth

deployment. They highlight a number of key topics

including data protection, security of health data,

informed consent, big data management, patient

safety and transparency of information across the EU

and, ultimately, on the need to regulate mHealth

applications.

One of the main concerns across industry is the

lack of a unified model which can incorporate all of

HEALTHINF 2017 - 10th International Conference on Health Informatics

454

the best practices for healthcare software

development. There is a need to formulate a

healthcare software model that can accurately

propagate trustworthiness throughout the process.

5.2 Towards Developing a

Trustworthiness Healthcare

Software Model

By carefully reviewing the appropriateness of the

attributes summarized through a software lens, we

suggest that the trustworthiness of healthcare software

model is related to the following set of properties:

 Security: inclusion of security mechanisms in

the model with respect to access control

processes.

 Efficiency: effectiveness of the model

construction that is able to give a quick

response or reaction with minimal resources

and/or time taken.

 Safety: inclusion of semantics that represent

process requirements related to safety, and the

ability to highlight inconsistencies in the

process model with respect to safety-related

processes.

 Functionality: the functions at the level

expressed in functional requirements of the

model, emphasizing at the level of final user

functionality

 Reliability: the probability of the process

model delivering results that is consistent with

the model assumptions.

 Regulation: decision support reference model

that will ensure that healthcare software

products are safe and effective to protect and

promote public health through various

standards and regulations.

 Validity: the ability of the process model to

reflect the assumptions and constraints about

the software process specified by process

stakeholders.

 Accuracy: the measurement tolerance, or

transmission of the process model that defines

or removes the limits of the errors.

6 FUTURE RESEARCH

Having established a foundation for the

Trustworthiness Healthcare Software Model by

identifying the key attributes of trustworthiness from

a healthcare software perspective, we will continue to

build on this to establish key processes and metrics

within the model.

As part of our future research, we will examine

and modify existing trustworthiness models. The

subsequent focus will be on extending and modifying

existing techniques based on our identified attributes

for the analysis. Then our next step will be to take the

trustworthiness model that we develop, to test and

refine it on a large scale with healthcare software

sector. This way, we will also be able to say which

attributes are the most important.

7 CONCLUSIONS

The primary goal for adopting healthcare software is

to provide patients the best service possible by

gathering and interpreting accurate information. This

should help them to take correct and timely decisions

which reduces cost, time and effort, thereby resulting

in the timely treatment of the patient. But, there are

apparent concerns regarding whether we can trust

healthcare software solutions.

We have identified that there is a gap, and

therefore, a consequent need to introduce a

Trustworthiness Healthcare Software Model. In this

study, we have focused on an initial definition of

trustworthiness attributes from literature. We

highlight our next steps towards the development of

the Trustworthiness Healthcare Software Model and

its validation across the healthcare software sector.

ACKNOWLEDGEMENTS

“This work was supported with the financial support

of the Science Foundation Ireland grant 13/RC/2094

and co-funded under the European Regional

Development Fund through the Southern & Eastern

Regional Operational Programme to Lero - the Irish

Software Research Centre (www.lero.ie)”.

REFERENCES

Amoroso, E., Taylor, C., Watson, J. and Weiss, J., 1994,

November. A process-oriented methodology for

assessing and improving software trustworthiness.

In Proceedings of the 2nd ACM Conference on

Computer and communications security pp. 39-50.

Armitage, A and Keeble-Allen, D. 2008."Undertaking a

structured literature review or structuring a literature

review: Tales from the field", Electronic Journal of

The Need for Trustworthiness Models in Healthcare Software Solutions

455

Business Research Methods 6 (2), pp. 103-114.

Available www.ejbrm.com.

Boehm, B., 2006, May. A view of 20th and 21st century

software engineering. In Proceedings of the 28th

international conference on Software engineering pp.

12-29. ACM.

Burton, J., McCaffery, F. and Richardson, I., 2006, May. A

risk management capability model for use in medical

device companies. In Proceedings of the 2006

international workshop on Software quality pp. 3-

8.ACM.

Carbone, M., Christensen, A.S., Nielson, F., Nielson, H.R.,

Hildebrandt, T. and Sølvkjær, M., 2013, August. ICT-

powered Health Care Processes. In International

Symposium on Foundations of Health Informatics

Engineering and Systems pp. 59-68. Springer Berlin

Heidelberg.

Carroll, N. 2016. Key Success Factors for Smart and

Connected Health Software Solutions, Computer, Vol.

49, No. 11, pp. 32-38.

Carroll, N. and Richardson, I., 2016. Software-as-a-Medical

Device: Demystifying Connected Health

Regulations. Journal of Systems and Information

Technology, 18(2).

Catwell, L. and Sheikh, A., 2009. Evaluating eHealth

interventions: the need for continuous systemic

evaluation. PLoS Med, 6(8), p.e1000126.

Clarke, P., Lepmets, M., McCaffery, F., Finnegan, A.,

Dorling, A. and Flood, D., 2014, November.

MDevSPICE-a comprehensive solution for

manufacturers and assessors of safety-critical medical

device software. In International Conference on

Software Process Improvement and Capability

Determination pp. 274-278. Springer International

Publishing.

European Commission (2014), “Green paper on mobile

health (‘mHealth’)”, COM 219 final, SWD 135 Final,

available at: http://ec.europa.eu/digital-agenda/en/

news/green-paper-mobilehealth- mhealth (accessed

November, 2016).

Huckvale, K., Adomaviciute, S., Prieto, J.T., Leow, M.K.S.

and Car, J., 2015. Smartphone apps for calculating

insulin dose: a systematic assessment. BMC

medicine, 13(1), p.1.

Huckvale, K., Car, M., Morrison, C. and Car, J., 2012. Apps

for asthma self-management: a systematic assessment

of content and tools. BMC medicine, 10(1), p.1.

Huckvale, K., Car, M., Morrison, C. and Car, J., 2012. Apps

for asthma self-management: a systematic assessment

of content and tools. BMC medicine, 10(1), p.1.

Institute of Medicine (2007) Preventing medication errors.

Washington (D.C.): National Academy Press.

Jepsen, T., 2003. IT in healthcare: progress report. IT

professional, 5(1), pp.8-14.

Larsen, M.E., Nicholas, J. and Christensen, H., 2016. A

systematic assessment of smartphone tools for suicide

prevention. PloS one, 11(4), p.e0152285.

Leroy, G., Chen, H. and Rindflesch, T.C., 2014. Smart and

connected health. IEEE Intelligent Systems, 29(3), pp.2-

McCaffery, F., Dorling, A. and Casey, V. (2010), “Medi

SPICE: an update”, available at: http://

eprints.dkit.ie/48/ (accessed 14 November 2016).

McHugh, M., McCaffery, F. and Casey, V., 2012. Software

process improvement to assist medical device software

development organisations to comply with the

amendments to the medical device directive. IET

software, 6(5), pp.431-437.

Merriam-Webster, 2004. Merriam-Webster's Collegiate

Dictionary. Merriam-Webster.

Nicholas, J., Larsen, M.E., Proudfoot, J. and Christensen,

H., 2015. Mobile apps for bipolar disorder: a systematic

review of features and content quality. Journal of

medical Internet research, 17(8).

Olff, M., 2015. Mobile mental health: A challenging

research agenda. European journal of

psychotraumatology, 6.

Petticrew, M.A. (2001)."Systematic literature reviews from

astronomy to zoology: Myths and misconceptions",

British Medical Journal 322 (7278), pp. 98-101.

Available www.bmj.com/cgi/contact/full/322/7278/98.

Shojania, K.G. and Dixon-Woods, M., 2016. Estimating

deaths due to medical error: the ongoing controversy

and why it matters. BMJ Quality & Safety, pp.bmjqs-

2016.

The Daily Mail: Health warning over blood pressure

monitoring apps as doctors warn they are ‘untested,

inaccurate and potentially dangerous’. Available

at: http://www.dailymail.co.uk/sciencetech/article-

2887791/Health-warning-blood-pressure-apps-doctors-

warn-untested-inaccurate-potentially-dangerous.html.

United Nations. International Standard Industrial

Classification of All Economic Activities, Rev.3. New

York.

Van Velsen, L., Wildevuur, S., Flierman, I., Van Schooten,

B., Tabak, M. and Hermens, H., 2016. Trust in

telemedicine portals for rehabilitation care: an

exploratory focus group study with patients and

healthcare professionals. BMC medical informatics and

decision making, 16(1), p.1.

Wu, R., Ahn, G.J. and Hu, H., 2012, January. Towards

HIPAA-compliant healthcare systems. In Proceedings

of the 2nd ACM SIGHIT International Health

Informatics Symposium pp. 593-602. ACM.

Yang, Y., Wang, Q. and Li, M., 2009, May. Process

trustworthiness as a capability indicator for measuring

and improving software trustworthiness.

In International Conference on Software Process pp.

389-401. Springer Berlin Heidelberg.

Zhang, H., Kitchenham, B. and Jeffery, R., 2012. Toward

trustworthy software process models: an exploratory

study on transformable process modeling. Journal of

Software: Evolution and Process, 24(7), pp.741-763.

HEALTHINF 2017 - 10th International Conference on Health Informatics

456