For decades, mankind has relied on experimental data gathered by traditional research means to conquer disease. Unfortunately, that can take decades or longer to render results. New artificial intelligence (AI) powered bioinformatics, or computational biology, can predict how diseases and symptoms will react to possible treatments and cures in hours, days or weeks instead.
Indeed, AI can identify potential treatments and cures just as it can model how cells will react to them. Further, AI can also predict disease occurrences and reoccurrences, as well as disease evolution and its effect on current treatments and cures.
Bioinformatics, or computational biology, can do all that just using computational models. It can tap into hidden knowledge in experimental data gathered over many years in seemingly unrelated research projects. Or, it can work from simulated data conjured by scientific minds and couched with known factors such as the rules of physics, genetics, epigenetics, and other qualifying circumstances.
Today, artificial intelligence (AI) technology makes all of that work faster, and it’s also capable of using much larger data sets and solving far more complex problems than human minds can.
However, general artificial intelligence like that which is depicted in science fiction movies does not yet exist. Instead, today we are using subsets of AI, namely machine learning (ML) and deep learning (DL).
The easy to understand video below explains how different technologies are blended together to create more innovations in many different industries, even though it all comes from computational biology.
Advanced analytics and automation complete the modern day “smart” toolbox. All of these tools are extremely powerful and are already changing the human experience in myriad ways. But they also present some challenges too.
The video below is a good, easy-to-understand introduction to the different stages of AI and its achievements and challenges.
While machine learning and other AI tools are already making great strides in medicine, more specialized tools are emerging on the scene at a steady pace. One of the latest is scGen, “an AI-powered tool for predicting a cell's behavior in silico. scGen will help map and study cellular response to disease and treatment beyond experimentally available data,” according to the report in Science Daily.
The interesting twist in scGen’s development is the discovery that smaller is better, at least in this case. Scientists recently discovered that the large-scale atlases of healthy human organs in the Human Cell Atlas isn’t going to cut it in making disease characterizations and treatments. That’s because there are too many disease possibilities and variations impacting organs to scale the analyses. But that’s not to say that the many aspects of the Human Cell Atlas are not useful. The video below explains how important this reference map, this Atlas, is.
However, by refocusing computations at the cellular level, disease and cell reactions can generally be more readily seen and understood.
"For the first time, we have the opportunity to use data generated in one model system such as mouse and use the data to predict disease or therapy response in human patients," said Mohammad Lotfollahi, PhD student (Helmholtz Zentrum München and Technische Universität München).
One of the central goals of computational biology is “accurately modeling cellular response to perturbations, e.g. disease, compounds, genetic interventions,” say the scGen scientists.
“Although models based on statistical and mechanistic approaches exist, no machine-learning based solution viable for unobserved, high-dimensional phenomena has yet been available. In addition, scGen is the first tool that predicts cellular response out-of-sample. This means that scGen, if trained on data that capture the effect of perturbations for a given system, is able to make reliable predictions for a different system,” the report explains.
Beyond scGen and other modern-day AI tools, work is continuing to create computing approaches that can predict diseases that don’t exist today and thus for which specific data are not available. Inference based on sophisticated simulated data will likely outweigh empirical data in the computations at that point. But the payoffs could be very high, particularly in prevention.
In the short-term things like flu vaccines will improve because the predictions for this year’s reigning strain will be more accurate. But in the long-term, it means that cures or genetic “tweaking” can prevent a predicted new disease from ever getting a hold on humans. That scenario is theoretical at the moment, but few doubt its inevitability.