A new study led by researchers from the University of St Andrews, the University of Copenhagen, and Drexel University has developed AI-powered computational models that predict how neural networks degenerate in amyotrophic lateral sclerosis (ALS). Published in Neurobiology of Disease, the research introduces a novel approach using biologically plausible neural networks to simulate disease progression and test potential treatments, offering a powerful complement to traditional animal and laboratory-based methods. ALS, also known as motor neuron disease or Lou Gehrig’s disease, affects motor neurons in the brain and spinal cord, leading to progressive muscle weakness, stiffness, and cramps. The disease typically begins in the spinal cord, with symptoms appearing first in the limbs. Globally, about two in 100,000 people are diagnosed annually, with around 200 new cases in Scotland each year. Traditionally, ALS research relies on genetically modified mice that display ALS-like symptoms. However, these models are limited by cost, time, and the difficulty of observing continuous disease progression. Researchers must analyze specific timepoints, missing the dynamic changes between them. Computational models overcome these constraints by simulating the entire disease timeline and allowing scientists to test the impact of single variables—such as neuron survival or synaptic strength—without the confounding factors present in living organisms. The new models use biologically plausible neural networks that mimic real neurons by communicating through spike signals, similar to how nerve cells function in the human body. These networks are structured based on known anatomical and connectivity data from the spinal cord. Each neuron’s activity is governed by mathematical equations that calculate its excitability. When a neuron receives a spike, its excitation level changes, and if it reaches a threshold, it fires, passing information to connected neurons. To model ALS, researchers gradually removed neurons from specific populations and reduced connections between them—simulating the neurodegeneration seen in patients. This allowed them to predict how the disease spreads through neural circuits over time. They also tested potential treatment strategies by simulating neuron preservation or enhanced connectivity. Co-author Beck Strohmer from the University of Copenhagen said the models can simulate both disease progression and therapeutic interventions. “We can test how saving certain neurons or strengthening connections affects the network’s overall function,” she explained. Dr. Ilary Alodi from the University of St Andrews, who led the study, emphasized the importance of validating model predictions in real biological systems. “We predicted that a certain treatment would protect a specific group of neurons. When we tested this in mice, the result matched the model’s prediction,” she said. This alignment demonstrates that computational models can guide and refine animal research, reducing the number of experiments needed and improving their precision. The findings suggest that AI-driven models can accelerate the discovery of effective treatments by identifying promising targets and strategies before testing in animals. The team is now expanding their approach to study brain networks in dementia, opening new avenues for understanding neurodegenerative diseases.