Nigeria Collects Vast Amounts of Health Data. Can AI Help Put It to Work? By John Uwaishe
Dedicated Nigerian health workers, laboratory staff, hospital administrators, government officials, and national survey teams collect data on births, deaths, vaccinations, diseases, treatments, nutrition, and access to essential services.
The 2025 Nigeria Health Systems and Services Profile, supported by WHO, reports that Nigeria’s District Health Information Software 2 collects routine data from about 38,500 health facilities, with a 92.3% reporting rate in 2023.
This raises an important question. Is the data genuinely accessible and usable? Information may exist in a database yet remain readily unavailable to decision-makers. It may be available but incomplete, complete but outdated, or technically accurate yet presented in a form that requires expertise or time they do not have.
Why collected data remains difficult to use
Nigeria’s health information challenge appears to be a web of interconnected barriers.
The first is fragmentation. Health information is distributed across routine reporting platforms, surveys, electronic medical records, laboratory systems, disease registries and programme-specific databases. Some disease programmes operate parallel systems, often reflecting different reporting requirements and funding sources. When systems use incompatible formats or cannot exchange information, producing a coherent picture becomes difficult.
This fragmentation is particularly significant in a federal system. Information must flow between facilities, local government areas, states, federal institutions and development partners. Responsibilities may overlap, and standards and implementation capacity vary. The result can be duplication in some areas and significant gaps in others.
Data quality presents an additional challenge. The WHO-supported profile highlights issues such as incomplete reporting from public facilities, ongoing underreporting by private providers, and limited implementation of DHIS2 at the facility level. A high national reporting rate does not necessarily reflect all local omissions or inconsistencies. When the source data is incomplete, even the most advanced dashboard will inevitably offer an incomplete representation of the actual situation.
Health information systems require functioning computers, reliable connectivity, secure storage, software maintenance, and people who can operate them. The same assessment identifies chronic underfunding and inadequate basic information and communications technology infrastructure as constraints on Nigeria’s system.
There is also a human-capacity issue; however, it should not be simplified to the notion that Nigeria lacks experts. Nigeria is endowed with competent health professionals, statisticians, researchers, engineers, and public administrators.
The concern is that institutions do not consistently have sufficient dedicated personnel for health records and information management when necessary. Moreover, the current staff may lack adequate resources, training, or time to effectively analyze the information they are tasked with collecting.
A related challenge is a culture that prioritises submitting reports over using data to improve decisions. Frontline workers often spend considerable time entering information but receive little feedback on local trends, service gaps, or how their data has shaped policy and resource allocation.
These problems are well recognised. Nigeria’s Second National Strategic Health Development Plan called for stronger institutions, integrated systems, better data quality, greater interoperability and improved analytical capacity. Their persistence suggests that the central challenge is not identifying solutions but financing, coordinating and sustaining them.
Turning collection into action
There are promising examples demonstrating what can be achieved through a redesign of the information process.
In a 2025 account of its SMART+ programme, UNICEF reported that traditional paper-based nutrition surveys in Nigeria could take three to four months to produce results. Such delays made the findings less useful for urgent planning.
SMART+ introduced an end-to-end digital approach to survey collection, aggregation, analysis and quality assurance. UNICEF reported that the programme made actionable insights available within approximately two to three weeks.
The importance of this example is not the fact that paper was replaced with software. The process connecting collection, quality assurance, analysis and access was reconsidered as a whole. Technology created value by reducing a defined operational delay.
The Nigeria Health in Digital Initiative points in a similar direction. According to the World Bank, the initiative aims to address data fragmentation by leveraging electronic medical records and data-exchange systems to improve information sharing and support policy decisions. Digitisation should make existing information easier to combine, verify and use.
Investment must extend beyond purchasing software. Nigeria needs consistently applied data standards, improved interoperability between authorised systems, sustained infrastructure funding, and clear responsibility for data governance. More trained personnel are also needed to turn records into useful analyses at the federal, state, and local levels.
Private healthcare providers must be included. A substantial share of care occurs outside government facilities, and incomplete reporting renders both national and local data unreliable. Improving participation will require practical reporting processes, clear expectations, adequate support and confidence that sensitive information will be handled responsibly.
Most importantly, analysis should benefit the facilities and communities that produce the data. Primary healthcare managers should be able to track shifts in demand, reporting gaps, local disparities and emerging pressures without waiting for an annual report produced elsewhere.
Where artificial intelligence may help
Artificial intelligence can help, but only when the underlying data systems work properly.
For my 2025 MSc research at Heriot-Watt University, I built and evaluated a conversational AI system that enabled users to query World Health Organization indicators in plain language. It combined retrieval-augmented generation with a fine-tuned model, making relevant data easier to find and interpret without requiring users to manually navigate complex datasets.
Fifteen participants, drawn from health, research, and policy-related backgrounds, completed public health decision tasks using both the conversational system and manual methods. Participants rated the system 4.47/5 for ease of use and 4.47/5 for making complex information easier to understand. It received 4.27 for being faster than the manual method, 4.40 for overall satisfaction and 4.53 for willingness to use it in the future. Responses were also rated highly for relevance, coherence and efficiency.
The results demonstrate a practical opportunity. Conversational systems can reduce the expertise and time required to locate indicators, compare figures and understand established datasets. They can help bridge the final gap between a technically available resource and a usable answer.
AI is only as reliable as the information behind it. Missing patient records, conflicting definitions between institutions and inaccurate source data cannot be solved by adding a model on top. When the underlying system is fragmented or unreliable, AI risks presenting weak conclusions with unwarranted confidence.
Any use of AI in public health should draw from approved sources, link its outputs to the underlying evidence and clearly show where uncertainty exists. Access controls are particularly important for patient-level information, while responsibility for interpreting results and making consequential decisions must remain with qualified professionals.
The goal is not to place a chatbot on top of every government database. It is to build reliable information systems and then use carefully evaluated tools, including AI, to make those systems easier to navigate.
Advanced technology will only be useful if the foundations are in place. Nigeria should prioritise the quality, interoperability, governance, and funding of its health information systems, while investing in the people who collect, maintain, and interpret the data. Useful analysis must also reach decision-makers at every level. Once these foundations are stronger, AI can be introduced to address specific problems in accessing or interpreting information.
Nigeria already produces vast amounts of health data. The challenge now is to turn it into timely insight that leads to better decisions and better health outcomes.
About Author
