How AI could change the way Kenyans access health insurance

 Artificial intelligence in healthcare. [Courtesy]

Artificial Intelligence is moving beyond the world of technology and into one of Kenya’s most important healthcare conversations: how patients access, understand and pay for medical care.

For health insurers, AI is increasingly being viewed as a tool that could simplify complicated insurance products, speed up claims processing and help patients make better decisions about where to seek treatment.

At AAR Insurance Kenya, the technology is already being integrated into several areas of the business, including medical claims, customer support and healthcare services.

But as the insurer expands its use of AI, questions around patient privacy, algorithmic bias and human oversight are becoming just as important as the technology itself.

According to AAR Insurance Kenya Group Head of Technology Eugene Asanya, the starting point should not be the technology but the healthcare problem it is intended to solve.

“When we speak about AI, the question is: what value are we able to derive from it?” Asanya said.

One area where AI could have a significant impact is helping patients understand health insurance.

Insurance policies can contain technical language that many customers struggle to interpret, particularly when trying to establish what is covered, what is excluded and how to access treatment.

AAR is developing AI-powered tools that can allow customers to ask questions in everyday language and receive explanations without having to navigate complex insurance terminology.

“If complexity of insurance products is one of the challenges to adoption, then AI can simplify that by allowing customers to ask questions in their own natural language and receive responses they can easily understand,” said Asanya.

For patients, the potential benefit is more than convenience. Better understanding of insurance coverage could help reduce situations where patients discover limitations to their cover only when they are already seeking treatment.

AI is also being explored in the healthcare journey after a customer has enrolled.

AAR is using historical claims information to develop tools that could help members compare hospitals based on factors such as service experience and value. Such systems could give patients more information when choosing healthcare providers while helping them make better use of their medical cover.

Another potential application is medical pre-authorization, a process that can sometimes delay access to treatment when an insurer must review and approve a procedure before it is undertaken.

AAR is exploring how AI could make the process faster while retaining the controls required to protect both patients and the insurer.

“We are looking at opportunities where we can hasten the pre-authorisation process while still maintaining the necessary insurance controls,” Asanya said.

The technology is already being used in claims processing.

Asanya said approximately 40 per cent of AAR’s outpatient medical claims are processed using machine-learning models. The insurer handles about 6,000 outpatient claims every day.

Automating part of this workload can reduce the amount of manual processing required and potentially allow straightforward claims to be settled faster.

“Resolving a significant proportion of those claims without necessarily having a human in the process is quite a big achievement,” he said.

However, the growing role of AI in healthcare raises an important question: what happens when a machine gets it wrong?

AAR says human oversight remains an essential part of its systems. Human experts continue to validate models, monitor automated decisions, and provide quality assurance before claims are finalised.

“A human is still key in that process; there is always a quality assurance layer to ensure the models are making appropriate decisions,” he said.

This human oversight is particularly important because healthcare data is highly sensitive and AI systems can reproduce errors or biases contained in the information used to train them.

An AI model trained on incomplete or poorly representative data could potentially produce inaccurate or unfair outcomes.

“It can happen,” Asanya said of algorithmic bias. “But it comes back to who trained the model.”

For insurers, this means that adopting AI is not simply a matter of purchasing software. Organisations must understand the data being used, assess its quality and continuously monitor how systems make decisions.

There are also concerns about AI-generated misinformation, commonly referred to as hallucinations, where systems produce convincing but inaccurate information.

In healthcare, the consequences of such errors can be particularly serious.

Asanya said organisations therefore need to establish clear safeguards before introducing AI into sensitive processes, including ensuring that personal information is not unnecessarily exposed to AI systems.

AAR has developed internal AI governance policies alongside its existing information security and data privacy frameworks.

The company also uses enterprise-grade AI systems that can be monitored and audited, allowing it to assess how automated decisions are made.

The debate comes as Kenya's healthcare system increasingly embraces digital technologies, from electronic medical records to telemedicine and digital health platforms. AI could become another layer in this transformation, potentially helping health insurers process information faster and interact with patients more effectively.

For AI to improve the patient experience, Kenyans must be confident that their medical and financial information is protected and that automated decisions can be challenged when necessary.

“Trust is the license that the customer has given us for us to advance these digital technologies,” Asanya said.

The future of AI in health insurance, therefore, may depend less on how sophisticated the technology becomes and more on whether insurers can demonstrate that innovation is being introduced responsibly.

If deployed with appropriate safeguards, AI could help make health insurance  easier to understand, claims faster to process and healthcare choices more informed.

But when the technology touches a patient's health, money or access to treatment, the human element remains critical.