Key Takeaways
- The Israel Medical Association’s position paper offers guiding principles—not patient‑specific protocols—for integrating AI into clinical settings while preserving physician responsibility.
- As of May 2024, the FDA has cleared 882 AI‑enabled medical devices, with radiology accounting for 76 % of approvals.
- Israeli health systems are already deploying AI tools such as Sheba’s Aidoc (radiology alert system), the Rounds voice‑to‑note platform, and Clalit’s AI‑PRO for proactive primary‑care screening.
- Core risks include biased or unrepresentative training data, a false sense of security, and AI “hallucinations” that could lead to erroneous recommendations.
- Ethical and legal safeguards stress mandatory disclosure of AI use to patients, clear liability frameworks, and informed‑consent considerations.
- Successful implementation demands a structured process: necessity assessment, regulatory vetting, risk‑management planning, practical staff training, and continuous performance monitoring.
- The paper concludes that AI should serve as a safety layer that augments—never replaces—the clinician’s judgment, emphasizing transparency, human oversight, and ongoing control.
Purpose and Scope of the Position Paper
The Israel Medical Association’s new position paper, released through the Institute for Quality in Medicine and the Israeli Society for Risk Management and Patient Safety in Medicine, sets out general principles for integrating artificial intelligence–based systems into medical practice. It explicitly states that the document is not a guideline for treating an individual patient or a recommendation for a specific product; rather, it defines how medical organizations should introduce AI tools into environments where a mistake can cost lives. As the paper notes, “position papers are intended to serve as a tool for medical staff members, and do not replace their judgment in any given situation.” This framing underscores the intent to support clinicians while preserving their ultimate responsibility for patient care.
Current Landscape of AI Medical Devices
The rapid spread of AI in medicine has accelerated since generative AI entered public use, but many applications predate that trend. According to the position paper, as of May 2024, the US Food and Drug Administration has approved 882 medical devices that use artificial intelligence. Radiology dominates the field, with 671 devices (76 % of all approvals), followed by cardiovascular devices (10 %), neurology (3 %), hematology (1.9 %), gastroenterology and urology (1.5 %), and anesthesia (1 %). These numbers illustrate how AI is already embedded in diagnostic imaging, risk assessment, and workflow prioritization across specialties.
Sheba Medical Center’s Aidoc: Enhancing Radiology Safety
In Israel, the change is visible in practice. At Sheba Medical Center, the Aidoc system—developed internally—assists radiologists by scanning CT and X‑ray studies, detecting suspicion of emergencies such as stroke, cerebral hemorrhage, pulmonary embolism, aortic dissection, or pneumothorax, and alerting radiologists. The system “scans imaging tests, detects suspicion of emergency situations … and alerts radiologists,” causing the suspicious test to rise to the top of the work‑list with a marked area. Importantly, Aidoc does not replace medical interpretation; it adds a safety layer that helps clinicians manage high volumes of images during demanding shifts.
Rounds: Alleviating Physician Burnout through Automated Note‑Taking
Another innovation addresses a major source of doctor burnout: documentation. The Rounds system records the medical visit, transcribes the conversation, and generates a full visit document or medical summary from it. From the physician’s perspective, this allows them to “look at the patient, listen, ask, examine, and at the end review the summary, correct it, and sign it,” rather than splitting attention between the patient and a screen. The paper highlights that, from a risk‑management standpoint, using such a system requires “strict maintenance of privacy, information security, clarification to the patient, and medical control ensuring that the generated summary accurately reflects what was said and done.”
Clalit Health Services’ AI‑PRO: Personalized Preventive Care
Clalit Health Services has rolled out an AI‑based platform called AI‑PRO, built on the C‑Pi platform, to support family doctors in proactive and personalized medicine. The system “scans medical information from computerized records every night, cross‑references it with clinical guidelines and knowledge bases, and surfaces recommendations and patients at risk to the doctor.” Examples include flagging a diabetic patient who has missed tests, a woman with uncontrolled hypertension, individuals at risk for osteoporosis, or medication combinations needing review. Crucially, the decision remains with the doctor: the system raises a flag, and the clinician decides whether to contact the patient, adjust treatment, order a test, or disregard the suggestion.
Inherent Risks: Bias, Over‑reliance, and Hallucinations
While the promise is great, the position paper warns that the first risk relates to the technology itself: an AI system’s output is only as good as the data on which it was trained. If the data are “partial, biased, or unrepresentative of the patient population,” recommendations may be skewed, potentially causing errors for under‑represented age groups, ethnicities, or complex comorbidities. A second risk is a false sense of security; when a computerized tool delivers a fast, confident answer, clinicians under time pressure may be tempted to accept it uncritically. Finally, the paper flags the danger of AI “hallucinations”—responses that appear reliable but are incorrect, which in medicine could translate into wrong drug advice, misinterpreted test results, or missed diagnoses.
Ethical and Legal Imperatives: Transparency and Accountability
The guidelines stress that ethical use of AI hinges on transparency. The position paper quotes the Chairman of the Ethics Bureau of the Medical Association, who said in a Knesset discussion: “The doctor must disclose that the answer to the patient is based on artificial intelligence.” This requirement links AI accuracy to the trust between doctor and patient. Legal questions remain unresolved: Who bears liability if a clinician follows an erroneous AI recommendation? Is informed consent needed for each AI‑assisted encounter? Must patients be told that a physician’s answer relied on a computerized tool? The paper calls for organizations to pre‑define permissible use cases, authorized operators, and required levels of human oversight to mitigate these uncertainties.
Implementation Framework: From Assessment to Ongoing Oversight
To responsibly adopt AI, the position paper outlines a step‑by‑tep process. First, organizations must verify that the system is truly necessary and improves a medical or administrative process without compromising safety. Next, they should confirm regulatory approval or professional endorsement, even for tools not classified as medical devices. A formal risk‑management exercise follows: identify failure scenarios, devise solutions, craft an implementation plan, establish event‑reporting mechanisms, and train end users. Training must be practical—covering system operation, alert meanings, data inputs, limitations, when to consult experts, and how to document overrides of AI advice. Finally, continuous monitoring is essential because model performance can drift as patient populations, treatment protocols, or data evolve; initial success does not guarantee sustained reliability.
Balancing Promise and Prudence: AI as a Supportive Tool
The authors conclude that artificial intelligence can serve as a safety layer, shortening interpretation times, restoring face‑to‑face interaction with patients, reminding clinicians of forgotten tests, identifying risks, alerting to deterioration, and streamlining workflows. However, these benefits hinge on clear boundaries, transparency, appropriate consent, information security, staff training, documentation, and ongoing control. As the paper reiterates, “the doctor remains the one who listens, examines, decides, explains, and bears the professional responsibility.” By positioning AI as an adjunct rather than an autonomous decision‑maker, the guidance seeks to harness technological advances while safeguarding the human core of medical care.
https://www.jpost.com/health-and-wellness/health-and-wellness-around-the-world/article-899364