Quris: AI & Patient-on-a-Chip for Faster Drug Development

The Quest for Alternatives to Animal Testing in Drug Discovery

The reliance on animal testing during the drug discovery process presents a significant ethical challenge. Despite this, a viable substitute for utilizing mice remains elusive, even acknowledging their limitations as accurate models of human physiology.

Quris proposes a groundbreaking solution by integrating artificial intelligence with data generated from a “patient on a chip” system. This combination promises robust testing capabilities and automation, all while substantially reducing costs – eliminating the need for animal subjects.

Securing Funding for Innovation

The company has successfully secured $9 million in seed funding to transition from pilot programs to full-scale production. A distinguished group of investors and advisors signals the potential merit of this innovative approach.

Update: The company announced in January an expansion of the round to $28M, with Welltech Ventures leading the full round, alongside contributions from iAngels, GlenRock Capital, and other investors. Further expansion in December brought the total to $37M, including a $9M investment from Softbank.

The Core Concept: Enhanced Human Body Simulation

The underlying principle is straightforward: develop a more refined, small-scale simulation of the human body. This simulation would generate data readily interpretable by a machine learning system.

While conceptually simple, execution is complex. However, Quris swiftly translated the initial research into tangible progress.

Leveraging "Organs on a Chip" Technology

The Israel-based company’s methodology builds upon significant research conducted at Harvard concerning “organs on a chip.” These systems, while relatively new, are now well-established in the field.

They utilize small quantities of stem cell-derived tissue – known as “organoids” – as a testing platform for drugs or treatments. This provides valuable insights into how human organs, such as the liver, might respond to various substances.

Creating a Functional Human System

Harvard researchers discovered that connecting multiple organ-on-a-chip systems (e.g., liver, kidney, and heart cells) creates a remarkably effective simulation of the human body.

Although not a perfect substitute for a living organism, this serial organoid system, or “patient on a chip,” offers a promising alternative to traditional mouse testing. Currently, treatments that pass mouse testing only achieve success in human trials approximately 10% of the time.

From Research to Scalable Product

According to Isaac Bentwich, CEO and co-founder, the team recognized the potential immediately upon the publication of the Harvard study. They began focusing on the engineering and AI advancements necessary to transform this experimental system into a scalable, commercially viable product.

This isn’t merely a replacement for mouse testing; it’s a cost-effective method for conducting limited human testing without involving actual humans, and without the inherent uncertainties associated with animal models.

A rendering illustrating the anticipated appearance of the fully automated “chip on chip” device. Image Credits: Quris

A Paradigm Shift for Pharmaceutical Companies

“Consider a pharmaceutical company,” Bentwich explained. “Would they prefer to wait until the cusp of clinical trials to discover that a promising molecule lacks efficacy? While genomic discoveries are valuable, they cannot circumvent the limitations of mouse experiments, where failure rates are as high as 90%. This allows you to identify the most promising candidates before investing in costly clinical trials.”

Given the substantial costs associated with drug candidate development – often exceeding hundreds of millions of dollars to reach the clinical stage – even a modest investment (tens of millions) to eliminate potential failures is justifiable. If the technique proves accurate, the return on investment is substantial, potentially avoiding the expense of pursuing ineffective compounds.

As Bentwich articulated, this approach introduces the “fail early, fail cheap” philosophy of software development to a field where such principles were previously unattainable.

The Quris System: Efficiency and Automation

The Quris system employs a “chip-on-chip” technique, utilizing multiple organoid systems (chips) in sequence. This design is significantly smaller and more efficient than current state-of-the-art laboratory systems.

Simulating hundreds of humans using Harvard’s methodology would require millions of dollars. However, the Quris system achieves this at a fraction of the cost, utilizing less biological material, incorporating automation, and leveraging a sophisticated machine learning model.

The Power of AI-Driven Interpretation

A key aspect of Quris’s approach is the unique dataset generated, which fuels a specialized AI capable of analyzing and interpreting experimental results. The AI is continuously trained using existing and emerging drugs, learning to correlate sensor signals with substance safety.

This enables effective testing with a limited number of chips, rather than the hundreds of mice traditionally required.

Personalized Medicine Through Chip Technology

The chips themselves are not uniform. Through careful manipulation and selection of stem cells and tissues, different patient types and conditions can be simulated. This allows for testing against diverse genetic predispositions and complicating factors in an automated environment.

For example, if a drug demonstrates efficacy but causes side effects in 10% of patients, testing against various genetic backgrounds may reveal the underlying causes of these adverse reactions.

Members of the Quris team working in a laboratory setting. Image Credits: Quris

AI as an Enabling Technology

As the AI catalogs and analyzes this data, it is expected to become highly proficient at predicting drug candidacy for human trials from a relatively small number of automated tests. Without AI interpretation, the data presents a complex challenge requiring extensive expertise.

However, Bentwich emphasized that they do not envision eliminating the biological component entirely. “It’s fundamental to our philosophy and biological understanding that the AI must collaborate with a biological counterpart,” he stated.

Expert Validation and Future Adoption

Robert Langer, co-founder of Moderna, a member of the scientific advisory board, expressed his concurrence and anticipates rapid adoption of this technique. However, he noted that larger, more conservative pharmaceutical companies may be slower to embrace it.

“This appears to be a significant opportunity,” Langer said. “I’ve explored similar concepts in other areas of chemistry, leveraging AI for predictive modeling. It won’t replace testing entirely, but it will significantly narrow the possibilities and accelerate the process.”

While support from figures like Langer (along with Nobel laureate Aaron Ciechanover) is valuable, Bentwich believes their patent portfolio and first-mover advantage will be crucial for gaining market entry. An agreement with the NY Stem Cell Foundation provides them with privileged access to the organization’s stem cell workflow.

A Dual Business Model

Quris operates under a dual business model. One involves providing drug screening services to pharmaceutical companies, with payment contingent on the accuracy of the results. For instance, if a drug cleared by the system successfully reaches a predetermined testing milestone, payment is triggered.

The other prong focuses on developing their own drugs. Currently, the company is developing a treatment for Fragile X condition, a genetic disorder linked to Autism, which is slated to enter clinical trials next year.

Beyond Traditional Drug Discovery

Bentwich highlighted that despite the proliferation of AI-powered drug discovery companies, few have a molecule resulting from their work entering clinical trials. This isn’t due to a lack of capability, such as identifying molecules with specific bioreactivity or developing efficient manufacturing methods, but rather the numerous hurdles in the lengthy discovery, testing, and approval process.

The $9 million seed round “provides ample funding to finalize the productization of our system, enhancing its efficiency and automation, and to test the first hundred or thousand drugs to train the AI,” Bentwich explained. The round was led by “Dr. Judith Richter and Dr. Kobi Richter, pioneers of cardiovascular intervention therapeutics, with participation from Moshe Yanai, a disruptive data-storage technology leader, and strategic angel investors.”

A Vision for Personalized Medicine

Bentwich envisions a future of “completely personalized medicine.” As the cost of stem cells continues to decline, new markets will emerge.

“It will no longer be limited to expensive experiments for pharmaceutical companies. In five to ten years, this technology may be accessible to hundreds of millions of people. Currently, the process is somewhat barbaric. You consult a pharmacist who lists potential side effects, but you remain uncertain. Are you the guinea pig? The answer is: yes, we all are. But this is a crucial step towards a more informed and personalized approach.”