WRAIR Identifies Biomarkers with AI-Powered Research Methods
Integrative Systems Biology at the Walter Reed Army Institute of Research utilizes artificial intelligence, commonly called AI, in research methods that identify biomarkers, biological indicators that a person is at greater risk for military relevant health issues.
In Integrative Systems Biology, scientists look for biomarkers that can assist with identifying psychiatric conditions such as post-traumatic stress disorder or acute stress in service members and veterans. These biomarkers provide useful information during diagnosis, prognosis, monitoring, and interventions to ensure soldier’s resilience and preparedness.
WRAIR’s Integrative Systems Biology has been refining and improving these AI-powered tools over many iterations, resulting in more than 200 publications and more than 40 oral or poster presentations at scientific conferences around the world.
“We have been developing these AI methods for more than 14 years,” said Dr. Ruoting Yang, a bioinformatician and principal investigator with WRAIR’s Integrated Systems Biology. “My research methods use deep learning and other advanced machine learning approaches tailored to biomarker discovery and systems biology in psychiatry.”
Large language models, which power tools such as ChatGPT, Gemini, and Claude, are AI systems built using deep learning and trained on large volumes of text to summarize, translate, and generate natural language. This type of AI is designed to look for reoccurring patterns in text and compare them to one another. Multimodal versions of these models can also interpret images and other visual information.
“We use large language models extensively to accelerate coding and improve documentation for our machine learning and deep learning research,” explained Dr. Yang. “Our AI models are trained to identify patterns in biological data.”
Machine learning, including deep learning, uses statistical methods to learn patterns from data. At WRAIR, scientists train and evaluate these models to analyze large datasets and look for biomarkers, which are biological indicators that are associated with different aspects of human health.
“We call this field precision psychiatry,” said Dr. Yang. “Being able to better identify psychiatric conditions is an important way to improve and evaluate the benefits of treatments for service members with mental health conditions.”
Scientists are pivoting to machine learning for data analysis across many fields because it can process larger volumes of data very quickly and efficiently once it is accurately and precisely trained.
“We break large volumes of data into two sets: one for training and one for validation,” said Dr. Yang. “First you ‘teach’ the AI what to do on the training data. Then you give it the validation data and evaluate how successfully it accomplishes the task.”
Once the AI is refined enough to successfully analyze the validation data, then the scientists can use it on independent test data sets, which are new data sets whose information hasn’t been used in the training process.
“Our data sets can be very large,” said Dr. Yang. “One of our largest data sets has to do with studying DNA methylation and PTSD. It’s more than 10,000 samples – and each individual sample has millions of data points.”
WRAIR scientists have turned to deep learning methods, because they allow for the processing of these large data sets quickly and with increasing complexity.
“On top of speed, AI offers refinement,” explained Dr. Yang. “In the early stages of this field, people had to look at spreadsheets and do their calculations for each sample by hand. Now we can run massive data sets very quickly. Training more complex AI algorithms allows us to break open black boxes and learn new things about the relationship between the brain and the body.”
Learn more about Integrated Systems Biology at the website: https://wrair.health.mil/Biomedical-Research/Center-for-Military-Psychiatry-and-Neuroscience/Integrated-Systems-Biology/
Legal Disclaimer:
EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
