IBM Research

Boehringer Ingelheim and IBM have announced a partnership aimed at advancing generative artificial intelligence (AI) and foundation models for therapeutic antibody development. The collaboration agreement will see Boehringer use an IBM-developed, pre-trained AI model that will be “further fine-tuned” on the German drugmaker's specific proprietary data to help accelerate the pace at which it can create new antibody therapeutics. The companies noted that, despite “major" technological advances, the discovery and development of therapeutic antibodies against diverse targets remains a "highly complex and time-consuming process". IBM’s foundation model technologies, which have already shown success in generating biologics and small molecules with relevant target affinities, are used to design antibody candidates for specific disease targets. These are then screened with AI-enhanced simulation to select and refine the best binders for the target. Boehringer Ingelheim outlined that it will produce small quantities of the candidates that can be tested experimentally. Andrew Nixon, global head of biotherapeutics discovery at Boehringer Ingelheim, said: “I am confident that by joining forces with IBM scientists we will develop an unprecedented platform for accelerated antibody discovery which will enable Boehringer to develop and deliver new treatments for patients with high unmet need.” Alessandro Curioni, vice president, accelerated discovery, IBM Research, added that the company was “thrilled” to bring its multi-modal foundation model technologies to Boehringer “to help accelerate the pace at which Boehringer can create new therapeutics”. The collaboration comes just days after Boehringer said it would be acquiring bacterial cancer therapy specialist T3 Pharmaceuticals in a deal worth over $500m, marking a significant boost to its immuno-oncology portfolio. The transaction gives the company access to T3’s proprietary therapy platform, which uses live bacteria to deliver immune-modulating proteins directly to cancer cells and tumour micro-environments while sparing healthy tissues. Boehringer also entered into an exclusive research collaboration and worldwide licensing agreement with Covant Therapeutics in March to develop new cancer immunotherapies, and announced a collaboration and licensing agreement with 3T Biosciences at the beginning of the year to discover and develop next-generation cancer therapies.

IBM’s foundation model technologies have already proven their worth when it comes to “generating biologics and small molecules with relevant target affinities,” the tech giant said. Boehringer Ingelheim will harness an artificial intelligence model trained in the computer labs of IBM as the German drugmaker becomes the latest Big Pharma to apply machine learning tools to its drug discovery efforts. The foundation model tech, which has been developed and trained at IBM, will be “further fine-tuned on additional proprietary data” provided by Boehringer so the pharma can discover new antibody therapeutics. “We are very excited to collaborate with the research team at IBM, who share our vision of making in silico biologic drug discovery a reality,” Andrew Nixon, global head of biotherapeutics discovery at Boehringer, said in the Nov. 28 release. “I am confident that by joining forces with IBM scientists we will develop an unprecedented platform for accelerated antibody discovery which will enable Boehringer to develop and deliver new treatments for patients with high unmet need.”  The idea is that by inputting information on the sequence, structure and molecular profile of specific disease targets, the AI model can conduct simulations of experiments to identify antibody molecules with the highest likelihood of success. These can then be tested in the lab, with the results fed back into the AI model to fine-tune the results further. IBM’s foundation model tech has already proven its worth when it comes to “generating biologics and small molecules with relevant target affinities,” the tech giant said. IBM is no stranger to drug discovery, having granted Pfizer access to its IBM Watson cognitive computing capabilities back in 2016. Earlier this year, Moderna signed on to explore the use of quantum computers in developing its future medicines. The company has been “at the forefront of creating generative AI models” that extend the technology’s use beyond language models, Alessandro Curioni, vice president of accelerated discovery at IBM Research, explained in this morning’s release. “We are thrilled to now bring IBM’s multimodal foundation model technologies to Boehringer, a leader in the development and manufacturing of antibody-based therapies, to help accelerate the pace at which Boehringer can create new therapeutics,” Curioni added.

From left to right: Taha Jangda, managing director, Ascension Ventures; Beth Andrews, chief digital health officer, life sciences & healthcare, Dell Technologies Data is vital in advancing healthcare, as it helps providers understand patients better and make more educated decisions on their care. However, the vast majority of health data isn’t used, according to one healthcare executive. “It’s difficult to connect, curate, normalize and work with data keys. Think of monitoring patients, there’s data in the EHR, you can run AI on that data. But what you’re not seeing or what you’re missing there is, for example, it’s estimated 900,000 data points of patient data on a critical care bed are thrown away per patient per hour. … 97% of the data today in healthcare is not used and it’s not used because it’s difficult to wrangle,” said Beth Andrews, chief digital health officer of life sciences and healthcare at Dell Technologies. Andrews made these comments during a fireside chat at the MedCity News INVEST Digital Health conference, which was held last week in Dallas. Prior to working at Dell Technologies, Andrews worked at Medtronic and Moving Analytics. The session was moderated by Taha Jangda, managing director of Ascension Ventures. Pharma and clinical trials are good examples of when data is often not being used because the data is “disparate and siloed,” Andrews added. There is a way to improve the process of using data, however. “Normally setting up to use federated learning or to kind of clean that data to get it that [protected health information], that can be a 12 to 18-month process. If you’re able to do something like confidential computing with a secure enclave, you can actually move really quickly within a couple of months,” Andrews said. “This is really key for research, say for rare disease and needing to tap into different sources of data and do that with confidential computing the whole way. You can work on [protected health information] and you can also speed up that process.” Federated learning is a process for training AI models “without anyone seeing or touching” the data, according to IBM Research. Confidential computing, meanwhile, refers to a cloud computing technology that separates sensitive data in a protected central processing unit enclave while it processes the data, IBM states. Andrews also talked about the importance of data in cancer care. Dell Technologies recently joined CancerX, a public-private partnership launched by the White House that aims to spur innovation in the treatment of cancer. “Three decades ago in 1990, there were 37 drug-related chemotherapy options with 60 indications,” Andrews said. “Today, there are well over 500 approved anti-neoplastic agents and it’ll keep growing. So we really need to be harnessing data and real-world evidence.”