AI Key to Tackling Incomplete Patient Records in Nigerian Hospitals — Data Strategist

Data strategist, Mr. Azeez Akinbode, has urged Nigerian healthcare institutions to adopt Artificial Intelligence (AI) as a solution to the widespread problem of incomplete patient records.

Speaking on Monday, Akinbode noted that while healthcare systems around the world increasingly rely on predictive algorithms to guide medical decisions, the effectiveness of these tools is threatened when patient data is incomplete. This concern, he said, inspired his latest research.

According to him, Nigeria’s longstanding challenges in healthcare infrastructure and service delivery underscore the urgent need for data science as a vital link between policy design and real-world implementation.

“Data gives us visibility. It helps us identify who is being left behind and why. When governments invest in analytics capacity—from local hospitals to national agencies—we begin to create policies that are not only functional but fair,” he said.

Akinbode explained that his research, titled “Impact of Incomplete Records on Predictive Models That Use Electronic Health Record Data,” is now archived in the OhioLINK Electronic Theses and Dissertations Centre. The work was also recently showcased at the 2025 INFORMS Annual Meeting in Atlanta, Georgia, where it attracted notable interest due to its practical relevance in healthcare delivery.

He said the study focused on a challenge common to health systems globally: inconsistent or missing patient data, particularly in resource-constrained environments where critical decisions must still be made.

“My research shows that even with minimal data, we can build models that remain reliable and fair,” he explained.

The study assessed 126 predictive models based on the seven American Diabetes Association (ADA) screening variables, testing how each performed when essential data points were removed. The results were compelling: using just three variables — age, race, and family history — the model maintained a strong predictive accuracy, recording an AUC of 0.95 across diverse patient groups.

“So, we don’t always need big data to make big impact. If we can pinpoint the key variables that truly matter, hospitals with limited resources can still make smarter, faster decisions,” he added.