Baylor College of Medicine Researchers Introduce A New Machine Learning Approach To Predict Schizophrenia Risk Using A Blood Test

454
Source: https://www.nature.com/articles/s41398-021-01496-3.pdf

Researchers at Baylor College of Medicine developed an innovative approach to analyzing the genome, which offers the diagnosis of Schizophrenia. Hoping to reveal an epigenetic marker for Schizophrenia, a method was obtained through a machine learning algorithm called SPLS-DA to analyze certain regions of a human genome called CoRSIVs. 

The team identifies epigenetic markers, a profile of methyl chemical groups in DNA collected from blood samples. The epigenetic markers differ between people diagnosed with Schizophrenia and people without the disease. The team developed a machine learning model to assess an individual’s odds of having the disease. Testing this machine learning model on a self-sufficient dataset shows that it can identify Schizophrenia patients with 80% accuracy.

The study was published in the journal Translational PsychiatryDr Robert A. Waterland, professor of paediatric – nutrition at the USDA/ARS Children’s Nutrition Research Center at Baylor, said, “Schizophrenia is a devastating disease that affects about 1% of the world’s population”. Even though environmental components and genetics seem to be responsible for this, ongoing proofs describe that other factors like epigenetic could also be accountable for it.

Epigenetics is a system of the molecular making of DNA that tells genes to turn on or off for different cells in the body cell type. Thus epigenetics markers can differentiate normal tissues with one individual. This process makes it difficult to evaluate whether epigenetics changes contribute to diseases involving the brain like Schizophrenia. 

To explain this problem, Waterland and his researchers recognized a specific region in the genome. DNA methylation is a standard epigenetic marker that differentiates between people but is consistent in different tissues in a person. CoRSIVs are the region that correlates with a systematic interindividual variation. Researchers suggest that examining CoRSIVs is a novel path to exhibit epigenetic causes of disease.

Source: https://www.nature.com/articles/s41398-021-01496-3.pdf

Methylation patterns in CoRSIVs in the blood sample can be analyzed as they are the same in an individual’s tissues to conclude epigenetic regulation on other parts of the body that are tough to evaluate, such as the brain. 

The researchers explained that numerous learners had analyzed methylation profiles in blood samples to recognize epigenetic differences amongst individuals with Schizophrenia. 

Researchers focused on CoRSIVs and applied the SPLS-DA machine learning algorithm for the first time to analyze DNA methylation. It was fascinating for scientists to use machine learning in medicine. Also, the results were very compelling. Machine learning algorithms suggested the possibility of risk of Schizophrenia in the early stages and found a new perspective that may apply to other diseases. 

Recent research is also innovative because it examines vital prospective factors that other studies did not consider. For example, methylation patterns in blood cells can be affected by elements such as smoking and taking antipsychotic medication, which is both common in schizophrenia patients. 

Researchers took numerous approaches to evaluate whether the methylation patterns detected at CoRSIVs were affected by smoking and medication use. New studies can rule out what Waterland said. DNA methylation at CoRSIVs is confirmed in a very early stage of life, indicating that the epigenetic differences identified between schizophrenia patients and healthy individuals 

there before the disease was diagnosed may contribute to the condition. 

The researchers were able to get much stronger epigenetic signals related to Schizophrenia than ever been done before using this novel approach. 

Paper: https://www.nature.com/articles/s41398-021-01496-3.pdf

Source: https://www.eurekalert.org/news-releases/924267