New horizons in cancer treatment: How spatial intelligence drives advances in precision medicine KellyOnTech
I have read a series of recent progress of AI in the medical field, which is very exciting. Today, I will introduce the latest progress of spatial biology.
Spatial Intelligence
Speaking of spatial biology, let’s first talk about the spatial intelligence brought to the forefront by Professor Feifei Li. Artificial intelligence has realized “listening” and “speaking” through natural language processing (NLP), and learned “seeing” through computer vision, but this is far from enough. Because “seeing” is for “doing” and “learning”. The core of human spatial intelligence lies in learning and acting through perception. This ability enables us to complete tasks efficiently in three-dimensional space, and it is also the key to the development of artificial intelligence to a higher level.
Take one of Professor Feifei Li’s pictures as an example. When you see the milk cup in the picture, do you subconsciously want to hold it? This impulse reflects the innate spatial intelligence of human beings, which closely connects our perception and action.
Professor Feifei Li’s new AI startup, World Labs, has raised $230 million from notable investors. Leading investors include Radical Ventures, a Canadian venture capital firm introduced before, where Feifei Li and Kai-Fu Lee are both scientific partners. Professor Geoffrey Hinton, who just won the Nobel Prize in Physics, is also an investor. In addition, Nvidia’s venture capital department also provided support. World Labs aims to develop software that can use images and other data to make decisions about the three-dimensional world and create a so-called “big world model”.
Spatial Biology Latest?Progress
Despite significant progress in cancer treatment in recent years, its efficacy remains low. Even when patients have known biomarkers, accurately predicting treatment outcome is still challenging. This difficulty stems from the complexity of cancer pathogenesis, as each individual has unique genomic, transcriptomic and proteomic characteristics, which may play a key role in disease progression and treatment response. How to provide highly personalized treatment plans for each patient and provide actionable insights into its efficacy has become a pressing issue.
One solution is to utilize spatial biology, artificial intelligence, and multimodal spatial analysis to reveal the hidden circuits of cancer.
领英推荐
Spatial biology is a relatively new research field, and papers on spatial multi-omics are growing exponentially. Studies have shown that the spatial positioning of cells is critical to understanding heterogeneity within tumors. By observing cells and tissues in a three-dimensional environment, similar to the way a global positioning system (GPS) records location coordinates to generate maps and track targets, we can use these techniques to map the spatial structure of cells and reveal the interaction between cells and their microenvironment, thereby discovering a large amount of information that cannot be captured by traditional sequencing or other technologies.
Spatial intelligence enables researchers to map the location of various cells and molecules within a tumor. This mapping is crucial because the organization of cancer cells and their surroundings significantly affects tumor growth and treatment response. By understanding this layout, scientists can better predict cancer behavior and tailor personalized treatment plans for patients.
In drug development, spatial intelligence is applied to study how drugs interact with cells in specific locations within the body. This precision is especially vital for diseases requiring targeted interventions, such as directing drugs to tumors or inflamed areas without affecting healthy tissue.
Moreover, cells do not function in isolation; they constantly communicate with one another. Spatial intelligence allows researchers to map these interactions, showing how the location of certain cells influences their function. For instance, the proximity of immune cells to a tumor impacts their ability to mount an effective response against cancer.
The practical application of this field is to generate artificial intelligence digital twins of patients, analyze tumor biopsies before cancer treatment, and predict response to treatment and drug resistance. By using multi-omics data to reveal the heterogeneous dynamics and hidden complexity of the tumor microenvironment, the project currently predicts patient outcomes with an accuracy of more than 90% and generates revenue by providing microdomain graph search models to biopharmaceutical companies.