IBM Re-enters Healthcare with Strategic AI Partnerships: Moderna and Boehringer Ingelheim Leverage MoLFormer and Quantum Computing for Drug Discovery

On November 28, Boehringer Ingelheim and IBM announced a collaborative agreement, enabling Boehringer Ingelheim to leverage IBM’s foundation model technologies to discover novel candidate antibodies for the development of effective therapies.

Under the agreement, Boehringer Ingelheim will utilize a pre-trained artificial intelligence model developed by IBM, which will be further fine-tuned using Boehringer Ingelheim’s other proprietary data to generate target candidate antibodies.

Andrew Nixon, Global Head of Biopharmaceutical R&D at Boehringer Ingelheim, stated, “The collaboration will develop an unprecedented platform for accelerating antibody discovery, thereby speeding up the company’s creation of new antibody therapies.”

Therapeutic antibodies are central to the treatment of diseases such as cancer, autoimmune disorders, and infectious diseases; however, the discovery and development of therapeutic antibodies targeting diverse epitopes is a highly complex and time-consuming process. Artificial intelligence (AI) and deep learning are transforming multiple aspects of antibody drug development, with AI increasingly impacting more sophisticated areas—namely, antibody design and optimization.

This collaboration between Boehringer Ingelheim and IBM will serve as an “accelerator” for its antibody R&D. Building on this partnership, researchers from Boehringer Ingelheim and IBM will jointly leverage IBM’s foundation model technologies to enhance the speed and efficiency of antibody discovery, as well as to predict the quality of candidate antibodies.

Boehringer Ingelheim’s “trust” in IBM stems from the latter’s prior release of MoLFormer, a model capable of effectively screening for and generating new molecules.

MoLFormer is a pre-trained AI model that leverages self-supervised learning to train a Transformer-based neural network on two public chemical databases—PubChem (containing approximately 100 million molecules) and ZINC (containing approximately 1 billion molecules)—to learn effective, low-dimensional molecular representations. Subsequently, through fine-tuning on task-specific datasets, the model is applied via transfer learning to various downstream tasks, including molecular property prediction, computation of molecular similarity, and analysis of the correspondence between interatomic spatial distances and attention values for given molecules.

In simple terms, MoLFormer can predict molecular physical properties (such as solubility), biophysical properties (such as antiviral activity), physiological properties (such as the ability to cross the blood-brain barrier), and quantum properties (such as molecular bandgap energy). Furthermore, MoLFormer can serve as a decoder to generate newly designed proteins and small molecules with desired properties.

In actual R&D, IBM and its partners will form a closed-loop workflow. Taking Boehringer Ingelheim as an example, IBM first leverages foundation model technologies such as MoLFormer to screen for target candidate antibodies; subsequently, during the validation phase, Boehringer Ingelheim produces these candidate antibodies at a small scale and evaluates them through experimental assays.

Finally, the results of laboratory experiments will be used to improve computer simulation methods through a feedback loop. IBM then fine-tunes the pre-trained model using its specific proprietary data, ultimately delivering target antibodies with the desired properties for new designs.

Prior to Boehringer Ingelheim, IBM had already established collaborations with Cleveland Clinic and Moderna in the areas of healthcare research and biopharmaceutical R&D.

In April 2023, IBM and Moderna signed a collaboration agreement to integrate AI and quantum computing into the development of mRNA therapies, aiming to accelerate the design of mRNA drugs with optimal safety and efficacy.

Under the agreement, Moderna will participate in the IBM Quantum Accelerator program and the IBM Quantum Network. IBM will provide access to quantum computing systems and the specialized expertise required by Moderna to advance its exploration of cutting-edge applications in life sciences driven by quantum technology. Additionally, Moderna will leverage MoLFormer to characterize potential mRNA therapeutics and design a new class of vaccines and therapies.

In March 2023, IBM and Cleveland Clinic officially announced the deployment of a medical-grade quantum computer in the United States. The IBM Quantum System One installed at Cleveland Clinic will become the world’s first quantum computer dedicated to healthcare research, helping Cleveland Clinic optimize complex healthcare systems and accelerate biomedical discoveries.

This deployment builds upon a prior collaboration between IBM and Cleveland Clinic. In March 2021, IBM and Cleveland Clinic announced a 10-year partnership to jointly establish a facility named Discovery Accelerator, aiming to fundamentally accelerate discoveries in healthcare and life sciences by leveraging high-performance computing across hybrid cloud, AI, and quantum computing technologies.

Through Discovery Accelerator, researchers plan to leverage advanced computational techniques to generate and analyze data, supporting the Cleveland Clinic’s Center for Global Pathogen Research and Human Health in conducting research across genomics, single-cell transcriptomics, population health, clinical applications, and chemistry and drug discovery.

Since installing its quantum computer, the Cleveland Clinic has launched nine quantum computing projects, including: developing quantum computing pipelines to screen and optimize drugs targeting specific proteins; enhancing quantum-powered predictive models for cardiovascular risk after non-cardiac surgery; and applying artificial intelligence to search genomic sequencing results and large drug target databases to identify effective existing medications for patients with Alzheimer’s disease and other conditions.

However, just one month before IBM announced its partnership with the Cleveland Clinic, IBM was considering spinning off its Watson Health business. In January 2022, IBM sold Watson Health to Francisco Partners, a San Francisco-based investment firm.

Watson Health is a business unit that IBM invested billions of dollars to build in the healthcare sector. It primarily leverages AI to help hospitals, insurance companies, and pharmaceutical manufacturers manage data and provide cancer treatment solutions, thereby assisting physicians worldwide in diagnosing and treating patients.

“Becoming a true AI doctor” was the goal behind IBM’s development of Watson Health, as the company sought to dominate the future medical AI market. Upon launching Watson Health, IBM boldly proclaimed that “humans would be replaced,” heavily promoting its applicability across the medical, financial, legal, and academic sectors, which feature rich and diverse linguistic data. Then-IBM CEO Virginia Rometty also described the initiative as IBM’s “moonshot” and one of the company’s strategic missions.

But contrary to expectations, Watson Health did not become the “flagship product that would bring revolutionary changes to the medical community”; instead, it became IBM’s misstep in its foray into the AI healthcare sector.

“The failure to truly leverage AI” emerged as one of the core reasons for Watson Health’s downfall, as IBM did not integrate the AI technologies it heavily promoted in its marketing into Watson Health products. Dr. David Ferrucci, who led the team that built Watson, once stated, “Watson was designed to recognize word patterns and predict correct answers in quiz shows; it is not a universal problem-solver for commercial applications, and it would likely fail a second-grade reading comprehension test.”

As one of IBM’s earliest and largest forays into AI-driven healthcare, Watson Health ultimately ended in failure.

However, IBM did not “collapse,” but rather regrouped to re-enter the AI healthcare arena.

This time, IBM has shifted away from its previous “go-it-alone” approach and philosophy by partnering with highly capable organizations with strong medical expertise to establish a presence in the AI healthcare sector. Leveraging its AI and quantum computing technologies, IBM empowers innovative pharmaceutical companies to address thorny issues and challenges in healthcare and drug research and development.

From the Cleveland Clinic to Boehringer Ingelheim, IBM is gradually expanding its commercial footprint in the healthcare sector. Whether IBM can successfully secure a place in this field remains to be seen!