WHO Recognizes Clalit Among Leading Models for AI in Healthcare
Israel’s largest healthcare organization, Clalit Health Services, is among four international approaches highlighted by the World Health Organization (WHO) for developing frameworks to guide the safe use of artificial intelligence in healthcare.
In a new report examining the practical implementation of AI across European health systems, the WHO looks at how healthcare organizations are moving from experimentation to integrating these technologies into clinical care. The report examines 11 case studies from different countries, including three from Israel, two of which involve Clalit.
A Framework for Evaluating AI Before Deployment
At the heart of Clalit’s contribution is OPTICA, a framework developed to evaluate artificial intelligence tools before they are deployed in clinical practice.
The model is based on a checklist of up to 77 criteria, divided into 13 sections and four main areas: clinical suitability, data assessment, development and performance evaluation, and deployment and monitoring.
The process involves five categories of stakeholders, ranging from clinical specialists to those responsible for the organization’s AI infrastructure.
Importantly, the evaluation does not end once an AI tool is approved. OPTICA also requires continuous monitoring after deployment to ensure that the system remains safe and effective over time.
Using AI to Identify Patients at Risk
Another Israeli case presented in the report concerns research conducted by the Clalit Research Institute on hepatitis C.
Researchers developed a model using 25 years of medical records to identify people at high risk of carrying the virus without knowing it.
In a screening campaign involving approximately 50,000 people, only a few dozen carriers had previously been identified. By targeting approximately 500 people whom the model identified as being at higher risk, physicians detected 38 new carriers.
The targeted approach therefore significantly improved the efficiency of screening.
However, the WHO emphasizes that the success of the system was not based solely on the performance of the algorithm. Its integration into physicians’ daily workflows and the system’s ability to provide understandable explanations also played a key role.
Addressing the Shortage of Radiologists
The third Israeli case concerns Aidoc, an Israeli company specializing in artificial intelligence for medical imaging.
Clalit uses a platform that allows multiple algorithms developed by different providers to operate simultaneously. The technology is particularly relevant amid a shortage of radiologists and the growing number of medical imaging examinations.
Once again, the WHO emphasizes the importance of governance. Healthcare professionals must remain responsible for the deployment of these tools, while AI systems must be continuously monitored, particularly following equipment or software updates.
Other European healthcare systems are reaching similar conclusions.
At Helsinki University Hospital, 60 AI tools were evaluated, but only 15 were ultimately deployed — approximately one quarter of the solutions assessed.
The figure illustrates an increasingly important distinction between artificial intelligence that works technically and technology that can operate safely and effectively in a real-world medical environment.
According to the WHO, only a minority of the projects examined use formal assessment frameworks capable of measuring the relationship between AI deployment and actual clinical outcomes.
For healthcare systems, this has become one of the major challenges of the AI revolution. A high-performing algorithm is only the starting point. It must also be integrated into existing clinical practices, understood by healthcare professionals, protect patient data and remain reliable over time.
Israel Seeks to Remain at the Forefront of Medical AI
The WHO’s recognition places Clalit among organizations developing structured approaches to managing the use of artificial intelligence in healthcare.
It also highlights Israel’s growing role in the development of medical technologies, at the intersection of its technology ecosystem and healthcare system.
Clalit’s model offers one possible answer: AI should not simply be approved and deployed. It must be evaluated beforehand and continuously monitored afterward.
The next stage of AI in medicine may therefore be less about developing the most powerful algorithm and more about building the strongest mechanisms for oversight, safety and accountability.
Caroline Haïat
Comments