There are plenty of cases where artificial intelligence (AI) is being used in the healthcare setting. However, perhaps the most effective one is disease tracking. By analyzing records over time, AI is able to spot things that human doctors may miss.
This principle is on display thanks to IBM’s new AI that tracks the progression of Huntington’s disease. The model can help predict when a patient might experience exacerbations of their symptoms and determine how quickly they will progress.
Finding New Connections
Huntington’s disease is a fairly rare genetic disorder that has strong hereditary ties. The Huntington’s Disease Society of America estimates that more than 200,000 Americans are currently at risk for inheriting the condition.
The neurodegenerative disease is both devastating and debilitating. Symptoms often begin as subtle cognitive changes but gradually progress to uncontrolled movements and memory loss. Eventually, Huntington’s leads to death.
Although there are genetic tests available to help diagnose the condition, people can begin displaying symptoms at almost any age. Some may appear in a 20-year-old while other people don’t experience them until their 60s. This leaves a lot of uncertainty for people who may be at risk of inheriting Huntington’s from a parent.
To help address this problem, IBM joined forces with the CHDI Foundation to create an AI model that is more effective than genetic testing. The algorithm can predict when a patient will begin developing Huntington’s symptoms.
The duo was able to accomplish this by training the AI with a series of MRI brain scans. They also utilized signals from the brain’s white matter to teach the program how cognitive changes occur. This is a novel approach compared to other traditional methods.
Unlike standard genetic testing, the team believes that its AI-based approach can also predict how a patient’s symptoms will evolve over time. That is huge news for the Huntington’s community. Until now, the precise evolution of symptoms has been more of a guessing game than a science.
The research team is optimistic that this is only the beginning of a breakthrough. They hope that they will soon be able to provide more accurate symptom forecasts for patients with as little as one scan.
An IBM blog post written by one of the study’s co-authors says, “The results demonstrate that measuring the potential progression of [Huntington’s disease] could eventually be done from a single brain scan, acquired in one visit.”
The team’s research was published in the journal Scientific Reports on Monday.
Although the AI can only predict symptoms right now, it may help guide treatments in the future. For example, doctors will have a better picture of a patient’s symptoms—possibly before they even occur. This will allow them to offer more individualized care and treatment options.
Then, researchers will be able to analyze new MRI scans to determine how the patient responds to that treatment. By repeating this process, scientists may be able to find an ideal treatment plan that helps stabilize Huntington’s patients.
For now, this breakthrough is novel. However, it may soon be used in clinical practice to battle one of the most devastating neurogenerative diseases we know of.