Imagine a world where we could predict the devastating progression of Amyotrophic Lateral Sclerosis (ALS) with precision, potentially leading to earlier interventions and improved treatments. This is no longer just a dream. Groundbreaking research from the University of St Andrews, the University of Copenhagen, and Drexel University has developed AI models that can predict the degeneration of neural networks in ALS, offering a glimpse into the future of this debilitating disease. But here's where it gets controversial: could these models eventually reduce our reliance on animal testing, a practice that has long been a subject of ethical debate? And this is the part most people miss: while animal models have been the cornerstone of ALS research, computational models can fill in the gaps, providing a more comprehensive understanding of disease progression.
Published in Neurobiology of Disease, this study introduces computational modeling as a powerful complement to traditional animal and in vitro methods. Motor neuron disease (MND), of which ALS is the most common subtype, affects approximately 2 out of 100,000 individuals globally each year. In Scotland alone, this translates to about 200 new diagnoses annually. ALS typically begins in the spinal cord, where motor neurons and specific neural circuits are first affected, leading to early symptoms like muscle weakness, stiffness, and cramps.
Traditionally, researchers rely on genetically modified mice to study ALS, observing their symptoms at specific timepoints due to time and budget constraints. However, computational models can predict what happens between these timepoints, offering a more continuous view of disease progression. Unlike animal models, which are influenced by numerous variables, computational models allow researchers to isolate and test the impact of specific changes with precision.
Here’s the game-changer: these models can also predict how neural circuits might respond to treatments, guiding future preclinical studies in mice. The researchers used biologically plausible neural networks—a far cry from the neural networks powering facial recognition or ChatGPT. These networks mimic the spike signals used by nerve cells in our nervous system and are structured based on known spinal cord biology.
The models, developed by the School of Psychology and Neuroscience, consist of mathematical equations that calculate the excitability of each neuron. When a neuron receives an electrical impulse (a spike), its excitability changes, and if it reaches a threshold, it passes the signal to the next neuron. Neurons are grouped into populations, connected based on biological data, to construct the network.
Co-author Beck Strohmer explains, “During ALS, neurons die, and communication between populations breaks down. We model this by removing neurons and reducing connections, allowing us to simulate disease progression. Similarly, we can test treatment strategies by saving neurons or strengthening communication.”
Dr. Ilary Alodi adds, “While hypotheses from models must be validated in animal studies, they provide invaluable guidance. In this study, our model predicted that a specific treatment would save a particular neuron population. When we examined treated mice, the prediction held true.”
These findings highlight the potential of computational models to refine experimental research and reduce the need for animal testing. Dr. Alodi notes, “We’re now applying these models to specific brain areas to understand neuronal communication changes during dementia, an exciting new direction for our lab.”
But here’s the question that sparks debate: As computational models become more sophisticated, could they one day replace animal testing entirely? Or will they always serve as a complementary tool? Share your thoughts in the comments—we’d love to hear your perspective on this groundbreaking yet controversial topic.
