From courtrooms to code: How one Nigerian researcher is building the tools to make AI safe for patients

Artificial intelligence is no longer a prospect in medicine. It is already here. Across hospitals in the United States, Europe, and increasingly in Africa, AI-powered tools are being used to read medical scans, flag potential diagnoses, predict patient deterioration, and even generate clinical notes from doctor-patient conversations. The global market for AI in healthcare is projected to exceed $180 billion by 2030, driven by the promise of faster, cheaper, and more accurate medical decisions. But as these tools multiply, a fundamental question remains largely unanswered: who is making sure they work safely, fairly, and within the law?

The difficulty is not simply a matter of writing better regulations. Even where regulations exist, hospitals lack the practical tools to verify whether an AI system complies with them. A diagnostic algorithm trained on data from one population may produce biased results when applied to another, yet most hospitals have no automated way to detect this. A clinical large language model may generate plausible-sounding but entirely fabricated medical information, yet there is no widely adopted method to measure how often this happens or how severe the errors are. And as genomic medicine advances, healthcare institutions are storing unprecedented volumes of sensitive genetic data with cybersecurity protections that were designed for a previous era. The gap is not between innovation and regulation alone. It is between regulation and the operational tools needed to enforce it at the point of care.

Addressing this gap requires a different kind of researcher. The scholars who can make the greatest impact in this space are not those working solely on improving AI performance or those writing policy analyses from a distance. They are the ones who can do both: understand how the technology works at a technical level and design governance systems that can be implemented in real clinical environments. Increasingly, this means researchers who are not only publishing papers but also building and patenting the tools that hospitals will need to deploy AI responsibly.

Valentina Palama, a Nigerian-born researcher and legal professional based in Houston, Texas, is among this emerging class of inventor-researchers. Palama holds a Bachelor of Laws from the University of Buckingham in the United Kingdom, was called to the Nigerian Bar in 2021 after completing the Nigerian Law School, and earned a Master of Science in Computer Information Systems with a perfect 4.0 GPA from Prairie View A&M University in the United States. Her unusual combination of legal training and technical education in computer science has positioned her to work at the precise intersection where AI governance is most urgently needed: the space between what the technology can do and what the law requires it to do safely.

In 2025, Palama filed a patent application with the Nigerian Patent Registry for a system she designed to audit and govern artificial intelligence in medical diagnostics. The invention, titled “System and Method for Auditing and Governing Artificial Intelligence in Medical Diagnostics through Bias Detection, Explainability, and Regulatory Compliance Frameworks,” proposes an integrated platform that performs three functions simultaneously. First, it detects bias across multiple dimensions in AI-generated diagnostic outputs, identifying whether an algorithm produces systematically different results based on patient demographics. Second, it provides hierarchical explainability, giving clinicians layered explanations of how an AI system arrived at a particular recommendation, from high-level summaries to detailed technical breakdowns. Third, it automates regulatory compliance verification, checking whether the AI system’s outputs and processes conform to the requirements of the FDA, the EU Medical Device Regulation, HIPAA, and Nigeria’s NDPR. No single tool currently on the market offers this combination of capabilities in an integrated system. If implemented, it would allow hospitals to adopt AI diagnostic tools with a level of oversight and accountability that does not currently exist in most clinical settings.

Palama’s second patent application addresses a different but equally pressing vulnerability. As genomic medicine expands, healthcare systems are generating and storing vast quantities of genetic data used for diagnosing hereditary conditions, tailoring treatments, and advancing research into rare diseases. This data is among the most sensitive information a hospital can hold, yet the cybersecurity frameworks protecting it were not designed for the scale or complexity of modern genomic databases. Palama’s invention, developed as a co-inventor, proposes an AI-governed cybersecurity framework that integrates predictive threat intelligence to anticipate attacks before they occur, intelligent access governance to control who can view and use genetic data and under what conditions, automated compliance monitoring to ensure ongoing adherence to data protection regulations, and secure collaborative research capabilities that allow institutions to share genomic data for research purposes without exposing it to unauthorised access. The framework is designed to support both the security and the scientific utility of genetic data, recognising that overly restrictive protections can impede the medical research that patients ultimately benefit from.

These inventions did not emerge in isolation. They are grounded in a substantial body of published research. Palama has authored more than ten peer-reviewed articles in international journals on topics including bias and transparency in clinical AI, the implementation of governance frameworks in hospital settings, regulatory fragmentation between HIPAA and FDA oversight, and the alignment of technical robustness with legal accountability in high-risk AI systems. Her research on evaluation harnesses for clinical large language models, which provides methods to quantify hallucination, bias, and the leakage of protected health information, earned the Best Research Paper Award from Well Testing Journal in 2025. She also serves as a peer reviewer for Scientific Reports, published by Springer Nature, and PLOS ONE. The patents, in other words, represent the applied extension of a research programme that has already been validated by the international scholarly community.

For Nigeria and much of Africa, where hospitals are beginning to adopt AI-assisted diagnostic tools often developed abroad, the questions Palama’s work addresses are not theoretical. They are immediate and practical. Who verifies that an AI tool trained on data from American or European patients produces safe results for African patients? Who ensures that the genetic data collected in a Lagos hospital is protected against breaches? Who audits the algorithms making recommendations that doctors rely on? The tools to answer these questions are still being built. Palama is among those building them.