AI Tool Modeled on Fraud Detection Identifies Proteins Linked to Diseases: Study

Jerusalem: Israeli scientists have developed a novel artificial intelligence (AI) tool that can identify key proteins involved in human diseases, potentially paving the way for more targeted and effective medical treatments. Drawing inspiration from fraud detection systems used in social networks, the tool, named Weighted Graph Anomalous Node Detection (WGAND), analyzes complex biological interactions to highlight proteins that play a central role in various bodily functions.

The study, published in the journal GigaScience, was conducted by researchers from Ben-Gurion University of the Negev. WGAND uses graph-based machine learning techniques similar to those employed in cybersecurity to detect suspicious behavior. Instead of analyzing social connections, the algorithm examines protein-protein interaction (PPI) networks—intricate maps that illustrate how proteins communicate and work together in the body.

Proteins are essential to nearly every process in the human body. They interact with other proteins in highly dynamic and intricate networks, which are often disrupted in disease. WGAND identifies "anomalous" proteins—those that display unusually high or unique interaction patterns within these networks. According to the researchers, such proteins are likely to be central to critical biological processes, especially in disease development or progression.

By focusing on these outlier proteins, the algorithm highlights molecular targets that might otherwise go unnoticed. In tests, WGAND outperformed existing detection methods by successfully identifying proteins linked to neurological and cardiovascular disorders. It also revealed proteins essential to nerve signaling and muscle movement, demonstrating its potential for broad application in biomedical research.

Professor Esti Yeger-Lotem, a senior researcher on the project, emphasized the tool's clinical potential: “This innovative algorithm has the potential to pinpoint which proteins are important in specific contexts, helping scientists to develop more targeted and effective treatments for various conditions.”

Co-researcher Dr. Michael Fire added, “It’s exciting to see how merging biology and cybersecurity expertise can lead to breakthroughs in understanding human biology. WGAND is a prime example of how interdisciplinary approaches can drive innovation in healthcare.”

The development of WGAND highlights the growing power of AI in biomedical discovery, offering new avenues to decode disease mechanisms and accelerate drug development.