Introduction—A Supervised Approach to Training an Artificial Intelligence to Extract Relevant Genomic Data from Literature
Adam B -
Psychiatry is one of the most fascinating yet frustrating fields of medicine. Unlike other specialties (take oncology, the study and treatment of cancer, which relies on measurable biological markers and lab tests), psychiatry still depends heavily on observed symptoms, patient reporting, and the subjective observations of a doctor. There are no simple blood tests or scans that can definitively diagnose depression, schizophrenia, or bipolar disorder, which means that psychiatric diagnoses are more ambiguous, more often incorrect, and more inconsistent with treatments.
Advancements in understanding genomics (and proteomics), the study of genes (proteins) and how they influence health, have shown how even specific genes (or proteins) and their up-regulation or down-regulation can be the root cause of disease. Understanding what causes this up-regulation or down-regulation of genes and proteins (external stimuli like medicine, trauma, etc.) is the goal of epigenetics. By combining these two fields, scientists can draw relationships between specific external stimuli TO genes TO disease which may uncover points to understand potential treatments to psychiatric disease.
There are hundreds of papers with evidence that certain genes, proteins, and molecular processes could play a role in psychiatric disorders. Buried among these research papers likely are recurrently reinforced relationships between genes, disease, and treatment. The challenge is that assessing all of these papers simultaneously produces enormous amounts of data due to how complex both phenotypic expression is (when we say someone has depression, what does that mean empirically compared to a “normal” person? What is a normal psychiatric state?) and what specifically changes with a particular experimental condition is (when a doctor prescribes an antidepressant, it doesn’t only act on one protein in one place through one mechanism). This is where artificial intelligence (AI) comes in.
This project focuses on developing a supervised AI model, meaning it will learn by comparing its predictions to human-reviewed data, gradually improving its ability to identify patterns. The ultimate goal is to create a tool that could help make psychiatric diagnoses more objective and precise, bringing the field closer to the standards of other medical specialties. As the weeks go by, I’ll explain more about how this process works experimentally.
Comments:
All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.