Many of the participants observed that their major challenges are “inadequacy of modern neuro-behavioural evaluation equipment,” “lack of facilities and expertise” and “facilities and expertise for large dataset acquisition are not available.” These direct quotes corroborate the insights we got from our data collection exercise. One participant captured it this way; “I believe the manual approach to gathering, collecting, storing the available data as a field as against digitised methods coupled with not having central (national, regional and continental) repositories like other developed nations has not helped matters.” These views validate what we found as we were collecting data. In our neuroimaging data collection exercise, one striking challenge was the scarcity of MRI scanning machines. At the time of completing this research, it should be noted that there is only one dedicated research-dedicated Magnetic Resonance (MRI) scanner in Africa - situated at the University of Cape Town South Africa. All available machines are for clinical purposes and even these are difficult to access for research. This lack of research scanners is a major problem because owing to both ethical and legal issues clinical datasets are more difficult to be made FAIR.

In Africa, even the clinical scanners are difficult to come by. For instance, within the six geopolitical zones in Nigeria where the neuroimaging data collection took place, the number of functioning MRI scanners per zone ranges from 1 to 4 and these scanners are only situated at the metropolitan cities in each zone with the South-western geopolitical zone having the highest number of MRI scanners while some other zones have about 1–3 MRI scanners. In the South–South zone of Nigeria which consists of a minimum of 6 states, only about three MRI scanners were found in good working condition with some states not having a single MRI scanner. Hence patients often embark on interstate travels to have access to an MRI facility. This delays access, creates inconvenience and possibly discourages patients from seeking such services.

Diagnostic centres are therefore often the primary source of MRI data in Africa. The scanners are of older technology and often do not meet the modern standards for research MRI. For instance, of the three collection sites we used in Nigeria during the experiential approach, two have 1.5 Tesla (T) MRI units and one has 0.3 Tesla (T) units. In comparison, the standard MRI scanner for research in the high-income countries is 3 T or 7 T which can provide better quality image, higher spatial resolution and more complex MR sequences, richer information on the microstructure and function of the brain. This has overall effects on the type and quality of data that can be collected. Figure 2 demonstrates the clear differences in technical quality.

A striking disparity exists in the quality of MRI data obtained from high income countries (such as the HCP data) and that obtained from low-income countries (such as a clinical set up MRI scan in Nigeria; see Figure 2). The HCP data has a higher image quality and spatial resolution (3D) which makes it a better fit for microstructural and functional analysis while the clinical set up scan data has a lower image quality and lower spatial resolution (2D) which is good for clinical diagnostic use but not the best fit for obtaining microstructural and functional brain information. Indeed, as we attempted to use major data tools developed in high-income countries (e.g., Europe, UK and the USA) to pseudonymize or deface the subjects’ data for Wogu et al. (2023), we discovered that none of the tools was designed to handle data such as the one that we were able to collect in Nigeria. As a result, we ended up resorting to manual defacing of the MRI data. In addition, advanced tools designed to segment and classify different tissues and estimate morphometric statistics of the brain (Fischl, 2012) failed to accurately segment and estimate morphometric tissue properties. Finally, attempts to transform the clinical-quality data to research quality data using the most advanced AI algorithms (Iglesias et al., 2023) failed, highlighting potential current biases in training data for the most advanced neuroimaging analysis tools. Whether tools developed in high-income countries should be designed to work on low quality data such as that coming from low-and medium-income countries is a question that the community might need to address. In the future, the scientific community might need to evaluate the potential need for major analysis software to be able to handle data from low-, medium-and high-income countries, especially if these countries start generating data shared globally.

Furthermore, there is a dearth of skills necessary for the use of digital technologies for data generation, processing, application and sharing. As one participant echoed; “as lecturers we do not have the technical skills and expertise in the use of technologies for brain data to impact on our students. I have tried to collect data from diagnostic centres in Ethiopia, but the radiologists did not know what I was talking about.” This confirms what we saw on the ground. MRI technicians are not yet trained to collect research data but have experience in clinical scans only. They are often trained to generate images that are good for clinical diagnosis but may not be suitable for advanced research analysis. The lack of focus on research also means that data on patients (including images) are not properly stored or archived. A record keeping culture is often missing; images are deleted due to lack of storage systems. It was also observed that a good number of patients’s brain MRI datasets were saved in formats (XML/AVOL and JPG) not compatible with neuroimaging data analytic tools. Only a limited number of brain MRI data were stored in the appropriate format (DICOM) which limited the quantity of datasets collected.

The training of the young generation is indeed impaired. Lack of knowledge of new experimental approaches, methods of brain mapping and computational and theoretical models is extremely limited, effectively putting Africa several years behind for what pertains to medical imaging and neuroimaging data analysis. Furthermore, the lack of expertise in data, especially statistics and data science within neuroscience and related disciplines, also limits the application of the latest, cutting-edge data analysis approaches to brain data in Africa.

In addition to challenges around the measurement devices, challenges exist even after data collection. For example, the lack of network bandwidth to share data from the imaging centre to the research institutes and universities require that data be collected on CDROM and be transported and saved on personal computers. Hard drives to store data are limited and the computational power to manage, process and analyse large datasets is non-existent. Overall, this situation creates siloed dataset of potential utility that remain closed inside institutional repositories, creating an unFAIR situation for the new generation of African students and researchers.

Finally, the behavioural data sets needed in association with imaging data are still unstructured, manually collected, manually stored often with pen and paper, without digitization. One participant linked the technical challenges to economic challenges; thus, “you need to spend a lot of money. These devices are not widely available in Sudan, this not only means problems for ageing patients’ due shortage of necessary devices such as MRI, but also for research.” This demonstrates that some of the technical challenges identified above are linked to micro and macro financial constraints.

Many of the participants pointed out that economic challenges are critical. According to one participant, “Africans are too broke to spend money on brain data generation and there is a lack of established funding bodies unlike in Europe and America.” Data collection requires financial resources that are not available in Africa. As another participant opined; “I think Africa is full of brilliant minds and the economic barrier is the top reason that is holding us back.” In other parts of the world evidenced in the formation of the international brain initiatives (IBI) comprising large-scale projects (Adams et al., 2020), there are public and private research grants available for data generation and for infrastructure that can facilitate curation and sharing. These are lacking in Africa. Researchers most often spend their own money and time to acquire brain data. This was the case for us. Our data collection, processing and sharing process were not funded. We funded this with our personal money, and it is an approach that is not sustainable considering the economic situation in Africa. As a participant observed, FAIR brain data; “would require lots of investment to be sustainable considering the digital infrastructure challenges we face …” Further connections between economic challenges and technical challenges were also made. One participant observed that; “the lack of financial means is a real obstacle which prevents us from having equipped laboratories as well as imaging and neurophysiology devices adapted to our needs” Laboratories in higher institutions in Africa are not well equipped to generate, process, store and share digital data due to lack of funds.

Additionally in clinical settings, inadequate financial resources and an under-resourced health care system contribute to delays in accessing medical facilities and in some cases, absolute inaccessibility to these healthcare facilities and services. As one participant said, ‘even the diagnostic centres do not have sufficient data because not many patients can pay for the high cost of scanning’. In most African countries, the available health insurance (subscribed to by a handful of people) does not cover the cost of neuroimaging services. Patients often pay out of their pockets and the costs are high. That contributes to a dearth of scans available for research from hospitals and diagnostic centres.

As a participant observed, the socio-cultural contexts and interests in Africa prevents the prioritisation of FAIR brain data. This was captured in this response; “The behaviour and cultural thinking affects the actions, abilities and interpretation of how people see the world. Thus, the African socio-culturally is engaged with other problems and this compounds their basic problems towards action and behaviour. Finally, making them understand FAIR as a problem is difficult to do.”

It is also identified that a number of neurological disorders are not considered clinical problems that require clinical solutions or research owing to socio-cultural beliefs, perceptions and values. One participant observed; “oftentimes, Parkinson Disease (PD), Alzheimer’s and other mental disorders associated with cognitive decline which often occur in elderly people are considered as part of the ageing process which do not require treatment.” The extreme version of this belief is that the person is considered as mentally deranged. “We still regard people as mad people that belong to the marketplace” as a participant observed. The implication of these is that families generally feel that there is nothing they can do in terms of treatment and clinical care. It is important to note that in African culture, elderly people, especially those with declining mental capacity are communally cared for by family members. It is believed that love, care and companionship are what the sick aged persons need at this phase of their lives. These perceptions mean that less people seek clinical answers to possible cases of neurodegenerative diseases. The effect of this is that only a small number of MRI scans are conducted in hospitals and private diagnostic centres.

These beliefs can be religious beliefs which affect African understanding of brain diseases and corroborates what Okpalauwaekwe et al. (2017) observed in literature that, it is common to attribute mental illness to supernatural factors such as witchcraft, ‘juju’ (sorcery), evil spirits, divine punishment. These beliefs drive patients with mental health diseases such as major depressive disorders, schizophrenia, psychosis and epilepsy and their families to religious houses, rather than seek medical solutions. Prayer houses and healing or herbalist centres are major places many people seek remedies. It is often difficult to convince them otherwise because they have been indoctrinated in their beliefs and perceptions. This belief in supernatural causes of mental health disorders exists among people of different educational status. In a case study presented by Asare and Danquah (2017) caregivers of rich and well-educated families attributed dissociative amnesia to spiritual causes. Moreso, deep cultural and religious beliefs in the reincarnation also affect tissue donations. This makes the establishment of open access biobank difficult. Reincarnation is the belief that at death, the soul is reborn in new physical forms and any form of mutilation of the body can affect the rebirth of the soul. This belief influences people’s rejection of postmortem examination as well as organ donation (Ulasi et al., 2020). Many cultures in Nigeria and African in general have cultural or religious beliefs in reincarnation, which prevents human tissue donations. That means that there are no available tissue banks that can inform diagnosis and research into many neurodegenerative diseases in Africa.

Huge amounts of brain research in Europe and the United States rely strongly on deceased brain and spinal cord tissues donated by the public for diagnosis and research. There are a good number of brain or bio banks that form good resources for brain research that demonstrate this point. For instance, the Brains for Dementia Research initiative has been founded and funded jointly by Alzheimer’s Society and Alzheimer’s Research UK to support brain donation and provide much-needed brain tissue for researchers (Francis et al., 2019). And since its establishment, the initiative has supplied over 80,000 samples of brain tissue for research and these have led to impressive advances including advances to improve the diagnosis of vascular dementia and the identification of new risk factors for dementia (Aoife, 2018). This type of resource that supports research as well as clinical diagnosis is still lacking in Africa.

Another defining socio-cultural factor that affects conduct of brain scans and subsequently the availability of African brain data globally is mistrust. As a participant observed; “In many African communities, there is a deep-seated mistrust of external entities, especially when it comes to sharing sensitive data such as brain data. This mistrust can stem from historical exploitation and a fear that data might be misused, misinterpreted, or used for purposes that do not benefit the community from which it originated.” This level of mistrust is at the root of the emerging ecosystem of data localisation policies in Africa (Babalola, 2024).

There are also several ethical challenges that influence brain data generation in Africa. According to one of the participants, the ethics approval process is a tedious process; “in my country, obtaining ethical approval for clinical trials and utilising human data is quite cumbersome with concerns relating to the legal implications.” Another participant echoed the same sentiment; “Federal hospitals make it hard because before you can get approval it would take a while. Other research documents would get approval quickly but when it comes to neuroimaging it is always a hassle. You will be asked different unnecessary questions.” These insights align with data practices on the continent following a number of unethical clinical trials in Africa (Lurie and Wolfe, 1997; Stephens, 2006). There is a strong awareness of the importance of ethical conduct of biomedical research. However, while most institutional research ethics boards do exist, implementation remains questionable. In the clinical setting where we collected data, ethics approval processes were complex, expensive and sometimes unclear. Many of the diagnostic centres requested money to grant ethics approval, even though we had ethics approval from the University Research Ethics Board. This delayed the process unnecessarily and prevented collection from certain identified collection sites.

Additionally, legal issues regarding the processing of brain data in Africa remain foggy without clear regulatory frameworks. One participant noted; “I know there is a regulation on data but what it means for brain data is what I do not know.” This means that in many cases, whereas informed consent is obtained as part of the research protocol, consent for further processing and sharing of the data is lacking. This is particularly important in the face of an emerging data protection regulatory ecosystem in Africa.

In many countries of Africa, data protection regulations and laws are beginning to implement data localization provisions (Babalola, 2024). These countries have enacted laws requiring the storage of data locally and cross-border transfers of personal data are banned unless authorised by a data protection authority, designated entities or by the consent of the data subject (CIPESA, 2022). Justifications for this include the need to protect national security (Kugler, 2022), promote the local digital economy, and to ensure adequate data security and users’ privacy (Kimumwe, 2021). Despite the fact that many of these regulations often do not consider the dynamics of data processing in health research (Eke et al., 2021), this affects all brain data considered under the regulations as personal data.

According to one of the participants, “African researchers cannot continue to work in silos. There is a need for reproducibility.” Neuroscience research is not immune to the reproducibility crisis (Kelly and Hoptman, 2022; Miyakawa, 2020; Poldrack et al., 2017). Reproducibility is a concept that describes the ability of producing the same results from the same data (Nichols et al., 2017). This is different from replicability that refers to the ability of producing consistent results from new data (Klapwijk et al., 2021). Lack of availability of datasets behind published research results is one of the major causes of scientific irreproducibility. That is why FAIR data principles have become a crucial part of general project planning as well as a core funding requirement. Making FAIRness of African brain data will, thus, contribute to the overall scientific reproducibility of neuroscience research conducted in Africa and on Africans.

Scientific Research and innovation are fundamentally shaped by data. As one of the participants acknowledged, FAIR African brain data will help in aligning research and innovation in neuroscience to the needs of Africans; “without our data, global research outputs may not work for us. The COVID-19 taught us a lot of lessons.” The degree to which research results or innovations are generalisable to all contexts and populations depends critically on the degree of diversity or heterogeneity of data. Applications of brain data in research and innovation not only lead to understandings of brain structure and function but they also lead to the development of clinical solutions, therapies and medical technologies (Kellmeyer, 2021). Non-representativeness of data from a group or population can lead to biassed or discriminatory research or innovation outcomes. This issue can be seen in racial biases in modern technologies, including eye tracking (Blignaut and Wium, 2014), face detection algorithms (Buolamwini and Gebru, 2018; Furl et al., 2002), and medical artificial intelligence (AI) algorithms (Obermeyer et al., 2019). Neurotechnologies or therapies developed may not be able to work for populations whose datasets are missing research and innovation processes. FAIR African brain data will ensure that African populations are represented in the global brain data ecosystem; making sure that global brain research and innovation outcomes are generalisable to populations in Africa.

Research is often an expensive endeavour. One of the participants pointed this out; “data collection is costly. If we can share data, we can save a lot of money.” A considerable portion of the cost can be attributed to the data creation and collection phase which for brain research often includes the use of expensive equipment and tools such as MRI Scanners and EEG machines. It also includes the time costs for reading, analysing and refining and classifying data to make it understandable and useful for research purposes. Efficient neuro-data storage also results in additional financial costs as in many cases it includes the cost of indexing and querying, and the technical difficulty of big data visualisation (Hider et al., 2022). The application of FAIR principles to research data can enable such costs to be minimised. This is because of the reduced need for the collection of new research data as FAIR encourages data reuse. Also, as FAIR data is well documented, the time lost due to unfindable, unstructured and incomplete data is limited. Similarly, time spent on data analysis by researchers is reduced with the use of relevant software to read, understand and automatically recognise patterns. In fact, a cost–benefit analysis conducted for the European Commission (PwC EU Services, 2018) has shown that the annual cost of not having FAIR research data could cost the European economy at least €10bn each year. This figure gives some indication of the cost savings that could result from making African brain research FAIR.

“Our understanding of brain disorders in our communities will surely change if we can share and apply brain data in Africa. Maybe we can help to diagnose better.” This was observed by one of the participants. Brain research is a data-driven endeavour but despite the advances made over the years, the diagnosis and prognosis in brain related diseases remain unclear and uncertain in many situations. These are areas where the application of FAIR principles can enable improvement; it results in a better quality of data management and amplifies the quality of metadata and datasets along with their machine readability. Like other biomedical fields (Wise et al., 2019) the use of FAIR data makes the data more suitable for use with advanced analytical tools developed with the aid of AI and machine learning. Using such advanced analytical tools can potentially improve the diagnosis and enable a personalised precision approach. Application areas include medical image quantification and analysis, signal processing, automated analysis of genetic data and disease prediction (EPRS, 2022). AI can also be used to better understand diagnosis for example in situations where there are overlapping clinical presentations but different treatment options (Dwyer et al., 2018). However, the potential for AI to support brain research can only be fully realised through the application of FAIR data principles to brain data as this makes them available in suitable formats for use in modern computing environments.

As one of the participants mentioned, FAIR data can also increase citations. This participant said, “I also heard that it could help increase citations which is a good thing’. This aligns with evidence from literature because data citation helps to give proper credit to the creators of the datasets, encourages data sharing, and increases the visibility of the research (Cousijn et al., 2018). FAIR data principles can increase a researcher’s citations, where datasets are treated as citable entities. Researchers can assign persistent identifiers, such as Digital Object Identifiers (DOIs), to their datasets, making them citable in the same way as scholarly articles (Wittenburg, 2019). Also, other principles supported by FAIR such as the use of standardised metadata and open access practices enable researchers to make their data more visible to the scientific community.

By adopting FAIR data principles, researchers make their data more accessible and reusable for the broader scientific community. This increased availability and usability of data can lead to more researchers citing the original work, acknowledging the data contributors, and facilitating a culture of data sharing and collaboration. Nevertheless, while FAIR data can create favourable conditions for increased citations, other factors also come into play, such as the significance of the research, the quality of the data, and the impact of the findings.

FAIR data helps researchers to harness the potential of various technologies, including AI, ML, data analytics, the Internet of Things (IoT), and cloud computing. By using FAIR data researchers can access, integrate, and analyse data effectively, leading to enhanced research outcomes and enabling them to take advantage of the benefits offered by these technologies. However, a participant observed that; “we have a serious dearth of datasets in Africa. This has a negative effect on the development of AI and ML research in Africa.” For instance, scientists can enhance their ability to harness the power of AI technologies and utilise them to derive significant insights from their data. One reason for this is the interoperability enabled by FAIR data which makes it easier to combine and integrate different datasets from various sources. When data is interoperable, researchers can leverage AI techniques to efficiently extract insights, identify patterns, and derive new knowledge by analysing diverse datasets together.

Similarly, for IoT technologies where vast amounts of data are generated from interconnected devices and sensors, FAIR data principles also promote data interoperability, enabling researchers to integrate and analyse IoT data alongside other data sources. By making IoT data findable, accessible, and reusable, researchers can gain insights, monitor systems, and develop innovative applications for brain research.

Likewise, FAIR data aligns well with cloud computing, facilitating the storage, sharing, and processing of data within the cloud environment. Cloud platforms provide scalable infrastructure and robust computing power, enabling researchers to efficiently manage substantial datasets. By integrating FAIR data into cloud workflows, researchers gain the capability to leverage the scalability, accessibility, and computational resources of the cloud for advanced data analysis and processing tasks.