In a study published in Cell Reports, scientists at Gladstone Institutes employed a new video-based machine learning approach to uncover previously undetectable signals of early disease in mice designed to mimic key characteristics of Alzheimer's.
Senior Investigator Jorge Palop (left) and staff scientist Stephanie Miller (right) review data from an open-source machine learning tool called VAME. The tool analyzed video footage of mice to provide new insights into behaviors indicative of early Alzheimer’s disease. Image Credit: Gladstone Institutes
Subtle indications of Alzheimer’s disease might appear decades before a diagnosis, frequently in the form of erratic behaviors that indicate very early stages of brain dysfunction.
However, scientifically, finding and evaluating these little behavioral changes has been impossible, even when investigating Alzheimer’s in mice.
Their findings offer insight into a novel technique for detecting neurological disease earlier than previously conceivable and tracking its progression over time.
We have shown the potential of machine learning to revolutionize how we analyze behaviors indicative of early abnormalities in brain function. We have leveraged a valuable tool that opens the door to a more complete understanding of devastating brain disorders and how they begin.
Jorge Palop, PhD, Study Senior Author and Associate Professor, Gladstone Institutes
To examine video footage of mice exploring an open arena, the scientists employed a machine learning framework called VAME, which stands for “Variational Animal Motion Embedding.” The open-source technology recognized minor behavioral patterns captured on camera—changes that could not be seen just by glancing at the mice.
Tracking Disorganized Behavior
VAME’s deep learning platform differs from conventional behavioral tests in mice, which frequently involve predefined tasks that the animals are asked to execute.
According to Stephanie Miller, PhD, a Gladstone staff scientist and the study's first author, one disadvantage of such tests is that they cannot capture the whole spectrum of spontaneous behavioral changes induced by disease, particularly in the early stages. They also lack scalability and frequently employ labor-intensive procedures.
The Gladstone study used VAME to analyze two groups of mice that represented various features of Alzheimer'’. In both mouse models, the machine learning technology detected a considerable rise in “disorganized behavior” as the animals aged. For example, the mice had odd behavior patterns and switched between different activities more frequently, which could be linked to memory and attention problems.
Similar machine learning approaches could be used one day to study spontaneous behaviors in humans, potentially providing early diagnosis of neurological diseases. I envision this technology will be used to assess patients in the clinic and even in their homes. It gives scientists and doctors a way to solve the very hard problem of diagnosing preclinical stages of disease.
Stephanie Miller, PhD, Study First Author and Staff Scientist, Gladstone Institute
Miller first experimented with VAME several years ago when the technology was still in its infancy. She and Palop worked with Stefan Remy, MD, and his team in Germany to design the platform from the beginning. They collaborated to demonstrate VAME’s relevance for neurological research in a report published in Communications Biology.
Evaluating a Potential Treatment
The Gladstone team added another layer to their new investigation by using VAME to determine whether a potential therapeutic intervention for Alzheimer's might prevent disorganized behavior in mice.
The scientists used previous research from Gladstone investigator Katerina Akassoglou, PhD, who revealed that a blood-clotting protein called fibrin causes a chain reaction of hazardous effects when it spills into the brain via damaged blood vessels. By preventing fibrin's harmful effects, Akassoglou’s lab has prevented neurodegeneration that leads to cognitive decline and protects animals against Alzheimer’s.
To determine whether this treatment technique could protect mice against Alzheimer's-related behaviors, the researchers genetically inhibited fibrin from causing harmful inflammation in the brain. The intervention was effective, minimizing the development of aberrant behaviors.
It was highly encouraging to see that blocking fibrin’s inflammatory activity in the brain reduced virtually all of the spontaneous behavioral changes in Alzheimer’s mice, reaffirming that fibrin and the ensuing neuroinflammation are key drivers of the disease. Machine learning can offer an unbiased way to evaluate potential treatments in the lab—and I believe it may ultimately become an invaluable clinical tool, as well.
Katerina Akassoglou, PhD, Senior Investigator, Gladstone Institute
Palop and Miller are collaborating with other Gladstone teams studying neurological diseases to enable them to use the VAME technology for new behavioral research.
Miller added, “My goal is to make this tool and similar approaches more accessible to biologists and clinicians in order to shorten the time it takes to develop powerful new medicines.”
Journal Reference:
Miller, S. R. et. al. (2024) Machine learning reveals prominent spontaneous behavioral changes and treatment efficacy in humanized and transgenic Alzheimer’s disease models. Cell Reports. doi.org/10.1016/j.celrep.2024.114870