How Will Neuro-Symbolic AI Impact the Pharmaceutical Industry in 2025?

Artificial intelligence has transformed many industries and processes, and some decision-makers are exploring how to apply AI to meet specific needs. They are also interested in how different types of AI could help them.

Some business leaders have begun relying on neuro-symbolic AI. Its algorithms mimic the structure and function of neural networks in human brains and recognize symbols associated with representations or relationships. While neural networks excel at tasks such as image recognition and language processing, symbolic AI uses rule-based systems and knowledge graphs to demonstrate reasoning and explain its conclusions.

These characteristics make neuro-symbolic AI an exciting prospect for many demanding business applications. Pharmaceutical leaders are among those examining the best ways to use it. How can this technology help?

1. Enabling More Effective Drug Repositioning

Drug repositioning uses a pharmaceutical product to address a need other than the one for which the company initially marketed it. Sometimes, particularly during clinical trials, participants experience therapeutic effects other than what drug developers expect. Such occurrences open new treatment pathways for patients with specific symptoms and increase profitability potential for the responsible companies.

Researchers believe neuro-symbolic AI will speed drug repositioning efforts. They explained that it can predict a specific drug’s ability to treat a certain condition.

This type of AI could also help product development teams understand the therapeutic mechanism that causes desirable patient results. Confirming that aspect could facilitate the further exploration of the effects, such as how they might positively affect similar conditions.

Another potential application is to use this AI to study differences between cell and tissue types, such as the abundance of particular proteins across tissue types and how they affect each other.

This targeted information could show pharmaceutical experts where to focus when exploring whether drugs well-established for their ability to treat one condition could also improve others. In addition to helping patients, these circumstances can be cost-effective for companies because the associated clinical trials are often shorter and less expensive.

The pharmaceutical industry is not alone in harnessing AI’s predictive capabilities. Manufacturers often depend on it to learn about impending equipment breakdowns or repair needs before they happen. That approach allows them to prevent costly and disruptive outages that cause production delays.

2. Identifying Appropriate Patient Cohorts More Efficiently

A cohort is a group of people with shared characteristics. In the pharmaceutical industry, patient cohorts are groups who simultaneously receive medical treatments, often during clinical trials. Because these participants usually share health conditions, their experiences can be crucial to patient trial outcomes and the likelihood of specific drugs becoming commercially available.

In July 2024, a company offering a neuro-symbolic AI system for cohort identification published results indicating that its tool is superior to traditional methods of finding patients for clinical trials. Its model demonstrates longitudinal and large-scale reasoning, enabling it to screen electronic medical records and track information throughout a person’s whole health care experience. In experiments, this option performed more than 42% better than competitors during some tasks.

These tests involved a 1,400-patient dataset and showed neuro-symbolic AI found relevant patients by combining large language models with domain-specific knowledge. Those familiar with the tool believe it could also enhance patient stratification and targeted interventions in addition to its applicability to clinical trials.

3. Furthering Personalized Medicine

Many physicians, patients and pharmaceutical decision-makers are highly interested in personalized medicine. Although many drugs are medical miracles for those who take them, they can also cause bothersome side effects because every patient is different and may react badly to a prescribed drug. However, personalized medicine could reduce that possibility because it creates therapeutics for individuals rather than the masses.

Although AI does not possess humans' consciousness or emotional reasoning capabilities, it can assist with data-driven decision-making. That makes it excellent for the extremely detailed content of medical records.

In one case, researchers created a neuro-symbolic system to interpret the medical notes of patients with breast cancer. Their approach combined large language models with symbolic reasoning. The resulting tool could understand disease-related descriptions in clinical documentation. It then creates a knowledge base that connects precise medical terminology to its corresponding context.

The team hopes to devote additional research to making this resource less costly to train and more applicable to resource-constrained computing environments. They also identified its potential to assist those treating or learning about other diseases. However, this tool is already off to a good start since it is 58% more accurate than a similar model.

Once pharmaceutical development teams can interpret medical notes in the correct context, they can increase their understanding of previous treatments’ effectiveness and pitfalls. Patient-specific information could become invaluable for creating personalized treatments, although protecting people’s privacy is crucial. 

A Bright Future for Neuro-Symbolic AI

Neuro-symbolic applications in the pharmaceutical industry are not yet widespread. However, the promising examples here and elsewhere make it worthwhile for executives to remain open to them and continue evaluating how to incorporate them into existing processes.