The requirements elicitation reported in this paper concerns development of a blockchain vaccine supply chain management web/mobile software application, the aim of which is to provide end-to-end visibility of the vaccine supply chain. This project has been necessitated by the needs of low- and medium-income countries (LMICs), where the vaccines supply chain is faced with challenges, which leads to problems including expiration and/or lack of needed vaccines, wastages and stock-outs ( Yadav, 2015). Faced with shortage in resources and shrinking development partners support, LMICs need to employ strategies that are more efficient and effective. This will enable countries to achieve universal health and save many lives, especially those of children.

This paper posits that one way of achieving this efficiency is by employing blockchain technology. Blockchain technology has attracted tremendous interest from a wide range of stakeholders, which include finance, healthcare, utilities, real estate and government agencies. This is due to its characteristics as a shared, distributed and fault-tolerant database/ledger that every participant in the network can share and where any single transaction is witnessed and verified by the network nodes. Information exchanged is resilient to manipulation. Blockchain’s decentralized architecture can be leveraged to develop an assured data provenance capability for vaccine supply chain management ( Liang et al., 2017). The project for which requirement elicitation is reported is based on permissioned blockchain, since the vaccine supply chain participants in healthcare are more likely to be altruistic and operate under real identities than are users of a highly anonymous, unregulated system like Bitcoin, which employs public blockchain. The application will be deployed in a private network owned by actors in the vaccine supply chain. Cryptographic data structure and a consensus algorithm will be employed. Self-executing and self-enforcing smart contracts are relevant in the vaccine supply chain permissioned blockchain because there will be need to verify correctness of transactions and enforcement of predefined rules.

There are several requirement elicitation techniques which can be used to define the scope of a project and gather user requirement. Abbasi et al., 2015 categorizes them into four namely: classical traditional techniques (interviews, questionnaires and surveys), cognitive analytical techniques (card sorting, laddering and repertory grids), modern and group elicitation techniques (brainstorming, joint application development (JAD), prototyping) and social analysis (ethnography, direct observation, passive observation) ( Abbasi et al., 2015).

The project adopted mixed methods ( Creswell, 2003), using different methods from the categories discussed by Abbasi et al. (2015). This was based on the fact that no single requirements elicitation method is capable of giving all the software requirements, meaning it is important to use several methods ( Yousuf & Asger, 2015).

The requirements elicitation methods were:

Table 1 shows the range of participants in all of the seven methods employed in the requirements elicitation activity.

Stock and flow diagramming from the system dynamics field will be used to represent the analysis of the methods of requirements elicitation discussed. The goals, processes, rules and challenges, as described by the stakeholders in the vaccine supply chain, are represented in the stock and flow diagramming. Besides, this method is specifically important because vaccine supply chain management involves complex, dynamic decision-making tasks. The complexity is due to the environmental context, which is characterized by a number of variables to manage temporal spacing of consecutive decisions and availability of needed information. The second source of complexity is the structure of the underlying system. This systemic complexity is the result of interactions between system components in such a way that the system produces outputs that are not easily predictable ( Özgün & Barlas, 2015). This paper posits that by appreciating time delays, feedback and remaining sensitive to nonlinearities, vaccine supply chain management can be efficient and accurate, saving countries investment and enabling vaccines to be at the right place at the right time in correct quantities, hence saving more lives.

The dynamic model of vaccine stock management is unique for several reasons: First, there should be no “backlog of orders” in this context as children need to get the vaccine in the right time. Second, in cases of temperature variations, the whole vaccine stock risks losing potency. Third, in cases of disease outbreaks, there is the possibility of whole-stock being consumed in a short time. Fourth, the final customers are varied and need to be reached even when they have not “ordered” through vaccine drives.

The work builds upon the model proposed by Gonçalves (2003), capturing a supply chain with a single supplier offering a unique, non-substitutable product to retailers ( Villa et al., 2015). Figure 1 displays the simplest structure of our supply chain structure. The structure shows the supplier, who in most LMICs is the UNICEF Supply Division. The retailer represents the specific country and the final customers are the last mile health facilities, which is where vaccines are administered to children.

Literature review. A.N., J.M. and H.A.M., from Jomo Kenyatta University of Agriculture and Technology, held weekly workshops over a span of 3 months. Prior to every workshop every team member was tasked with carrying out research on the topic scheduled to be next discussed. After the specified period of time members were tasked with writing white papers on the topics. The topics that were discussed and white papers written concern:

All white papers can be accessed on OSF ( Mindila, 2018).

LimeSurvey questionnaire design. After completion of phase 1, A.N.M, J.M.W and H.A.M designed questions in LimeSurvey software based on part of the literature review in phase two and the World Health Organization (WHO) Effective Vaccine Management (EVM) criteria ( WHO, 2018). The questionnaires were sent to the NVIP department staff for verification before administration. The questionnaires were administered to the 30 members who later also attended the Requirements elicitation workshop. The list of the questionnaire recipients can be accessed on OSF ( Mindila, 2018). The LimeSurvey questionnaire design can be accessed on OSF ( Mindila, 2018).

The themes of the survey were:

Focus group workshops with the NVIP department of the Ministry of Health (MoH). J.M.W and the NVIP head C.T. held a meeting on 8 ^th March 2018 at the NVIP offices where J.M.W introduced the project and obtained verbal permission from C.T., the NVIP head, to collaborate with a promise to issue a formal consent letter. NVIP head C.T accepted to be the project owner. Consent for the National Vaccine and Immunization Program (NVIP) focus group was obtained verbally from C.T., who also gave permission to the concerned staff to attend the focus group meetings. The participants were informed and invited by the head of NVIP. Consent was not sought from participants as these was considered low-risk interactions

The NVIP head C.T. invited A.N.M, J.M.W and H.A.M to present the project concept to the National Logistics Working Group meeting on 12 ^th March 2018 at the NVIP offices at DFH conference room and gathered their views.

Focus group workshop 1

A.N.M, J.M.W and H.A.M held a meeting on 25 ^th April 2018 with the NVIP department at NVIP offices Division of Family Health and introduced the Blockchain vaccine supply chain mobile/web application project. The minutes of what was discussed can be accessed on OSF ( Mindila, 2018).

Focus group workshop 2

A.N.M, J.M.W and H.A.M, C.S and J.C held a 1-day follow-up workshop on 26th April 2018 at DFH offices, where a team from NVIP was selected to work together with the project team. The project team shared the project timeline, a concept paper so as to enable the head of NVIP C.T. to obtain written consent. The project outline shared with NVIP can be accessed on OSF ( Mindila, 2018).

Focus group workshop 3

A.N.M, J.M.W and H.A.M. and the NVIP team held a one-day brainstorming session on 27th April 2018 that was guided by the LimeSurvey questionnaires designed in phase 2.

Focus group workshop 4

A.N, J.M and H.A.M and Joyce Charo of NVIP held a 1-day workshop on 17 ^th May 2018 where the survey structure was designed.

Discussion with KEMSA. The Project team had an introductory meeting with the CEO of KEMSA Mr. Fredrick Wanyonyi at the KEMSA headquarters on 19 ^th February 2018 where the team introduced the intention of the visit and the project. The CEO was receptive and linked us to the ICT manager, with whom we later organized for a meeting and thereafter held an interview with the ICT manager of KEMSA The communication that followed between J.M.W and Mr. Samuel Wataku can be accessed on OSF ( Mindila, 2018) as meeting with Prof. Wafula. The outcome of the meeting yielded collaboration themes which are captured as KEMSA CEO letter which can be accessed on OSF ( Mindila, 2018).

Focus group workshop with the National Logistics Working Group. The project team introduced the project to the National Logistics Working Group (NLWG), whose members were UNICEF, NVIP, John Snow, Inc. (JSI) and Clinton Health Access Initiative (CHAI) in a 1-day sensitization meeting made possible by the NVIP head. The presentation made by A.N.M. can be accessed on OSF ( Mindila, 2018).

Requirements elicitation workshop. The requirements elicitation workshop was held in Mombasa and involved 30 participants. The participants were drawn from Mombasa county, Migori county and Nakuru county. Besides the representation from the counties, there was representation from the NVIP department and the Regional Store managers from Mombasa, Kisumu and Nakuru and KEMSA. The project team were the facilitators. The aim of the workshop was to understand the processes, the SOPs, the actors, the stakeholders and the rules that govern vaccine supply chain. The workshop was guided by LimeSurvey questionnaires that were developed in phase 2.

The workshop adopted brainwriting or card sorting ( Abbasi et al., 2015) and brainstorming as the techniques for elicitation. It was guided by eight themes that had been prepared by the project team. A total of 12 participants from the three counties of Migori, Mombasa and Nakuru, whose positions were County EPIs, Sub-county EPIs, nurses in charge and Data clerks, took part in the workshop. There were three regional store managers drawn from Nakuru, Mombasa and Kisumu. There were six NVIP participants and eight project team members.

Four colors were selected and each color represented two themes. Every participant was assigned one of the selected colors randomly. Each participant was issued eight cards comprising of two green, two pink, two blue and two yellow in color to write responses to questions posed per color coded theme. The themes were coded as follows: Blue-Information system and supportive functions; Blue, Vaccine Management policies; Pink, Maintenance of Cold Chain Equipment; Pink, Temperature Monitoring; Yellow, Stock Management; Yellow, Storage; Green, Vaccine Distribution; Green, Transport.

In this workshop, a presenter prompted questions highlighting the important information needed on different themes. All participants then filled in their responses on cards provided. After posing four questions covering the first four themes, the participants broke out into 4 groups according to the color they were assigned randomly to discuss cards with color that match that of the group They would then discuss their responses and pin up a summary of answers generated. Anyone from the different groups added more answers on the pinned summary during rotation period. This process was repeated for the four remaining themes. After the group discussions, a plenary session was held where summaries and way forward were presented by the panelists.

The eight themes were as structured below:

Temperature monitoring

I am responsible for tracking vaccine temperature and at any one point in time I need to know the temperatures under which all vaccines are kept whether in stores or in transit. How may you make sure that you support me and what would be the challenges? How do we address them?

Quality, suitability and capacity of transport facilities

I am the officer in charge of vaccines transportation in the country. My duty is to guarantee quality and efficient transport services in the vaccines supply chain. What challenges may I face in ensuring vehicles deliver vaccines in good/favorable conditions within the appropriate timeframe? Also, what can you comment on the communication channels accompanying the vaccines transport services.

Information systems and supportive management functions

What tools and systems are used for vaccines data management? Are they sufficient in providing data for decision making?

Storage quality, suitability and capacity

There are various requirements for the various vaccines in terms of storage, how do I ensure that these requirements are not violated? If I walked into a particular vaccine store, how easy would it for me to locate a particular vaccine? What informs which vaccines I release upon requisition? Is the current storage sufficient, any contingency plans? Are there any security threats to vaccine storage?

Maintenance of cold chain equipment (CCE)

What important information and support would I require to provide effective maintenance of the cold chain equipment? Given that I would like to know the state of all CCE under my jurisdiction in terms of temperature and how much the equipment is holding and be able to respond when am called upon.

Vaccine distribution

I am the national/regional/sub-county vaccine distribution officer, what important information and support do I require to undertake effective vaccine delivery?

Vaccine management policies

What information do I require for effective vaccine management?

Stock management

I need to order the next stock of vaccines, where do I direct my requests to at the various levels right from the health facilities. Can I tell on the fly about various stock levels? What decisions inform the stock ordering policy? How do I know what amounts of stock to order (at the different levels)? How do I mitigate stock-outs and/or reorder delays? How do I know when orders arrive? How do I handle/prevent losses?

The visit was to the EPI logistician office, Mombasa subcounty store and Mombasa Regional Store on the 30 ^th of May 2018; it involved observations of how the vaccines were stored and labelled, how temperature monitoring was done, and how records were kept and reporting on vaccine movement was conducted. The staff were able to explain the vaccine supply chain processes and some of the challenges they face. Discussions were mostly oral, and pictorial images of what happens at these stores were captured. In a nutshell, the fact finding involved both observation and discussions.

The literature review method resulted in the output of white papers on different topics. The subheadings below contain important excerpts from the literature review. The full papers can be found on OSF ( Mindila, 2018).

International aspects of vaccines. The Global Vaccine Action Plan (GVAP) is a framework approved by the World Health Assembly in May 2012 to achieve the Decade of Vaccines vision by delivering universal access to immunization ( World Health Organization, 2013). African states have embraced immunization as a cornerstone for Sustainable Development Goals (SDGs) and AU agenda 2063 (Addis Declaration on Immunization, ADI 2017); SDGs call for “…access to safe, effective, quality and affordable medicines and vaccines for all by 2030”. ADI Strategy 2 states the need to “Address gaps in immunization and work with key partners to overcome barriers to access and utilization of immunization services”. Monitoring national vaccine coverage is key in understanding the execution of immunization programs so as to identify areas within immunization systems needing improvements, and prepare for the introduction of new vaccines ( World Health Organization, 2016).With the introduction of new vaccines, developing countries are facing serious challenges in their vaccine supply and logistics systems, especially in resource-constrained environments.

Use of provenance in the blockchain. Provenance provides the history of the origins of all modifications to an entity, agents that have modified the entity through processing it or viewing it, and the particular processes that have altered the entity. ( Liang et al., 2017) Using provenance for instance, one can be able to track pharmaceuticals from the manufacturing laboratory as they move all the way to the end consumer ( Hasan et al., 2009). In the implementation of the blockchain-based solution, the PROV Data Model, as specified under the PROV Standard released by W3C Provenance Working Group in April 2013, will be incorporated ( Moreau & Missier, 2013). The data model is composed of an agent, an entity and an activity at the core. It can store information about data, how it was modified, who modified it and when it was modified as well as the locations it has passed through. Therefore, incorporating it into the proposed solution will ensure that every vaccine in the supply chain can be isolated, analyzed, and all the activities associated with it identified hence ensuring full transparency and accountability in the supply chain.

Use of RFID and blockchain in cold chain. To improve traceability and monitoring of vaccines, use of RFID and blockchain technologies is recommended. RFID tags attached to vaccines will be read at pre-determined intervals by RFID readers fitted in transit vehicles and in storage facilities. The data read will be channeled to the blockchain. Small modifications can be done to the RFID tags such that they include some sensors, e.g. temperature sensors. An RFID tag with a sensor can now be aware of the temperature of its environment and can return such data once interrogated by a RFID reader. ( Ko et al., 2016). Beaconing RFID tags are consistently sending information out at predetermined intervals. These tags can be pre-programmed to send data following a pre-programmed schedule. Depending on application at hand, different tagging levels may be selected they include: Pallet level tagging, Case level tagging and Item level tagging ( Gaukler, 2011). Item level tagging combined with smart shelves can help prevent stock outs.

Blockchain. In a blockchain transaction, two cryptographic primitives are used to prevent malicious users breaking the system. A digital signature is used to make sure that the information is signed by the claimed person as well as to test whether the information is modified by some malicious people. The signature process contains signature generation and signature verification. Given a message, the signatory generates a signature by using his private key, and the verifier can use signatory’s public key to verify the messages authenticity. Instead of signing on the message directly, a cryptographic hash function is applied to the original message to produce a message digest for performance reason.

Merkle tree is an important element in blockchain security. It is a binary tree of hashes as proposed by Ralph Merkle ( Merkle, 1980; Merkle, 1988) that is used to verify data integrity efficiently and securely. One of the strengths of Merkle tree is that there is no need to recompute the hash of all data if one data block changes.

A network model based on multi-layer distributed accounting can be regarded as an organic combination of the centric network and decentralized network, which not only effectively utilizes the performance and capabilities of the cloud server, but also significantly improves the overall security and reliability of the IoT with taking advantages of blockchain techniques. Such a network is made up of entities and processes. The entities in an IoT environment are objects. This include equipment, facilities, users, APIs and other abstract things, so long as they have networking capabilities. These nodes have capabilities like data transmission, analysis and storage, edge layers that are made up of a certain number of objects with a central node that manages them and is able to interface to a superior layer. The node is self-independent in that it only communicates to the central node in order to achieve optimization. This can however be modified depending on the scenario and performance needs. A high level layer is a network entity consisting of a number of nodes that employ the blockchain protocol ( Li & Zhang, 2017)

Vaccine supply chain. Well-functioning supply chains, delivering medicines, vaccines and other health products, are critical to the provision of health services. Most governments in LMICs, especially in sub-Saharan Africa, choose a distribution model where the government procures drugs and distributes them to health clinics using a publicly run central medical store (CMS) and a government-owned transport fieet. There is considerable heterogeneity in the governance structure of the CMS in different countries. In some cases, products funded by some bilateral agencies are procured directly by the international agency or a procurement agent acting on its behalf. Once medicines arrive in the country, the most common structure for distribution involves warehousing and storage at the CMS. Depending upon the geography, administrative structure and the number of health facilities in a country, products are first distributed from the CMS to regional and district medical stores that supply hospitals, health centers and community health workers ( Yadav, 2015).

In LMICs, information about demand, stock levels and timely use of vaccines is poorly kept, affecting timely supply leading to expirees and/or lack of needed vaccines ( WHO et al., 2010; WHO, 2014).

There is also the risk of poor product quality and counterfeiting that LMICs face. Remedy would be to help manage this information along the supply chain by enabling visibility and transparency of information from the start to end of the supply chain for all authorized persons. This requires accurate data collection and secure data storage to enable a flow of trusted information between parties and to enable forecasting and any other analytics required ( WHO et al., 2010).

A.N.M, J.M.W and H.A.M held a meeting on 25 ^th April 2018 with the NVIP department at NVIP offices Division of Family Health and introduced the Blockchain vaccine supply chain mobile/web application project. The minutes from this meeting and the project outline shared with NVIP can be accessed on OSF ( Mindila, 2018).

Figure 2 represents the ecosystem of policies and regulations that guide vaccine management in Kenya. Policies developed by WHO which are applied globally, and GVAP and Decade of Vaccines 2010–2020 objectives and goals are aligned to reduce morbidity and mortality and address availability of potent vaccines for children. Kenya’s policies and plans are aligned to the global objectives and plans.

Figure 3 shows the organizations identified by NVIP as those that offer different forms of support, ranging from technical, financial and infrastructural support.

The brainstorming workshop 3 with the NVIP team provided insight into the processes of estimation of needs, procurement and arrival of vaccines. This was captured in the form of UML diagrams enabled by draw.io software and are represented in Figure 4– Figure 8.

Figure 4 shows the participants in the function of estimation of needs. They are health workers, health managers, depot managers and national logistician.

Estimation of vaccine needs happens at four levels, as illustrated in Figure 5: at the health facility, at the sub-county stores, at the regional store and finally at the primary store. At health facilities the people involved are the health workers; at the sub-county and regional stores, health workers at managerial level and store managers are involved; and the national logistician is found at the national level.

Figure 6 captures the process of estimation of vaccine needs as explained by the NVIP team. It is a sequential process that moves from right to left and time of events moves from top to bottom. Parameters considered include the target population, the minimum and maximum stock, stock available and amount of vaccine received last for each of the antigens. These parameters are then used to fill a vaccine order sheet. This process happens at all the four levels, every lower level propagates their order sheet to the higher level until the national level receives.

Figure 7 represents the participants in the procurement and arrival process. The process has eight participants, namely: UNICEF country office, NVIP head office, UNICEF Supply Division, Ministry of Health, National Regulatory Authority, Clearing Agent, Clearing Agent warehouse and the Cargo Handling centre. Figure 8 represents the procurement process while Figure 9 represents the processes involved in the arrival process as explained by the NVIP team.

The process is a sequential flow from left to right and in terms of time the events happen from top to bottom. Four parameters stand out as crucial to the process, this are, arrival dates, five pre-shipment documents, NRA permit and Tax Exemption. This four parameters facilitate the arrival and clearance process.

Lastly the NVIP team provided our team with insights into the final part of the arrival process that involves vaccine inspection and filling of the Vaccine Arrival Report (VAR). Figure 10 and Figure 11 show the participants and the process respectively.

The discussion with KEMSA gave us insight into the support they offer to NVIP. KEMSA offer infrastructural support to NVIP by way of providing cold chain storage. We further discussed with KEMSA on possibilities of hosting the blockchain application on their servers and we discussed possibilities of integrating the blockchain application with their systems.

The National Logistics working Group workshop provided a forum for us to create awareness of the blockchain solution for vaccine supply chain and the potential it has of enabling visibility, traceability and transparency.

The requirements elicitation workshop and the site visit enabled the research team to understand the process involved in the vaccine supply chain. The integration of the literature review already carried out, the workshop, the focus group meetings and the site visit enabled the team to capture the requirements as discussed in the figures below. The raw data of the requirements elicitation workshop and the data coding done by the research team as editable word tables and excel sheets are provided on OSF ( Mindila, 2018). They represent all the eight themes discussed in the Methods.

The requirements of the eight themes discussed and whose survey results are also provided on OSF ( Mindila, 2018), which also covers the literature review themes discussed are captured in Figure 12– Figure 20.

Temperature monitoring and storage. For every vaccine to be traced and remain visible within the entire supply chain, whether in storage or in transit, the Blockchain application needs to have a means of getting the temperature readings and associate the readings to a specific vaccine that has a unique digital identity within the system. At the time the vaccines arrive at the primary store, the personnel in charge will be responsible for getting the details of the vaccines and associated temperature readings. Figure 12 and Figure 13 represents a use case and activity diagram that will achieve this requirement. This will partly involve the tests done to ensure that the vaccine is in good condition at the time of arrival.

The application will employ RFID technology and QR codes to digitally capture details of each vaccine. The granularity of the capture during the testing can be the vial and the carton boxes or as agreed by the policy makers. The application will employ iBeacon and Eddystone technologies to provide details of temperature conditions and the adherence to FIFO and FEFO policies.

While on transit, the trucks involved will have iBeacon temperature sensors that will be connected to the drivers’ mobile phones and will then be able to transmit the temperature readings of the vaccines in transit. The assumption here is that as part of the contract for transportation the driver needs to have a smart phone with the application responsible for temperature readings and location installed and enabled respectively.

Figure 14 illustrates the process involved in capturing the vaccine details and location, and reading vaccine temperature.

Maintenance of CCE and transport vehicles. Maintenance of the CCE and transport vehicles is important in keeping the vaccines viable. Figure 15 and Figure 16 represent the activity diagram and the sequence diagram that captures the process involved. Figure 17 captures the objects involved in maintenance and clearly shows the attributes that would be important in maintenance.

Vaccine stock management. Figure 18 represents the processes involved in vaccine stock management.

Distribution and Transportation of vaccines. Figure 19 represents the processes involved in distribution of the vaccines, and Figure 20 shows the different components of the vaccine distribution process.

Open Science Framework: Blockchain Web/Mobile Application for Vaccine Supply Chain. https://doi.org/10.17605/OSF.IO/PTUAG ( Mindila, 2018).

This project contains the following underlying data:

Open Science Framework: Blockchain Web/Mobile Application for Vaccine Supply Chain. https://doi.org/10.17605/OSF.IO/PTUAG ( Mindila, 2018).

This project contains the following extended data:

Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).