Beyond Correlation: Feromics Unveils Causal AI to Decode the Immune System

BOSTON, MA – April 16, 2026 – In a field saturated with promises of artificial intelligence revolutionizing medicine, a Boston-based biotech is making a bold claim: the problem isn't the AI, it's the data. At the prestigious MIT AI Conference this week, Feromics Inc. presented its Intelligent Design™ platform, an approach that seeks to fundamentally rewire how AI learns about the human immune system.

Shai Schubert, PhD, Co-Founder and CEO of Feromics, articulated a core challenge plaguing AI-driven immunology. For years, researchers have fed AI models vast seas of molecular data from immune cells, hoping the algorithms could sift through the noise to find patterns related to disease. The issue, Schubert explained, is that these patterns often only reveal correlation, not the true biological cause. An AI might learn that a certain protein is present in a sick patient, but it can't know if that protein is driving the illness, a byproduct of it, or a complete coincidence.

"The problem isn't really the AI—it's the data behind it," Schubert stated during his presentation. "If the data doesn't reflect what cells actually do, the models miss what matters. We've solved the problem at the source."

Feromics' solution is a paradigm shift that it calls "functional immunomics." Instead of starting with a cell's molecular contents, the company's platform starts by observing what the cell does.

Solving AI's 'Garbage In, Garbage Out' Problem

The Intelligent Design™ platform is built on a "biology-first" principle. It uses proprietary technology, stemming from over a decade of research by co-founder Dr. Tania Konry at Northeastern University, to isolate individual immune cells and observe their functional behavior in real-time. For instance, the platform can identify which specific T-cells are actively killing cancer cells, which are merely present, and which are exhausted.

Only after a cell's function is observed and verified is it subjected to deep molecular analysis, such as single-cell RNA sequencing. This creates what Feromics calls "function-anchored data"—a pristine dataset where every molecular profile is already labeled with a confirmed biological action. This process effectively removes the guesswork for the AI. Instead of searching for a needle of causation in a haystack of correlation, the AI is trained on data where the causal link is already established.

This approach stands in stark contrast to conventional methods that generate massive, undifferentiated datasets and rely on algorithms to infer function. By cleaning and structuring the data at the source, Feromics aims to build more accurate, predictive, and causally-aware AI models that can be trained faster and provide more translatable insights.

Backed by Government and Industry Leaders

The novelty of this approach has not gone unnoticed. Feromics is bolstered by significant validation from both government and industry. The company recently secured a $4 million, three-year contract from the Advanced Research Projects Agency for Health (ARPA-H), the high-risk, high-reward biomedical research agency. The funding is specifically targeted at using the Intelligent Design™ platform to develop precision medicine technologies that can better select treatments for patients undergoing cancer immunotherapy.

An endorsement from ARPA-H signifies a belief that the technology has the potential for transformative, not just incremental, impact on public health. Further validation comes from a Sanofi iDEA Award, recognizing the innovative potential of the platform's underlying science, which was developed in the lab of Dr. Konry, a pioneer in single-cell functional assays and microfluidics.

The company's leadership combines deep scientific expertise with proven biotech experience. Dr. Konry's academic work provides the scientific bedrock, while Dr. Schubert, a graduate of the Harvard-MIT Program in Health Sciences and Technology, brings a track record as an accomplished biotech founder, focused on translating breakthrough science into commercially viable applications.

A New Toolkit for Modern Medicine

Feromics is deploying its platform across three core areas where a causal understanding of the immune system can have an outsized impact: drug discovery, clinical development, and diagnostics.

In drug discovery, the platform promises to accelerate the identification of new therapeutic targets and drug candidates. By understanding the precise molecular pathways that drive a desired immune function—like cytotoxicity against tumors—researchers can design more effective drugs and screen them more efficiently. This moves beyond simply finding a compound that binds to a target, to understanding if that compound elicits the right biological action.

For clinical development, a notoriously expensive and failure-prone process, the implications are profound. Feromics' technology can be used to develop functional biomarkers that predict which patients will respond to a given therapy. This allows for smarter clinical trial design, enrolling patients most likely to benefit and increasing the probability of success. A particularly powerful application is in allogeneic cell therapies, where the platform can screen donors to select those whose immune cells exhibit the highest functional performance, potentially leading to more durable and effective treatments.

In diagnostics, the platform enables the creation of AI-driven models that can predict clinical outcomes. By analyzing a patient's immune cells for functional signatures rather than just static molecular markers, it may be possible to develop companion diagnostics that more accurately stratify patients and guide treatment decisions before a therapy even begins. This represents a critical step toward a future of truly personalized medicine, where treatments are tailored to the unique functional biology of an individual's immune system. By training its causal AI models and then applying them to standard sequencing data, the company ensures its breakthroughs are scalable and can be integrated into existing clinical and research workflows.

By teaching AI to understand the 'why' behind cellular behavior, this new approach may finally provide the map needed to navigate the complex terrain of human immunity.