TEL AVIV, Israel — Clalit Health Services has been highlighted in a new World Health Organization report examining how artificial intelligence can move from experimentation into routine health care while maintaining patient safety, clinical oversight, and trust.
The WHO Regional Office for Europe published the report, Bridging Theory and Practice: Implementation Insights on Artificial Intelligence in Health Care, on June 26. It examines 11 case studies and practical approaches to implementing AI in health systems. Three of the case studies are from Israel, including two involving Clalit.
Among the approaches highlighted is OPTICA, a framework Clalit developed to systematically assess AI applications before incorporating them into clinical practice.
The model contains 77 checklist items across 13 sections. It assesses AI systems across areas including clinical suitability, organizational data, development and performance, and deployment and post-implementation monitoring. Clalit researchers previously detailed the framework in the peer-reviewed journal NEJM AI.
Clalit says every AI product intended for use within the organization undergoes its assessment and risk-stratification process.
The approach addresses a rapidly emerging challenge for health systems worldwide. Artificial intelligence is moving into hospitals, clinics and diagnostic services faster than many governments and health organizations are developing mechanisms to govern it.
WHO/Europe reported in July that nearly two-thirds of countries in its European region were already deploying AI in diagnostics, while only 8 percent had a health-specific AI strategy. Only 8 percent had liability standards defining responsibility when an AI system fails.
A 77-Point Check Before AI Reaches Patients
Unlike a conventional approval process that focuses primarily on whether an algorithm performs adequately during testing, OPTICA is designed to examine whether a particular AI application is appropriate for the health organization and clinical environment in which it will operate.
The framework brings together different participants in the assessment process, including clinical experts, AI developers, organizational data specialists, machine-learning operations experts and organizational AI leadership.
It also extends scrutiny beyond initial deployment.
Prof. Ran Balicer, Clalit’s chief innovation officer and head of its Innovation Division, said the organization developed OPTICA because AI systems require continued oversight after entering clinical practice.
“We created an organizational process called OPTICA, in which every AI product intended for implementation at Clalit goes through a checklist of up to 77 items,” Balicer said in a statement provided by Clalit.
“These items address every aspect of the product, including safety, privacy protection, accuracy, and continuous monitoring over the years, because the product itself may change.”
That emphasis on continuing evaluation reflects one of the central difficulties surrounding medical AI. Performance can vary by patient population, clinical setting, available data, and changes made to an AI product after deployment.
The OPTICA framework therefore requires organizations to monitor stability, output consistency, and continued performance once an application has been implemented.
AI Targets Undiagnosed Hepatitis C
Another Israeli example involves predictive technology developed by researchers at the Clalit Research Institute to identify people at elevated risk of carrying hepatitis C without knowing they are infected.
Clalit traditionally screened approximately 50,000 people to identify previously undiagnosed carriers. According to Balicer, that approach detected only 38 positive cases among 50,000 first-time screenings.
Researchers then developed a predictive model using Clalit’s extensive longitudinal patient information to identify people considered most likely to have hepatitis C.
When approximately 500 people identified by the model as being at elevated risk were screened, 38 positive cases were found.
That represented roughly a 100-fold improvement in screening efficiency compared with the broader screening approach.
The significance extends beyond the algorithm’s predictive performance. Successful implementation requires physicians to understand why a patient has been flagged and to incorporate the recommendation into normal clinical practice.
Balicer has said explainability has been important to gaining physician confidence in Clalit’s predictive systems. The organization now uses AI-generated recommendations in primary care to identify patients who may benefit from preventive interventions or treatment changes.
AI Moves Into Medical Imaging
A third Israeli case described in material surrounding the report involves infrastructure developed by Israeli medical AI company Aidoc, which enables multiple medical-imaging algorithms to operate within a common clinical environment.
Such systems can help prioritize scans and assist clinicians as health systems contend with growing imaging volumes and workforce constraints.
The implementation challenge, however, extends beyond installing an algorithm.
Clinical responsibility, integration into existing workflows, and ongoing performance monitoring remain essential, particularly when imaging equipment, software, or underlying clinical conditions change.
That reflects the broader message emerging from WHO’s work on AI governance.
WHO Warns Governance Is Lagging Behind AI
WHO/Europe has increasingly focused on the gap between the rapid adoption of artificial intelligence and the systems needed to govern it.
Regional Director Dr. Hans Henri P. Kluge warned in July that almost 40 percent of countries in the WHO European Region still lacked ethical guidance governing AI in health care.
“The future of AI in health won’t be decided by algorithms,” Kluge said. “It will be decided by the frameworks we build now, the partnerships we forge, and the political will we bring to making sure this technology serves everyone.”
WHO has identified governance, legal accountability, data management, workforce preparedness and responsible investment as central issues as artificial intelligence becomes increasingly integrated into health systems.
The organization’s April assessment of AI across the European Union found that 74 percent of EU countries were already using AI-assisted diagnostics. It also stressed that health professionals remain legally and ethically responsible for clinical decisions involving technologies they may not fully understand.
That reality makes organizational assessment frameworks increasingly important.
For Clalit, the objective is not simply to demonstrate that artificial intelligence can produce accurate predictions. It is to establish whether a system is clinically appropriate, whether the underlying data are suitable, whether physicians can use it effectively, and whether its performance can be continuously monitored.
The WHO report’s case studies point toward a broader transition already underway in digital medicine.
The central question is increasingly not whether artificial intelligence can work in health care. It is whether health systems can build the governance, clinical processes and human oversight necessary to ensure that it continues to work safely when placed in the hands of physicians and used in the care of real patients.
TheJ.ca receives no government money. Independent journalism depends on readers who believe these stories must be told. Consider supporting our efforts.
Stand Proud. Stand Strong. Stand Zionist.
