Better than experts .. Artificial Intelligence performs in the detection of ovarian cancer

A new study led by researchers at the Swedish “Carolinska” institute showed that artificial intelligence models could be better than human experts in identifying ovarian cancer by ultrasound images. The study published in the “Nature Medicine” journal provides promising results in the field of medical diagnosis using artificial intelligence. The first author of the study, Elizabeth Epstein, a professor of clinical science at the Carolinska Institute, said that ovaries are common and often discovered by chance. Epstein added that “there is a significant decline in the number of ultrasound experts in many parts of the world,” raising concerns about unnecessary interventions and delays in the diagnosis of cancer. She explained: “So we wanted to verify whether artificial intelligence can complete human experts in this field.” The researchers have developed a nerve network models that can distinguish between benign and malignant crops in the ovaries, after training and testing artificial intelligence on more than 17 thousand ultrasound photos of 3652 patients in 20 hospitals in 8 countries. The researchers compared the ability of diagnostic models with a large group of experts and less experienced ultrasound. The results showed that the models based on artificial intelligence performed better than experts and doctors in determining ovarian cancer, as they achieved a resolution of 86.3%, compared to 82.6% for experts and 77.7% for non -experiences. The results indicate that the nerve network models can provide valuable support in the diagnosis of ovarian cancer, especially in cases that are difficult to diagnose and in places suffering from a lack of experts in the field of ultrasound. Artificial intelligence models can also reduce the need for transfers to experts. Faster and more effective care in a simulation scenario to identify cases, reduce the support of artificial intelligence from the number of transfers by 63%, and from the wrong diagnosis rate by 18%, which can lead to faster and more cost -effective care for patients with ovarian problems. Despite the promising results, the researchers emphasize the need to perform more studies before understanding the full capabilities of the nerve network models and clinical limitations. “While ongoing research and development, artificial intelligence instruments can become a significant part of the future healthcare, which reduces the burden on experts and improves hospital resources, but we need to make sure they are adaptable to clinical environments and different groups,” says Philip Christiansen, a Ph.D. In the Epsin Research Group. Currently, researchers are conducting proactive clinical studies to determine the safety and practical effectiveness of the artificial intelligence instrument in clinics. Future research will also include a random study of multi -centers to study its impact on patient management and healthcare costs.