US Superintelligence Force resigns at OpenAI and the retina-reading AI

The study, published in the European Heart Journal-Digital Health, included 931 people without prior cardiovascular disease. 50.8% already had subclinical atherosclerosis. Aitheroscope identified 93.4% of positive cases and provided a reliable negative result more than 90% of the time. Performance was particularly good in young people and those at low or intermediate risk, precisely where traditional risk assessment tools tend to underestimate damage.

The project unites Internal Medicine, Ophthalmology, and Occupational Risk Prevention at the Infanta Leonor Hospital with Horus ML, responsible for the AI ​​development. The head of Internal Medicine, Juan Torres Macho, points out that the eye is one of the few harmless windows into the body's blood vessels: tortuous arteries and arteriovenous crossings reveal systemic damage that the AI ​​detects beyond what the human eye can see with the naked eye.

The test is quick, non-invasive, and uses equipment already available in many hospitals and health centers for diabetes monitoring. The authors recommend prioritizing healthy adults aged 18 to 50 with risk factors (hypertension, smoking, high cholesterol). A high-probability positive result should be confirmed with vascular ultrasound before treatment. The next step: external validation, integration into healthcare services, and certification for the healthcare system.

Source: The Newspaper

What connects these three fronts?

The SIF institutionalizes US self-regulation. Robinson demands that this self-regulation cease to be a marketing slogan. Aitheroscope shows the other extreme: when AI is applied with clinical evidence, it can save lives without needing a "frontier" model. Three news stories, one dilemma: who sets the pace—the product, the state, or the evidence?