NicheCompass, a groundbreaking artificial intelligence (AI)-based neural network, is revolutionising how scientists monitor and interpret millions of cells to predict chemical changes in tissues. This breakthrough system, developed by researchers at the Wellcome Sanger Institute, the Institute of AI for Health, Helmholtz Munich, and the University of Würzburg, can predict where customised treatments for diseases such as cancer will be most effective.

NicheCompass, as described in a research study published in Nature Genetics on March 18, uses spatial genetic data to map cell types, locations, and communication patterns. It reads a cell's social network, revealing alterations that aid in determining how different individuals may respotherapies – all - all within an hour.

Every cell in the human body is part of a broader communication network, and each is identified by distinct surface proteins. Single-cell and spatial genomic technologies have dramatically improved our understanding of the human body, resulting in the creation of comprehensive cell atlases that specify cell types, locations, and how genetic changes affect their interactions.

Despite this advancement, it is still difficult to comprehend cell neighbourhoods and measure their social connections. NicheCompass solves this problem by training deep-learning models to understand how cells communicate and align comparable cellular networks. This skill allows researchers to pose important questions like, "How do cancer cells interact with surrounding tissue in lung cancer patients?"

Using NicheCompass, researchers examined data from ten lung cancer patients to detect similarities and differences between them. These findings improve our overall understanding of cancer while suggesting possible targets for tailored therapy. Dr. Carlos Talavera-López, a co-senior author at the University of Würzburg, emphasised NicheCompass's capacity to identify changes in immune cell interactions with lung cancer tumours in patients. This real-world application not only revealed fresh information that contributes to our overall understanding of cancer, but it also highlighted one patient whose disease worked differently on the immune system. In the future, NicheCompass could aid in the discovery of new strategies to harness the immune system in certain tumours, resulting in personalised treatments that enable a patient's immune system to specifically attack cancer pathways.

Researchers also effectively applied the method to breast cancer tissue, demonstrating its potential for treating various types of cancer. In addition, researchers used NicheCompass to analyse a mouse brain's spatial atlas with 8.4 million cells. The model recognised brain regions quickly and accurately, resulting in a visual representation of the entire organ and demonstrating its potential for spatial atlas analysis globally.

The most innovative feature of NicheCompass is its capacity to allow clinicians to enter patient data and receive specific insights about an individual ailment in under an hour. This speed and accuracy could greatly improve clinical decision-making and patient outcomes.

Sebastian Birk, the first author at the Institute of AI for Health, Helmholtz Munich, and the Wellcome Sanger Institute, underlined the need for a large amount of data about the human body to better understand, prevent, and cure diseases. However, we also require tools that enable us to reap all of the benefits that this information may give. NicheCompass represents a tremendous advancement in this field, using the power of AI while also providing interpretability, allowing researchers and clinicians to ask questions about their data and better understand and treat diseases."

Dr. Mohammad Lotfollahi, co-senior author of the Wellcome Sanger Institute, made a comparison between cell communication and human social networks:

"People frequently communicate to their networks with a variety of information. They may discuss business or vacation photos and send them to different friends, but they can all lead to the same person. Cell-to-cell communication is similar; cells may employ different qualities to interact with their social network, forming communities or networks in their immediate surroundings. NicheCompass is the first AI model of its sort capable of interpreting these networks and answering questions that could have a direct influence on patient lives, such as identifying when and how health issues began and predicting how they could respond to specific treatments."

As researchers continue to enhance and expand NicheCompass, it is poised to become a cornerstone of precision medicine, allowing physicians to personalise treatments more effectively and improve patient outcomes worldwide.