Big Data Analytics in healthcare: The path to a business case

(1)

Big Data Analytics in healthcare:

The path to a business case

Naoufal Boussallami Student number: 10783296 University of Amsterdam Faculty of Science

Thesis Master Information Studies: Business Information Systems Final version: 12-07-2018

Supervisor: T.M. (Tom) van Engers Examiner: A.M. (Loek) Stolwijk

Abstract. Big Data Analytics projects offer the opportunity to improve care and become increasingly an important strategic choice for healthcare organizations. This study developed a Big Data Analytics business case framework that helps healthcare organizations with the structured approach of an (explorative) Big Data Analytics project. The framework is built on the basis of a generic IT business case framework that has been tested and adapted for Big Data Analytics projects on the basis of interviews with 14 Big Data Analytics experts in healthcare. This framework contains the different aspects that healthcare organizations should take into account during the business case phase before starting a Big Data Analytics project. Focus areas within the framework are marked to increase the chances of success of Big Data Analytics. The framework contains the following eight dimensions: (1) Process, (2) Data Governance, (3) Ethics, (4) Stakeholders, (5) Risks, (6) Content, (7) Goals and (8) Application Area. Of these, the first four dimensions are highlighted as focus areas.

Keywords. Big Data Analytics investments, Big Data Analytics business cases, Big Data Analytics approach, Healthcare.

(2)

2

2.4 Business cases ... 5 2.5 IT business cases ... 6 3. Methodology ... 6 3.1 Methods ... 7 3.2 Participants ... 7 4. Results ... 8 4.1 Application area ... 9 4.2 Process ... 10 4.3 Goals ... 13 4.4 Stakeholders ... 13 4.5 Risk factors ... 13 4.6 Content ... 14 4.7 Data Governance ... 16 4.8 Ethics ... 17

4.9 Critical success factors ... 17

4.10 Focus areas ... 19

6. Discussion ... 19

7. Conclusion ... 20

8. Acknowledgments ... 21

References ... 22

Appendix 1 - IT business case framework of Maes, Van Grembergen & De Haes (2014) ... 24

Appendix 2 - Interview questions ... 30

Appendix 3 - Interview questions case study ... 33

Appendix 4 - Big Data Analytics business case framework ... 34

Appendix 5 - Feasibility study ... 40

Appendix 6 - Cost analysis ... 41

Appendix 7 - Benefit analysis ... 42

Appendix 8 - Team composition ... 44

(3)

3

1. Introduction

Big Data initiatives provide potential through data-driven decisions, exceptional insights, the ability to measure and monitor factors that were previously considered abstract, better understanding of products and customers and discovering new sales opportunities (Rajpurohit, 2013). Manyika et al. (2011) estimate that the US health care expenditures could be reduced by eight percent by using Big Data effectively and creatively to improve the efficiency and quality. IDC (2017) expects that the Big Data market will have a compound annual growth rate (CAGR) of 11.9% through 2020 with revenues over $210 billion. Three of the top 10 technology trends for 2017 given by Gartner (2016) are explicitly related to the use of Big Data. According to Gartner, most of the world’s largest top 200 companies will utilize the full toolkit of Big Data Analytics in order to refine their offers (Gartner, 2016). Big Data is not just relevant for big organizations. Due to the fact that tools and platforms for Big Data Analytics are relatively affordable, many small-to-midsize businesses also are able to manage and leverage Big Data (Rossum, 2011).

Big Data Analytics is one of the most important IT innovations for healthcare organizations

(Raghupathi & Raghupathi, 2014).Alexander & Wang (2018) state the following: “data-driven management in healthcare has become a strategic choice in achieving sustainable growth, meeting the challenges of global competition, and explore the potential innovation for the future”. Big Data Analytics solutions are according to Alexander & Wang (2018) key to advancing healthcare systems. Multiple scholars state that the healthcare industry lags behind other industries in taking advantage of analytical tools and methods (Fihn et al., 2014; Ferranti, Langman, Tanaka, McCall, & Ahmad, 2010), which implies that healthcare organizations have to excel in the area of Big Data Analytics in order to be able to use the full potential.

Despite the fact that the Big Data market is growing fast and the positive predictions about the growth of this market, organizations are struggling to deliver the promised value (Return On Investment) (Rajpurohit, 2013). Different scholars discuss the importance of making a business case for a Big Data investment in organizations in order to estimate the added value and expectation in advance (Rajpurohit, 2013; Jokonya, 2015; Shim, French, Shim & Jablonski, 2015). A business case is used as a structural proposal for improvement that functions as a decision package for decision makers within organizations. It contains an analysis of the needs or problems of business processes, proposed alternative solutions, constraints and risk- adjusted cost-benefit analysis (Weaver & Sorrells-Jones, 2007). Gartner (2015) estimated that 60 percent of Big Data projects fail to go beyond piloting and experimentation and will be abandoned afterwards. Furthermore, organizations target a return of $3,50 for every dollar invested in Big Data project, but only manage to yield a return of $0,50 (Shim et al., 2015). This implicates that there is a mismatch between the expectations on the returns of Big Data investments and practice. Multiple scholars state that the comprehensiveness of a business case

significantly affects the success of IT investments and that a business case is a key criterion for project success (Ward, Daniel & Peppard, 2008; Fortune & White, 2006; Berghout & Tan, 2013).

Current research about methods for building a business case are mainly focused on information systems investments in general (Remenyi & Sherwood-Smith, 2012; Schmidt, 2003; Gambles, 2009; Ward et al., 2008). These methods offer an overview of different elements that should be present in a business case for an IT investment. This paper extends the existing body of knowledge of business case methodologies by creating a Business Case Framework for Big Data Analytics investments within healthcare organizations. This framework is focused specifically on the elements that constitute a business case for Big Data Analytics investments in healthcare, including focus areas based on critical success factors for Big Data Analytics projects. The

framework will assist project managers in healthcare organizations with the constitution of a business case and provides insight into the most important facets of a Big Data Analytics project and how to approach it. The framework should lead to better expectation management of stakeholders and gives healthcare organizations more grip on how to approach Big Data Analytics in a structured way.

This research explores the dimensions for a business case of a Big Data Analytics investment with the following research question: What aspects should a healthcare organization take into account during the business case phase before starting a Big Data Analytics project?

At first, a literature review is performed for framing the applied definitions concerning Big Data Analytics. This is necessary because of the ambiguous definitions on this subject. In addition, applications of Big Data Analytics projects in the healthcare sector are discussed to gain basic knowledge about possible

applications within healthcare. Because of the recent attention about privacy and security in the media and its relevancy as indicated by scholars, privacy and security are discussed in the context of Big Data Analytics projects. Furthermore, the current fragmented body of knowledge about IT business cases is discussed to get a good picture of the available knowledge about this subject. Based on the gained knowledge from the literature review, semi-structured interview questions will be drawn in order to develop the Big Data Analytics business case framework. Section three contains the research method, where the scoping of the research is determined,

(4)

4

including the design of the research. The fourth section outlines the results based on the performed interviews and finally contains the newly developed Big Data Analytics business case framework. Finally, the discussion contains critical reflections on this research, after which the conclusions of this research are drawn.

2. Literature review 2.1 Big Data Analytics

This section defines the concept of Big Data Analytics in order to get a clear understanding of this concept within this research context. This is necessary to frame the ambiguous definitions and buzzwords surrounding this field.

The concept of Big Data is closely related to Big Data Analytics, which are needed to create value of the data (Janssen, Van der Voort & Wahyudi, 2016). Big Data Analytics can be separated in two technical entities: Big Data and Analytics. Big Data refers to the massive amounts of detailed information, which is described further in detail in the next section. Analytics refers to Advanced Analytics, which actually is a collection of different tool types, including tools based on data mining, predictive analytics, statistics, artificial intelligence and natural language processing. These two technical elements together form the concept of Big Data Analytics (Russom, 2011).

Big Data refers to datasets whose size is beyond the ability of typical database software tools to store, manage and analyze (Manyika et al., 2011). Big Data cannot be defined in terms of a certain number of terabytes. Manyika et al. (2011) state that the size of the datasets that are qualified as Big Data change as technology advances over time. A commonly used definition of Big Data are the three V’s of Gartner. Gartner stated that Big Data can be characterized and identified by the following three V’s: Volume, Variety and Velocity. Volume is the amount of data which is stored within information systems that is caused by transaction volumes and other common data types. Variety is the diversity of the data. Examples are databases, audio, images, documents and e-mails. Velocity is the speed at which the data is retrieved and how fast it has to be processed to meet the demand (Stamford, 2011). IT company IBM extended this definition with the fourth V: Veracity, which describes the uncertainty about the quality of the stored data and the data coming out of the predictive model (Puget, 2015).

2.2 Big Data Analytics in healthcare

This section describes the role of data within healthcare organizations and examples of Big Data Analytics use cases. Raghupathi & Raghupathi (2014) applied the 4 V’s of Gartner, which are described in the previous section, to Big Data within healthcare organizations. The volume within healthcare context refers to the health-related data that will be created an accumulated continuously, which results in preposterous volumes. The veracity of these existing healthcare data include the different types of data, like structured, unstructured and semi-structured data. Structured data are data that can be easily stored, queried, analyzed and manipulated by computers (Raghupathi & Raghupathi, 2014). Examples of structured and semi-structured data are personal medical records, national health register data and clinical data from clinical decision support systems. Examples of unstructured data are handwritten physician notes, paper prescriptions, X-Ray reports, laboratory data, MRI, CT and other images. Other examples of unstructured data are patient data that are machine generated and/or sensor data and social media data (Archenaa & Mary Anita, 2015; Raghupathi & Raghupathi, 2014). The velocity of healthcare data refer to the constant flow of new data accumulating at extraordinary rates presenting new challenges. These challenges lie primarily in retrieving, analyzing, comparing and making medical related decisions based on the output of these data, while the data has the high velocity at which it is generated (Raghupathi & Raghupathi, 2014).

Big Data within healthcare organizations is overpowering, because of the volume of the data, the diversity of data types and the speed at which it has to be managed (Raghupathi & Raghupathi, 2014).

In general, healthcare organizations rely on Big Data Analytics technologies to capture all of these data about a patient, in order to get a complete view for insight into care coordination (Archenaa & Mary Anita, 2015) or lower costs while improving the care process (Patil & Seshadri, 2014). Healthcare organizations, ranging from single-physician offices to large hospital networks, can stand to realize significant benefits by digitizing, combining and effectively using Big Data (Raghupathi & Raghupathi, 2014). Leveraging the collection of patient and practitioner data offer the possibility to improve the quality and delivery of healthcare (Murdoch & Detsky, 2013). Potential benefits are expanding the capacity to generate new knowledge and detecting diseases at earlier stages in order to treat them more easily and effectively (Archenaa & Mary Anita, 2015; Raghupathi & Raghupathi, 2014; Murdoch & Detsky, 2013). Furthermore, Big Data Analytics offers the

(5)

5

possibility to predict or estimate certain developments or outcomes based on historical data. Examples are predicting the patients at risk for certain medical complications, length of stay of patients, choices about elective surgery, pinpointing patients who are likely to not benefit from certain surgery and diseases/illness progression (Raghupathi & Raghupathi, 2014). Besides analyzing historical data, real-time data can be applied to detect diseases as early as possible, identifying them expeditiously and applying the right treatments. Performing this real-time could reduce patient mortality and morbidity and even prevent hospital outbreaks (Raghupathi & Raghupathi, 2014).

2.3 Big Data Analytics Data Protection

With the recent implementation of the General Data Protection Regulation (GDPR) in Europe, the processing and storage of data has become hot topic in the media. In addition to the fact that data is widely discussed in general in the media, the relevance of data protection is discussed in academic literature in the context of Big Data Analytics projects (Murdoch & Detsky, 2013; Strang & Sun, 2017; Chen et al., 2014; Kambatla et al., 2014; Pence, 2014; Patil & Seshadri, 2014, Tene & Polonetsky, 2011, Alexander & Wang, 2018). While the healthcare sector increases the use of Big Data Analytics, security and privacy issues are important as emerging threats and vulnerabilities continue to grow (Patil & Seshadri, 2014). Due to this increased importance, this section discusses the aspect of data protection within the context of Big Data Analytics.

According to Pence (2014), the use of Big Data can be a serious threat to personal privacy. As stated before in this research, Big Data is characterized by the large amounts and variety of data. Data such as phone calls, social network posts, digital search and cell phone geolocation can be combined to create an

unprecedented window into the private life of people (Pence, 2014). Sullivan (2012) states that almost all of the personal information is provided online will end up being bought and sold, segmented, packaged, analyzed, repackaged, and sold again. Furthermore, public records like birth data, criminal records, political affiliation and real estate records have been analyzed and are being combined with online personal data (Pence, 2014). Besides these online data, healthcare organizations face an increase in the volume of data in terms of diversity, complexity and timeliness (Patil & Seshadri, 2014). Furthermore, the security of data is also an important issue. There were multiple incidents of cyber security hackers who broke into electronic databases, linking fields together and leverage data to obtain confidential information (Strang & Sun, 2017). This implies that combining data can lead to unwanted privacy-sensitive insights, which is very possible with the sensitive data that

healthcare organizations possess.

According to Patil & Seshadri (2014), traditional security solutions are not directly applicable to large and inherently diverse data sets. Several data protection related examples of solution directions have been given in the literature. Firstly, (Big) Data Governance is indicated by Patil & Seshadri (2014) as required prior exposing data to analytics. According to Patil & Seshadri (2014), Data Governance will be the first step in regulating and managing healthcare data, which is seen as necessary before analyzing the data. Meingast, Roosta and Sastry (2006) indicate that the following aspects need to be considered when managing data in healthcare: (1) data ownership, (2) what type of data and how much data should be stored, (3) type of data storage and (4) who can access what (access control). In addition, various methods for de-identification can be used to distance data from real identities in order to prevent privacy-related issues. Examples of usable techniques are: anonymization, pseudonymization, encryption, key-coding and data sharing (Tene & Polonetsky, 2011).

2.4 Business cases

This section provides background information about business cases in general. This indicates the importance of business cases in general within organizations and positions business cases within the processes within

organizations.

A business case outlines the justification of a project and is a document that sets out the perceived benefits, requirements and constraints of a project. In effect, a business case outlines the “why”, “what”, “when” and “how much” questions of a project (Lester, 2014; Maes, Grembergen & De Haes, 2014).

Furthermore, the business case describes the impact that the project is expected to have on the organization from a business standpoint. Business cases are often obligatory within organizations and rely upon an economic analysis in order to make a selection between competing alternatives and to determine whether a specific project is justifiable from a cost-benefit standpoint (Heerkens, 2002; Weaver & Sorrells-Jones, 2007; Nielsen & Persson, 2017). In essence, a business case provides evidence that the sponsors who invest in the recommended initiative will see a return on investment within a reasonable timeframe (Weaver &

(6)

Sorrells-6

Jones, 2007). The document itself is the responsibility of the sponsor but is often a combined effort by the sponsor of the project and the project manager (Lester, 2014).

2.5 IT business cases

This section elaborates on the previous section which contains general knowledge about business cases. Based on that knowledge, this section discusses business cases for IT projects and explains the importance and added value of IT business cases. Finally, the existing and fragmented knowledge about the constitution of IT business cases is structured and discussed.

Different scholars discuss the added value of a business case in IT projects. IT business cases capture the rationale behind IT projects and safeguard against resource wastages by ensuring that project outcomes satisfy important business motives (Berghout & Tan, 2013). Business cases determine the success of IT investments by empowering organizations during three critical project phases: (1) taking informed decisions regarding IT projects, (2) monitoring the progress of projects and (3) evaluate project outcomes (Berghout & Tan, 2013; Ward, Daniel & Peppard, 2008; Maes, Van Grembergen & De Haes, 2014). Berghout & Tan (2013) concluded that comprehensive IT business cases have higher initial cost estimates and they suggest that this improves investment decisions. Furthermore, business cases should lead to improvements in project governance and decrease the failure rates of IT investments (Berghout & Tan, 2013).

Current research of IT business cases are focused on the different elements that should be present in the business case document. Few studies have tried to clarify how business cases can be developed

comprehensively (Remenyi & Sherwood-Smith, 2012; Schmidt, 2003; Gambles, 2009; Ward et al., 2008). Although these multiple attempts, knowledge on business case research is scattered throughout literature (Maes, Van Grembergen & De Haes, 2014; Berghout & Tan, 2013). In order to tackle the problem of scattered knowledge about what actually constitutes a business case and to resolve misunderstanding among scholars, Maes, Van Grembergen & De Haes (2014) have conducted a systematic literature review in a selection of top academic and practitioner journals based on journal ranking publications. The most important findings were that the applications of business cases is useful in a broad range of investment contexts. They also found sufficient argumentation that using a business case continuously throughout an entire project life cycle can increase the investment success rate. This concurs with the findings of Berghout & Tan (2013). Furthermore, the business case should contain a richer set of information rather than only financial numbers. The

involvement of stakeholder is key when developing and using business cases.

Maes, Van Grembergen & De Haes (2014) constituted a holistic IT business case framework. Their IT business case framework organized fragmented knowledge about business cases in six dimensions: (1) Application area, (2) Process, (3) Content, (4) Risk factors, (5) Stakeholders and (6) Goals. Each dimension is further subdivided into sub-dimensions to structure text fragments within each dimension. A complete overview of the content elements of each sub-dimension can be found in Appendix 1.

Some of the dimensions are directly linked to the three stages of an investment life cycle: before, during and after implementation (Maes, Van Grembergen & De Haes, 2014). The stage “before

implementation” is focused on the feasibility precede the investment approval. The stage “during

implementation” starts when the investment decision has been made and the organization starts to invest the appropriate resources in the project. This stage ends then the investment has officially been launched. This launch implicated that the planned applications have been implemented and all organizational changes have been produced. The organization can now start using the result of the project. Finally, the post-implementation review is considered to be part of the “after implementation” stage (Maes, Van Grembergen & De Haes, 2014). 3. Methodology

This section addresses the demarcation and approach of this research. The literature review offered a clear overview of the role of a business case within the project lifecycle. The six dimensions from the IT business case framework of Maes, Van Grembergen & De Haes (2014) are used as a theoretical framework. The applicability and completeness of these six dimensions in the context of Big Data Analytics projects in healthcare are tested during this research. In order to build a comprehensive and practically usable Big Data Analytics business case framework, the content of the required business case elements were also determined. In addition to the constitution of a Big Data Analytics business case, focus areas were designated on the basis of critical success factors of Big Data Analytics projects. The focus areas aim to highlight the most important aspects of a business case in order to increase the success rate of Big Data Analytics projects. The Big Data Analytics business case framework provides therefore insight into the most important facets of a Big Data Analytics project and ho.

(7)

7

3.1 Methods

Semi-structured in-depth interviews with experts were held in order to develop the Big Data Analytics business case framework. Details about the participants can be found in section 3.2. The interviews had a duration varying from one to one and a half hour. The interview questions were derived from the literature review. The questions have been formulated to test the framework of Maes, Van Grembergen & De Haes (2014).

Furthermore, questions have been formulated to determine the critical success factors. Finally there are questions formulated about the positioning of privacy and security in the context of Big Data Analytics projects. There has deliberately chosen to ask questions explicitly about the positioning of privacy and security within the framework, because the literature emphasizes the importance of these aspects. The coding scheme that was used for coding the interviews can be found in Appendix 9.

The interviews were divided into three parts: (1) general questions for background information about the interviewee (2) identifying the Big Data Analytics business case dimensions, including their content and (3) determining the critical success factors. The complete questionnaire of the interview is included in Appendix 2.

The first part of the interview was in the context of obtaining background information about the expert. These questions were focused on testing the extent to which the person is actually familiar with the Big Data Analytics concept as used within this research context. Furthermore, questions about recent project experience related to Big Data Analytics were asked. This was necessary to test the practical experience of the expert, so that afterwards could be determined if the interview provided sufficient added value for this research.

The second part of the interview consisted of questions around the six dimensions of the IT business case framework of Maes, Van Grembergen & De Haes (2014). During that part, the existing IT business case framework was tested in the context of Big Data Analytics projects. Superfluous and missing (sub)dimensions were identified during this phase in order to build the new business case framework. Furthermore, the content of the dimensions were determined. Questions during this part of the interview were more directive, so that the interviewee had the space to address points that were not explicitly included in the existing framework. This part of the interview was therefore clearly (consciously) less structured. The structure of the interviews regarding to the questions asked differed therefore from person to person.

During the third part of the interview, the critical success factors were identified and discussed in order to determine the focus areas. This part of the interview was unstructured since many follow-up questions were asked about the mentioned critical success factors. Due to the emerging stage Big Data Analytics projects are currently in, with the result that there are few leading projects within the healthcare sector, there is deliberately chosen not to substantiate this part of the research statistically with for example surveys.

At first, one interview was held to lay the foundation in terms of basic knowledge about Big Data Analytics projects, so that the right questions could be asked to the subsequent participants. The questions during this interview were more substantive focused on the project and their critical success factors, so the interviewee could indicate how Big Data Analytics projects are approached in practice and what the most important facets are. This can be seen as a light version of a case study, but only has the purpose to provide basic knowledge about Big Data Analytics projects to ask specific questions to subsequent participants. This case study has not been further elaborated in this study. The list of interview questions for this specific interview is included in Appendix 3.Furthermore, an iterative approach was used to ensure the quality of the developed the Big Data Analytics business case framework. The framework of Maes, Van Grembergen & De Haes (2014) was validated during each interview. In parallel to the interviews, the new framework was

iteratively developed on the background to keep track of which new insights have been obtained per interview. The iteratively developed framework was therefore continuously validated on the background after each interview until theoretical saturation was reached. The requirement for theoretical saturation was that the last two interviews did not offer new insights on the framework. A total of 14 interviewees were required until theoretical saturation was reached.

3.2 Participants

In order to develop the Big Data Analytics business case framework, 23 Big Data Analytics experts were contacted for interviews. Of these, in total 14 were available and interviewed. These Big Data Analytics experts met at least one of the following criteria within the context of a healthcare organization:

1. Practical experience with constituting a business case for Big Data Analytics projects 2. Practical experience with leading Big Data Analytics projects

3. Practical experience with implementing Big Data Analytics projects

An obligatory requirement was that the participant had in-depth knowledge of the concept of Big Data Analytics.

(8)

8

At first, experts within the personal network were approached. Started from there, a snowball

sampling approach is used to collect subsequent experts. The criteria described above have been applied during snowball sampling. The already interviewed experts were also requested to introduce other experts within their network who could contribute to this research. Because these experts already had knowledge about this research, they could better target who was relevant. For this method of sample collection, the previous named criteria have also been passed on to introduce subsequent experts. Furthermore, experts were approached via LinkedIn. These experts have been found after querying on the following job titles in

combination with the keyword “healthcare”: 1. Big Data expert

2. Big Data lead

3. Big Data project manager 4. Senior Big Data consultant 5. Senior Big Data engineer 6. Big Data architect 7. Big Data Analytics 8. Data scientist 9. E-health

Each LinkedIn profile has been extensively studied and assessed before contact has been made. Due to the ambiguity of the definition Big Data Analytics, an extra query is performed on job titles that contain the following keywords in combination with healthcare:

1. Data & Analytics 2. Data Analytics 3. Advanced analytics 4. Analytics

These keywords were combined with the function groups as described in the above list with job titles. This to not exclude people due to a perception difference of the definition of Big Data Analytics. For all of the contacted experts via LinkedIn, as well via snowball sampling, the audit question was raised whether they meet the aforementioned criteria about their practical experience.

The 14 interviewees are included anonymously in this study. Table 1 contains a subdivision of the roles/functions of the participants. All participants have experience or a specialization in healthcare and are working at prominent consultancy or healthcare organizations. The literature review emphasized the importance of privacy and security in Big Data Analytics. Therefore, one Big Data Analytics legal expert in healthcare has been interviewed in order to validate statements from other interviewees about privacy and security processes and to provide further information about privacy and security processes in healthcare.

Table 1. Subdivision roles/functions of the participants

Roles/functions Number of participants

(Senior) Data Scientist 4

Project manager 3

Big Data Analytics lead/manager 3 (Senior) Big Data Analytics consultant 3 Big Data Analytics legal expert 1 Total: 14

4. Results

The section provides the results based on the interviews that have been conducted. The aim of this section is to develop the new Big Data Analytics business case framework based on the results derived from the interviews. Each of the existing dimensions and sub-dimensions in the business case framework of Maes, Van Grembergen & De Haes (2014) were tested on their applicability within Big Data Analytics projects and expanded or adapted where necessary. Figure 1 contains a graphical overview of the dimensions of the new Big Data Analytics business case framework developed in this research. The dimensions in which a red asterisk is included are new dimensions that have been added based on the interview results. The green dimensions are the focus areas of the framework determined on the basis of the critical success factors mentioned during the interviews. A complete overview of the sub-dimensions and their content elements of each dimension of the new Big Data Analytics business case framework can be found in Appendix 4.

(9)

9

Firstly, the applicability to Big Data Analytics projects and the content of each dimension by Maes, Van Grembergen & De Haes (2014) are tested on the basis of the interview results. Thereafter, the two new dimensions, Data Governance and Ethics, which can be derived from the interview results are discussed and the content of these dimensions is determined. Thereafter, the critical success factors for Big Data Analytics project are discussed that were mentioned during the interviews. Finally, these critical success factors are translated into the focus areas of the framework.

Figure 1. Graphical overview dimensions new Big Data Analytics business case framework 4.1 Application area

“Technological orientation” and “Organizational orientation” are the two sub-dimensions that have already been included in the framework of Maes, Van Grembergen & De Haes (2014). The interviews have shown that these two sub-dimensions are still applicable within Big Data analytics projects, although it is indicated that the current given content is not specific enough. The content that can be applied according to the respondents is described later in this section. The respondents indicated that a third and new dimension applies within the healthcare sector. This sub-dimension is called "Medical orientation" and contains examples of common medical processes where Big Data Analytics can be applied according to the respondents. The new content elements of each sub-dimension is described below.

Technological orientation

“E-health records” and “Global data synchronization network” are the only two technological applications that have already been included in the framework of Maes, Van Grembergen & De Haes (2014) and are marked by the interviewees as still applicable. A general remark is made about “E-health records”, because E-health is a broader definition of the use of technology to support or improve healthcare. E-health is more an umbrella term for different technological applications. Examples of further specification of E-health given during the interviews are: gathering and processing care data, lifestyle data and medicine data. “Global data

synchronization network” is further specified as: sharing (big) data within a network of healthcare organizations or other organizations within a (private) network, like healthcare insurance organizations or pharmaceutical companies. The following new applicable technological applications are according to the interviewees the most common in practice: developing a data platform (infrastructure). This infrastructure is required to store (big) data efficiently so that it can be used by different end-users. Another technological application is developing an analytics platform to be able to analyze the stored (big) data. This platform makes it possible to convert the data into useful information for the development of, for example, predictive models. Finally, the implementation of analytics cloud solutions make it possible for an organization to start quickly, without having to purchase the necessary hardware.

(10)

10

Organizational orientation

The existing mentioned content within the sub-dimension “Organizational orientation” have been further specified by the respondents. The interviewees have stated that the content within this sub-dimension should be focused on the different business processes that are present in a healthcare organization. The medical processes are divided into a separate dimension (see next section). Examples of organizational oriented applications are: optimizing (medical) administrative processes, optimizing planning processes and optimizing financial processes.

Medical orientation

The content within this new sub-dimension are focused on the application of Big Data Analytics within medical processes. Examples of these application that have emerged during the interviews are: monitoring systems for discharge forecasts, predicting psychiatry treatments, determining oncological treatments, digital radiology (image analysis), digital pathology (tissue samples) and predicting residivity.

4.2 Process

This dimension is focused on the various process steps that need to be taken to develop a business case. “Before implementation”, “During implementation” and “After implementation” are the three sub-dimensions that have already been included in the framework of Maes, Van Grembergen & De Haes (2014). The already existing content elements within the “Process” dimension are according to the respondents generally focused on Waterfall oriented projects. These projects are more performed sequentially and are often carried out according to the project management methodology "Prince2". Prince2 is characterized by planning in detail and identifying any other project details in advance. The interviewees indicated that, as opposed to

"traditional" IT projects, it is often difficult to determine in advance what the actual (tangible) result will be of Big Data Analytics projects, because it is often discovered in the data. It is also indicated as a "dream" that several stakeholders must work on together. In a more (innovative) R & D setting, together with the

stakeholders, new insights are sought in the data and together with the stakeholders will be determined how these insights can be implemented in the business or medical processes. The interviews have shown that Big Data Analytics projects should therefore be approached more iteratively instead of sequentially to achieve an optimal learning curve. The interviewees have furthermore indicated that the entire project should be done in a short cycle, so that ambiguities can emerge at an early stage. The results about the necessary process steps for Big Data Analytics business cases emerged during the interviews surrounding each of the three sub-dimensions will be discussed below.

Before implementation

The already existing process steps that need to be taken before the implementation and to develop a business case were assessed during the interviews. The interviewees were asked how Big Data Analytics project can be approach the best and what aspects should be taken into account in advance. The process steps that have been marked as important by the respondents can be grouped in the following four phases: (1) Exploratory phase, (2) Assessment phase, (3) Formalization phase and (4) Decision-making phase. Figure 2 shows a summary of the most important aspects of each phase. Each phase, including a brief overview of the required process steps that need to be taken before an implementation, will be discussed below. Despite the fact that the steps appear to be sequential at first glance, it should be emphasized that these steps should be performed short-cyclically.

(11)

11

Figure 2. Overview identified process business case process steps

Phase 1 - Exploratory phase

The process steps in the first phase stand in the context of exploring the why, who and what part. At first, the project manager should understand the investment relevance by understanding the reason why the project should be implemented. Possible reasons that are mentioned during the interviews are: market trends, strategic foundations or care issues. In addition, the interviewees also indicated that there are two most common reasons for a Big Data Analytics project in healthcare. The first one is that there is already a (business) question available. Healthcare organizations already have a problem that they want to tackle with Big Data Analytics. The other reason is that a healthcare organization has a lot of data in a certain area and want to explore what the possibilities are by using that available data (explorative approach). According to the

interviewees, this explorative approach often takes place in academic hospitals. After the reason of the project is clear, the project manager must then carry out the stakeholder analysis in order to identify the relevant stakeholders, including the project sponsor, and what their importance is for the project. All interviewees indicated that this is a crucial step and should be carried out at the very beginning of the project and that the stakeholder should stay involved from the beginning. Details about the stakeholder analysis can be found in the dimension “Stakeholders” in section 4.4. In the extension with the stakeholder analysis, the problems must be identified, including the requirements from the different stakeholders. Based on the input from the

stakeholders, a global design must be made of the possible solution for a solution direction, including the solution alternatives. The respondents have indicated that during the initial phase of a project, in which solutions are devised, data protection and ethical issues should already be taken into account. This can be done by, for example, involving the Data Protection Officer and an ethicist. In order to be able to determine the chances of success of the project, the project manager has to assess what the current level of support is within the organization and increase this as much as possible. The current maturity in the field of Big Data Analytics should be a starting point for determining the project objectives, so that realistic expectations and ambitions can be created. According to the respondents, good expectation management and realistic objectives

contribute to the success of the project. Furthermore, the project manager must identify various obstacles and resistance and try to eliminate them as much as possible. Due to the fact that Big Data Analytics projects are often new to organizations, the project manager must also remove as much uncertainty as possible by, for example, taking stakeholders to referees, develop contingency plans, gaining experience with one specific use-case or executing a proof of concept (see next phase). At the end of the first phase, the project manager has to check whether the project has a chance of success, to determine whether it pays to further shape the business case.

Phase 2 – Assessment phase

During the second phase, the project manager has to determine whether the project is feasible in practice. In order to prevent reinventing the wheel, the respondents have advised that project managers should

investigate what other healthcare organizations have done in this area. In consultation with the IT department, a technology choice should be made. Furthermore, all respondents have indicated that a proof of concept is

(12)

12

almost always carried out, so that the feasibility can be determined and that the organization obtains the first experiences with the solution directions. Before the proof of concept is executed, the project manager must determine which resources are needed for this. A feasibility study is performed, in order to determine the organizational feasibility, technical feasibility and legal and ethical feasibility. The financial feasibility is also tested at a high level. The detailed cost benefit analyses takes place in the next step. Details about the aspects to be assessed during the feasibility study can be found in Appendix 5. After the feasibility study has been carried out and the project is considered feasible, the project manager must determine which resources are needed for the actual project (team composition) and check whether they are available. These are resources from both IT and medical specialists. The respondents emphasized that these resources must commit to the project and have to reserve time for the project.

Phase 3 – Formalization phase

After the previous phase has shown that the project is feasible in practice, the project manager needs to formalize the business case so that it can be offered for review. For this, a cost-benefit analysis must be performed in detail. More details on the aspects that have been identified as important during the cost analysis can be found in Appendix 6. By carrying out a realistic cost analysis, it becomes clear in advance what costs the healthcare organization should take into account when carrying out the project and who will pay for the solution. By performing the benefit analysis, the project manager can then gather the right arguments to convince decision-makers to carry out this project. More details relevant aspects about the benefit analysis can be found in Appendix 7. Furthermore, the project manager should perform a what-if analysis and failure analysis, including a suitable backup plan. This is by two respondents labeled as required, so that an

organizations can fall back on the old situation if the project results are disappointing in practice. Finally, the project manager should finalize the business case by writing the business case document based on the obtained insights during the preliminary phase of the project. The content of this document is discussed in section 4.6.

Phase 4 – Decision-making phase

After the business case has been written, a decision must be made by the decision makers and project sponsors. The business case should be evaluated by the stakeholders before it is communicated with the decision makers. After the business case has been communicated with the makers, official decision-making takes place. Despite the fact that the interviewees had overall knowledge about the decision-decision-making process, none of the interviewees were decision-makers who had in-depth knowledge about the most

important aspects of the decision-making process and the actual assessment. Therefore, there are no details on the actual assessment of a Big Data Analytics business case derived from the research results, which means that this phase, in contrast to the other three phases, cannot be further developed in detail.

During implementation

The interviewees emphasized the fact that the business case needs to be updated regularly on the basis of the new insights obtained during the project phase. These changes are mainly the changing needs of the

stakeholders or the upscaling and downscaling of resources. These elements have already been named in the existing framework but have been emphasized during the interviews. What also is emphasized is the fact that the stakeholder must remain involved during the implementation phase. Their opinion, commitment and conformation of the business case should be identified and assured during each iteration from the beginning.

Something that is not explicitly mentioned within this sub-dimension in the existing framework, but which has been indicated by the interviewees, is the fact that the detailed planning of the project needs to be adapted short-cyclically and is drawn up during the implementation phase. The interviewees also indicated that the detailed (technical) solutions of the project should be conceived and adapted in a short-cyclical way, based on the changing needs. However, the interviewees indicated that both the detailed planning and detailed solutions should not be included in the business case document itself. Finally, the interviewees extended the current content elements with the fact that healthcare organizations should check after each iteration whether the project should be continued and what the scope will be.

After implementation

In contrast to the necessary steps before the implementation of a Big Data Analytics solution, the steps after the implementation of Big Data Analytics solution were not discussed extensively during the interviews. This is mainly due to the fact that the focus of this research was more in the run-up to a project. The interviewees also

(13)

13

indicated that after each iteration it must be determined whether or not to continue with a project. However, the interviewees indicated that the lessons learned should be collected and shared within and outside the organization.

4.3 Goals

“Before implementation”, “During implementation” and “After implementation” are the three sub-dimensions of the dimension “Business case goals” that have already been included in the framework of Maes, Van Grembergen & De Haes (2014). The interview results showed that there are no changes to the content of this dimensions. However, new insights were given about the role of a business case in Big Data Analytics project in healthcare. One of the comments, for example, is that in a business case the rationale of the project must be discussed, but not the complete implementation and step-by-step plan of the project. The approach must be scalable and adaptable, as the project requires at that moment. Furthermore, more insight was given into the decision-making process surrounding Big Data Analytics projects in healthcare. In practice, Big Data Analytics is something that healthcare organizations should do and it is being more of a no-brainer. Therefore, the business case for Big Data Analytics will be more on sub investment level and not about to assess whether they should start a Big Data Analytics project or not. Business cases can therefore be used to determine whether the healthcare organization should select a certain use-case or not. In addition, business cases can be used to approach a project in a structured manner and to decide which sub investment a healthcare organization should do or not. The question of whether Big Data Analytics projects can have added value is therefore, in contrast to other branches, less of the order. Assessing Big Data Analytics investments is mentioned as more a formality. Most healthcare organizations have budgets for Big Data Analytics projects. A business case according to one interviewee, is more a tool for checking stakeholder opinions, assigning responsibilities and starting a proof of concept. Since insufficient in-depth statements were made during the interviews

surrounding the decision-making process, these statements cannot be fully generalized within this research. 4.4 Stakeholders

The dimension “Stakeholders” contains in the current framework the following three sub dimensions: “Before implementation”, “During implementation” and “After implementation”. The current subdivision of potential stakeholders per sub-dimension given by Maes, Van Grembergen & De Haes (2014), is according to the respondents implicitly focused on Waterfall oriented projects. As mentioned by the respondents can the Waterfall project management approach be characterized by planning in detail and eliminating all uncertainties, so that most stakeholders no longer need to be involved during the project. All respondents emphasize that stakeholder management is crucial within Big Data Analytics projects and that this should be done at the very beginning, especially before determining a solution. They have furthermore emphasized that all of the stakeholders must remain involved throughout each phase of the project. The current subdivision to the sub dimensions “Before implementation”, “During implementation” and “After implementation”, and the associated identified stakeholders per project phase, are according to the respondents therefore no longer applicable.This means that the current subdivision to different types of stakeholders per project phase no longer applies and that all of the stakeholders mentioned in the new framework must remain involved during each phase of the project. Therefore, no subdivision of sub dimensions are required in the new framework.

The interviewees have defined the goal of the stakeholder analysis as follows: “to identify all of the involved stakeholders and what their importance is for the project in question”. The interviewees have stated that the stakeholders differ per project and that it is context dependent. Although that the relevant

stakeholders are context dependent, there are certain stakeholder groups that often can be identified. During the stakeholder analysis, project managers should at first identify the group who would benefit from the proposed project and who will use the data. This could be a patient group, nurses or medical experts. Other common important stakeholders are: health insurers, IT department, the government, cooperation partners and external IT suppliers. Furthermore, project managers should identify who the decision makers are and who is going to sponsor the project. These sponsors or decision makers could be someone in the higher

management (director) or someone in the middle management. But also in this case, the exact stakeholders depend on the context and organization.

4.5 Risk factors

The current existing sub-dimensions within the dimension “Risk factors” are: “Before implementation”, “During implementation and After implementation”, “Environmental influences to business case” and “Business case limitations”. No comments were made by the interviewees on the already existing risk factors in one of the

(14)

14

existing sub-dimensions. However, new risk factors including their potential impact have been identified to extend the current list with risk factors of the sub dimension “Before implementation”. These new risk factors will be discussed within this section. Furthermore, this dimension has been extended with one sub dimension called "Project risks". The risks within this sub-dimension are in the context of risks that can impede the success of the project. The risks, including their potential impact, identified for this new sub dimension are also

discussed.

Before implementation

Quantifying the benefits of Big Data Analytics projects within healthcare is marked as a risk, because there are a number of challenges in this area. The biggest challenge is the financial impact of health prevention. Healthcare organizations "earn" money by providing care. By reducing the number of treatments, certain Big Data Analytics solutions can provide a Business case that is less interesting from a financial perspective case. Assessing Big Data Analytics business cases in healthcare only from a financial perspective is therefore mentioned as not sufficient. The respondents indicated that Big Data Analytics business cases should be assessed more on qualitative criteria, such as increasing the quality of care and the added value for patients. Furthermore, the incompleteness of the cost analysis is marked as a risk. This can give a distorted picture of reality, which means that the project costs can turn out higher than expected. A Big Data Analytics solution that is not scalable is also marked as a risk. Because certain parts cannot be reused due to a lack of scalability, it could mean that certain costs have to be doubled during other Big Data Analytics projects within the same healthcare organization, which result in a less attractive business case. Another business case risk is when the ambition level is too high in the business case regarding the objectives and the project manager does not take into account the current maturity of the care organization in the area of Big Data Analytics. This could ensure that the uncertainty from the stakeholders increases and that there is less support and adoption.

Project risks

One of the project related risks mentioned is insufficient knowledge in the field of Big Data Analytics. This may partly be related to the recruitment policy but also the retention policy. The result is that the learning curve becomes flatter and that the maturity in the area of Big Data Analytics does not increase, because knowledge is lost from the organization. Lack of technical knowledge within the IT department in the area of Big Data Analytics is also mentioned as a project risk. This can provide an inhibiting effect for the project. Good collaboration between business and IT is also important. The interviewees emphasized that a good balance should be sought here and that the IT department should not be leading. Instead, it has to be driven more from the business and IT has to take a facilitating role. Furthermore, a bad collaboration between stakeholders is mentioned a risk because it can cause a delay and discussion. If the sponsor is not enthusiastic or when the problem owner is not involved, it might become a hurdle further on in the project. A lack of scalability of the chosen solution is also marked as potential project risk. The result of a non-scalable solution is that the solution cannot be used on a larger scale or in other situations. The lack of expectation management beforehand is also mentioned as a project risk as it can result in disappointing project results. Not checking whether the proposed Big Data Analytics solution is ethically responsible or in accordance with legislation, poses a risk to the actual implementation of the project, because the project will eventually be cut off or may not be implemented in the production environment. Finally, a lack of Data Governance is mentioned as a project risk, which result in unusable data and not being able to fully use the data

4.6 Content

This dimension contains the different elements that should be present in a business case document. Although the exact business case document template may differ per healthcare organization, the interviewees were asked which elements should be present in the business case document for a Big Data Analytics project in healthcare. The current framework contains the following seven sub dimensions: (1) Investment description, (2) Investment objectives, (3) Investment requirements, (4) Investment impact, (5) Investment risks, (6) Investment assumptions, considerations and scenarios and (7) Investment governance. No comments on the current division of sub-dimensions emerged from the interviews. However, new content has been added to the existing sub-dimensions. The biggest changes are the further specification of content already mentioned, such as the costs with which a healthcare organization must take into account and what the potential benefits are. The results from the interviews are discussed per sub dimension.

(15)

15

Investment description

“Project planning and roadmaps” is the only content element within this sub dimension where a change needs to be made. The interviews showed that Big Data Analytics projects should be the best approached iteratively, instead of the entire approach being mapped out in detail in advance. For this reason, this content element needs to be replaced by the objectives of the Big Data Analytics project. The detailed planning of the project will be determined during the project.

Investment objectives

Two new content elements within this sub dimension have been identified. The first one is that within this sub dimension the link with the strategic objectives should be taken into account. The project manager must indicate what the project contributes to the strategy of the healthcare organization. The second one is that the project manager must indicate how the proposed project will improve the healthcare quality. So what are the goals in terms of improving the care provided.

Investment requirements

“Customer needs” is one of the existing content elements within this sub dimension. Because this framework is focused on health care, two interviewees indicated that this may be renamed to "Patient needs". Furthermore, two new content elements have been identified. "Legal requirements" are requirements that are set by the government. These can be requirements in the broadest sense of the word. However, in practice at Big Data Analytics it will mainly be requirements in terms of data storage and data processing. The second new content element is called “Ethical requirements”. These can be ethical requirements specified by the ethicist involved. In healthcare there is a committee called "Medical Ethics Committee", which tests every project in terms of ethics before it is implemented. By adding this content element, the project manager can include these ethical requirements in advance.

Investment impact

None of the existing content elements within this sub dimension are marked as unnecessary. However, the existing elements have been further specified and new content elements have been identified. The “Benefits and costs” is an already existing content element. During the interviews, the various cost items associated with Big Data Analytics project in healthcare were identified. There are also a number of different general remarks about the cost analysis made by the interviewees. The list of the identified cost items and general comments about the cost analysis can be found in Appendix 6. Furthermore, the potential financial and non-financial benefits of Big Data Analytics projects in healthcare have been identified. General comments were also made by the interviewees on how the benefit analysis can be performed and what facets the project manager must take into account. These results can be found in Appendix 7. A new sub-element of the content element "Benefits and costs" is that is mentioned by one interviewee is the value of the data. The value of the data can be marked as a type of benefit according to this interviewee, because the data can be sold in a processed form, for example, to other parties within the network of a healthcare organization.

The interviewees indicated that in the content element "Financial plan" it should be clear how the project will be paid and who will bear these costs. It must therefore be clear beforehand who the project sponsor is. Furthermore, more body was given to the content element "Feasibility study" during the interviews. An extensive description has been given on the areas in which the feasibility of Big Data Analytics projects should be tested. This detailed description can be found in Appendix 5.

Finally, three new content elements have been identified within this sub dimension. The first one is that the project manager should identify the impact of the solution on the patient. The second content element is that the project manager should identify the impact on the current affected healthcare processes. Lastly, the impact identified during the proof of concept is included as a new content element. Within this content element, the responses from the proof of concept can be included.

Investment risks

None of the current content elements are marked as unnecessary. However, it has been emphasized that the Risk Management Plan must contain the mitigating measures. This has not yet been explicitly mentioned in the already existing framework. The various risk factors that apply are included in the "Risk factors" dimension within this framework.

(16)

16

Investment assumptions, considerations and scenarios

The results within this sub-dimension contain further specifications of the existing content elements. The interviewees indicated that the project manager for the content element “Best practices research” can carry out reference visits at other care organizations. One interviewee advises to also look outside the healthcare sector for a cross-sector reference visit. In addition to the content element "Realistic technology scenarios", it has been indicated that healthcare organizations must select the right technology supplier, by including this selection in the proof of concept. Another addition is that the description of the role of IT within the project can be given here.

Investment governance

The interviewees have indicated that Data Governance is an important aspect that needs to be taken into account beforehand. Privacy by design, Security by design, Data Quality and Metadata management should be included in the new content element "Data Governance".

4.7 Data Governance

The first new dimension added to the business case framework for Big Data Analytics is the dimension "Data governance". It was decided to make this a separate dimension, because in all interviews, as well in the literature review, it has become clear that Data Governance is an important and conditional aspect for a Big Data Analytics project. If a healthcare organization does not have its Data Governance in order and does not pay attention to it, then, according to the interviewees, it is impossible to perform a Big Data Analytics project properly. Based on the results from the interviews, this dimension can be subdivided into the following two sub-dimensions: “Data management” and “Data protection”. The description and content elements of each of these sub-dimensions are described below.

Data management

This sub dimension contains a collection of different data management elements that have emerged from the interview results. This collection of data management elements determine the degree of usability of data, and are a tool to organize processes to improve, for example, the data quality and completeness. “Meta data management”, “Data Quality” and “Data stewardship” are the three content elements within this sub

dimension that can be derived from the interview results. The interviews showed that metadata management is crucial for Big Data Analytics project, because otherwise it is impossible to know the meaning of the data and in what context it takes place. This is required in order to interpret the data correctly and process the data in the predictive models or other data models. Data quality is marked as an important aspect, because otherwise the Data Scientist will make incorrect predictions or cannot make any predictions due to the noise in the data. Data stewardship is the last content element. The interviewees indicated that it is important that the correct data is recorded so that the correct predictions can be made. This means that processes must be set up to register as much relevant data as possible about the medical processes, so that these data can be used in a later stage in Big Data Analytics projects. These are partly the tasks of a Data Steward.

Data Protection

This sub-dimension is in the context of data protection and the key elements around data protection. This includes topics such as privacy and security. This sub dimension contains two content elements that can be derived from the interview results: "Privacy by design" and "Security by design". The interviews showed that there are two possibilities for healthcare organizations to deal with (big) data.

The first possibility is that the data is irreversibly anonymous. In that case, the law on personal data no longer counts. However, it has been indicated that you must be able to demonstrate that the data is

irreversibly anonymous. In a proof of concept, for example, healthcare organizations can use also fake data, so that the feasibility and benefits of the project are checked, without having to go through the law.

The second possibility is that personal data are processed that are traceable to a person. In that case, the law does apply and you must comply with all legal requirements. In this case, the healthcare organization must also determine the legitimate grounds for using all those data. In order to meet the requirements, in-depth legal knowledge is required. The interviewees stated that by involving the Data Protection Officer in advance, this person can help to determine whether the law applies, and if the law is applicable, how the solution can be designed to meet the legal requirements. Concepts such as “Privacy by design” and “Security by design” then apply and must be implemented.

(17)

17

4.8 Ethics

The second new dimension added to the business case framework for Big Data Analytics is the dimension "Ethics". This has become a separate dimension, because in the interviews it emerged that this must be taken into account in advance in a project and that it can make or break a project. For example, the consequences of ethically irresponsible choices could lead, according to the interviewees, to reputational damage. Examples of ethical issues that were discussed during the interviews are: (1) which data you may or may not use or (2) who remains responsible for the treatment if the choice of treatment is made on the basis of data.

There are insufficient interview results to further shape this dimension in more detail with sub-dimensions, because substantive ethical issues need to be discussed. None of the interviewees was an ethicist to give more details about ethical substantive processes. However, comments were made about how

healthcare organizations can deal with ethics in Big Data Analytics projects. These comments will be discussed further in this section. These comments can be grouped under one single content element: "Validating ethical issues".

The interviewees indicated that care organizations should look in advance to see whether ethical issues arise with the chosen Big Data Analytics solution even if it does not play at that time or phase of the project. This can be the case once healthcare organizations scale up the project. Furthermore, the interviewees have indicated that healthcare organizations should, at all times when they start working with medical

substantive data, involve an ethicist. This person validates the chosen solution from an ethical perspective. This allows the project manager to solve ethically objectionable issues beforehand to prevent a rejection by the "Medical Ethics Committee" of a healthcare organization.

4.9 Critical success factors

To increase the chances of success of a project, questions have been asked about the critical success factors of Big Data Analytics projects in healthcare. The critical success factors can be grouped into the following four categories: (1) People, (2) Technology, (3) Organizational and (4) Approach. Figure 3 contains an overview of the mentioned critical success factors subdivided into these four categories. The critical success factors are briefly described below.

Figure 3. Graphical overview dimensions new Big Data Analytics business case framework People

All interviewees indicated that people are the main success factor of a Big Data Analytics project. Examples of two statements made by two different interviewees are “50% of the chance of success of Big Data Analytics projects comes from the people” and "without the right people, you cannot make it". The interviewees were asked which types of people are needed to increase the chances of success of the project. The respondents indicated that multidisciplinary people are important. They have indicated that people obviously have their own specialism but they must be able to act in a multidisciplinary manner. Furthermore, the respondents indicated that multidisciplinary teams are a key to success. The teams must be scalable and shaped according to the demand of the project (phase). From the interview results it can be concluded that the following team composition is according to the respondents the ideal (dedicated) team composition: Data Engineer, Data Scientist, medical specialist or nurse, end-user of the information, project manager, product

champion/problem owner and Data Architect. The description of these roles can be found in Appendix 8. The interviewees indicated that these people should commit to the project and that time should be made available for the project. Other roles that are important but which have been appointed by a few respondents and cannot therefore be generalized, are: critical people who look critically at the added value of the solution, user

Big Data Analytics in healthcare: The path to a business case