In addition to answering all kinds of questions or generating images that never existed, artificial intelligence (AI) is also becoming relevant in its application to medicine. Today, the magazine Nature publishes a study in which AI improves the results of humans in the evaluation of echocardiograms, a type of images used to diagnose cardiac ailments. The authors, a multidisciplinary team at Cedars-Sinai Medical Center in Los Angeles, conducted a randomized, blinded trial—the first of its kind with this technology—to assess the accuracy of the AI by analyzing 3,495 ultrasound scans used to view muscle function and the valves of the heart.
In the study, cardiologists were asked to rate the initial assessment of ultrasound technicians who typically do that job and those performed by artificial intelligence. Doctors made corrections to the machine in 16.8% of cases compared to 27.2% of human technicians. Furthermore, the cardiologists were not able to distinguish which were assessments made by the AI or by a person. As summarized by the authors, “for patients who underwent echocardiographic quantification of their cardiac function (…) the AI assessment was not inferior to that of echocardiographers.”
Those responsible for the work, led by cardiologist David Ouyang, do not believe that this type of technology will replace the work of professionals. “We hope it will help specialists save time and minimize the more tedious parts of the cardiac imaging job,” he says. The cardiologist, at least for now, will continue to have the last word.
At Cedars-Sinai, they are also applying AI to other aspects of cardiac prevention. A team led by Sumeet Chugh has begun an observational study with around 400,000 people in search of a more precise way to assess the risk of ventricular fibrillation, a type of arrhythmia that can cause sudden death. The accumulation of large amounts of information to train the AI would allow, according to those responsible for this project, to predict with greater precision who should undergo an intervention because their risk of heart failure is high. Other uses of AI put to the test by this same institution make it possible to improve the quality of images obtained by lower quality scanners until they match those of others obtained by higher quality devices.
Amparo Martínez, president of the Cardiac Imaging Association of the Spanish Society of Cardiology (SEC) and cardiologist at the Hospital Clínico Universitario de Santiago, explains that, now, in the latest generation workstations, “there are already semiautomatic programs that are used to analyze different measurements of an echocardiogram. It is something that saves time, but an expert assessment is still necessary, ”she adds. This type of technology, she assures, “is helping a lot and will help more, it can serve to speed up the work, because it is faster than if you carry out the measurement, but the work of the doctor will always be necessary,” she concludes. .
Ouyang believes that his technology can help make medical treatment more affordable, but his words, at least in the case of technicians who evaluate ultrasound scans, do not rule out that AI could replace human labor. “The cardiologists would still look at it and make sure it’s done right. That said, in our trial, the cardiologists were not able to distinguish between the work done by the technicians and the AI and saw that the preliminary assessment of the AI was better and required fewer changes, ”he summarizes.
As with technologies like ChatGPT, the possibility of leaving something as sensitive as medical diagnosis in the hands of systems whose operation is sometimes not fully understood has raised suspicions. However, experts like Regina Barzilai, who has developed algorithms to improve cancer diagnosis, believe that the danger lies in not taking advantage of the potential of a technology that can improve the accuracy of medical diagnosis and make it cheaper. “There is an overemphasis on the risks of this technology, but if you look at healthcare systems around the world, they often don’t include it. It is only in some hospitals where research is done. It is like talking about the dangers of a house on Mars when no one is building houses on Mars ”, she opined in an interview with EL PAÍS. “The danger of artificial intelligence”, she concluded, “is in delaying its application to medicine, because there are many people who suffer from insufficient care, medical errors or high health costs”.
You can follow THE COUNTRY Health and Well-being in Facebook, Twitter and instagram.
