AI and Birth Defects – A Groundbreaking Approach to Predict Drug Risks

Have you ever wondered about the power of artificial intelligence (AI)? Not just its role in creating more engaging video games or streamlining our online shopping, but its potential to profoundly impact our lives in areas such as health and wellbeing. Today, we delve into an AI application that could revolutionize prenatal care by predicting the risk of birth defects caused by medications, even before these drugs hit the market.

AI’s Detective in Healthcare

The Icahn School of Medicine at Mount Sinai in New York is leading the charge with a ground-breaking AI model, affectionately known as a ‘knowledge graph’. This AI detective sifts through mountains of data, seeking associations between genes, congenital disabilities, and drugs. The goal? To predict which existing medications, not currently classified as harmful, may actually lead to birth defects.

But that’s not all! This knowledge graph also has the potential to identify the risks associated with pre-clinical compounds, those drugs still in the testing phase, that could harm a developing fetus. The aim is to ensure safer medications and prevent the tragic outcome of drug-induced birth defects.

The Magnitude of the Problem – Birth Defects and Unknown Causes

Around 1 in 33 births in the United States involve birth defects, making them a significant health issue. These defects can be structural or functional abnormalities that occur during pregnancy, often resulting from a complex interaction of genetic and environmental factors.

Despite significant advances in medical research, the causes of most of these disabilities remain a mystery. Certain substances in medicines, cosmetics, food, and environmental pollutants can potentially cause birth defects if a pregnant person is exposed.

The AI Detective at Work – Digging Through the Data

The team of data scientists gathered information from several datasets about known associations between birth defects and various substances. They combined this data from reproductive health genetics, classification of medicines based on their risk during pregnancy, and the effects of drugs and pre-clinical compounds on human cells’ biological mechanisms.

One of the key elements of the AI model, ReproTox-KG, uses semi-supervised learning (SSL), a technique that uses a small amount of labeled data to guide predictions for much larger unlabeled data. This aspect of the model helped the team to prioritize 30,000 preclinical small molecule drugs for their potential to cross the placenta and cause birth defects.

Making Sense of the Data

While AI and data science may seem daunting, here’s a fun way to understand it: imagine you’re at a party, and there are various groups, or ‘cliques,’ where everyone knows each other. Now, imagine these cliques represent different bits of data – one clique could be the gene, another the birth defects, and a third the drug. The AI’s job is to find out how these cliques are connected, potentially revealing the mechanisms by which certain drugs might cause birth defects.

The Future of AI in Healthcare

The findings from this study are preliminary, and additional experiments are needed for validation. However, the initial results offer a promising glimpse into how AI could transform healthcare.

The team behind the study plans to use similar AI techniques for other projects focusing on the relationship between genes, drugs, and diseases. They also aim to use this data in training materials for bioinformatics courses, extending the reach of AI in medical research and education.

From Prediction to Prevention

Just imagine the possibilities. A future where AI helps evaluate the safety of new drugs, protecting unborn babies from harmful substances. A world where regulatory agencies like the U.S. Food and Drug Administration and the U.S. Environmental Protection Agency could use AI models to assess new drugs’ risks or other chemical applications.

This is more than just the stuff of science fiction. It’s the potential reality offered by the blending of AI and healthcare, marking another giant leap forward in our ongoing quest to improve human health and wellbeing.

John Erol Evangelista, Daniel J. B. Clarke, Zhuorui Xie, Giacomo B. Marino, Vivian Utti, Sherry L. Jenkins, Taha Mohseni Ahooyi, Cristian G. Bologa, Jeremy J. Yang, Jessica L. Binder, Praveen Kumar, Christophe G. Lambert, Jeffrey S. Grethe, Eric Wenger, Deanne Taylor, Tudor I. Oprea, Bernard de Bono, Avi Ma’ayan. Toxicology knowledge graph for structural birth defectsCommunications Medicine, 2023; 3 (1) Link

Leave a Reply

Your email address will not be published. Required fields are marked *