Scientists Use AI’s Learning Process to Decode Complex Cell Data todayheadline

Hebrew University researchers have developed a new method that examines how artificial intelligence learns to classify cells, turning the AI system’s learning process itself into a valuable source of biological insights. The approach could help scientists better understand cellular development and disease progression.

The method, called Annotatability, monitors how artificial neural networks learn to identify different cell types and states. Instead of focusing solely on the AI’s final classifications, the researchers analyze how quickly and confidently the system learns to recognize different cells.

This novel approach addresses a fundamental challenge in modern biology: how to accurately classify and understand the vast diversity of cells in living organisms. While current technologies can measure the activity of thousands of genes in individual cells, interpreting this data and accurately labeling cell types remains difficult.

The research team, led by Jonathan Karin, Reshef Mintz, Barak Raveh, and Mor Nitzan at Hebrew University, published their findings in Nature Computational Science. The study demonstrates how their method can identify misclassified cells, discover transitional cell states, and track disease progression at the cellular level.

Using Annotatability, researchers analyzed cells from various tissues, including blood, pancreas, and cancer samples. The system proved particularly effective at identifying cells in transitional states – those changing from one type to another – which are often difficult to classify using traditional methods.

In one application, the team examined pancreatic cells from diabetic mice receiving different treatments. The method revealed subtle differences in how individual cells responded to treatment, potentially offering new ways to evaluate treatment effectiveness.

The technique could also help researchers better understand how cells change during disease development. When studying cancer cells transitioning from one state to another, the system identified cells at various stages of transformation, providing insights into disease progression.

The researchers have made their method freely available to other scientists, potentially accelerating research across many areas of biology and medicine. While still in its early stages, the approach offers a new way to extract meaningful patterns from the increasingly complex data generated by modern biological research.

The development comes at a crucial time as biologists grapple with ever-larger datasets generated by new technologies that can measure thousands of features in individual cells. By leveraging AI’s learning process itself as a source of insight, the method offers a novel approach to understanding the complexity of living systems.