Why Stem Cells & AI Are Best Friends
A not-so-unlikely friendship promises to lead to fascinating medical discoveries
The stem cell biologist James Thomson and colleagues touted stem cells as “a potentially limitless source of cells for drug discovery and transplantation therapies” in the 1980s. Meanwhile, in the article that coined the term artificial intelligence (AI), John McCarthy suggested that “every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.” Not only do the promises of both fields continue to be pursued, but researchers have now described even greater aspirations.
So it’s no surprise that the combination of these two powerful fields is particularly awe-inspiring. Despite the relative infancy of this multi-disciplinary field, research at the intersection of stem cell biology and AI is bound to lead to fascinating discoveries — and has already made some interesting contributions.
One reason the combination of these two fields can be so powerful is the rise in data collection in biology. Think about the thousands or millions of cells that researchers often study in a single experiment. Then think about the thousands of parameters researchers may be looking at in each cell — this can include everything from how long a cell survives to proteins that are present in one cell but not the neighbouring cell.
In the past, we didn’t have the tools to study individual cells in such detail. Instead, scientists would look at populations of cells more broadly. As our biological tools have become more advanced, there’s been a rise in single-cell analysis techniques which can lead to amazing findings, but also generate immense amounts of data that can be impossible to decipher and analyze for a human.
AI to the rescue! Analyzing huge amounts of data is one of the many areas in which AI shines. Researchers have recently been using AI to:
- Create cell atlases
- Predict how stem cells will behave either in the lab or when transplanted into patients’ bodies
- Test which drugs work well on cells in the lab and which ones don’t
Cell Atlases: A Periodic Table for Biologists
Imagine trying to navigate the world without a map or better yet a new city without Google Maps. Without any references as to where things are, it’s very easy to get lost! This applies to studying cells too. That’s why scientists have been creating cell atlases — catalogues of cell types based on where the cell comes from, how they behave, the proteins they create, and much more. These large catalogues are put together by hundreds of researchers performing hundreds if not thousands of single-cell experiments which makes this a fantastic reference point. Fabian Theis, a renowned computational biologist, likened these atlases to a periodic table for cells.
The cell atlases can then be used to study new sets of data. For example, researchers could look at how older patients’ cells differ from younger ones that are already in the atlas. Scientists can also study which cells are responsible for a disease of interest.
But as we’ve already talked about, single-cell analysis methods generate lots of data and this is where AI comes in and makes the creation of the atlases possible.
Applying this to the stem cell field specifically, scientists have created cell atlases of cells involved in development (which continues to be an area scientists struggle to fully understand). Another approach has been creating an atlas of organoids — mini 3D structures of cells in the lab created from stem cells. Organoids are very useful in modelling and understanding how diseases impact specific patients without testing drugs on the patients themselves.
Using AI to Predict Stem Cell Biology
If organoids didn’t already sound amazing enough, what if I told you that they can self-assemble, meaning stem cells will begin to form 3D structures that are somewhat like mini organs when provided with the right conditions in the lab? Well, it’s true. However, it can be difficult to predict how changing different factors, like what proteins the cells produce, impacts how these organoids self-assemble and the patterns they form. Here’s where AI comes in once again with another of its strengths — prediction.
In one study, Ashley R.G. Libby and colleagues trained an AI model to predict how changing the proteins cells create will change the patterns of the stem cells; the model turned out to be very accurate.
Taking things to the clinic, researchers have created an AI model that can predict how leukemia patients will fare after a bone marrow transplant (which includes blood stem cells). This predictive tool can help clinicians continually assess the risks their patients face post-transplantation.
Drug Screening with AI
Another rapidly growing field that uses stem cells is drug screening — testing different drugs for various diseases on cells to predict how they’ll impact patients. There’s our magic word once again: predict. In a study by Martti Juhola and coworkers, using stem cells, they looked at cardiomyocytes, the muscle cells in the heart, from patients with a disease that causes irregular heartbeats.
After exposing the cells to different factors, they trained an AI model to be able to detect how cells behave in response to the factors. What makes this experiment and others like it powerful is that the cells are coming directly from patients with the specific disease and the AI models allow us to find new patterns and ideas researchers may not come up with on their own. While the model created was by no means perfect, it shows us how AI and stem cells can help with drug screening.
Nowadays, you can’t go one day without hearing the word AI at least once. Though stem cells aren’t talked about as much, they’ve also been a hot topic for decades now. Regardless of whether you’re sick of hearing about either of them, you’d be remiss not to admit they hold a lot of promise for changing the way we approach (and hopefully solve) medical problems.
About the Author
Parmin Sedigh is a 17-year-old stem cell and science communications enthusiast as well as a student researcher. She’s also an incoming first-year student at the University of Toronto, studying life sciences. You can usually find her on her computer following her curiosity. Connect with her on LinkedIn.
