The use of artificial intelligence (AI) to advance stem cell therapy has produced exciting results, with a key role in driving recent growth and innovation. Separated into three parts, this article provides an overview the promises and challenges of stem cell therapy before exploring the current uses of AI to address these challenges. In the third part and final section, it then outlines strategic considerations for addressing key patent eligibility issues likely to arise when patenting AI-enabled stem cell inventions. Identifying inventions involving a combination of AI and stem cell technology and ensuring appropriate patent protection of these inventions is crucial for safeguarding the advances.

Promises and Challenges of Stem Cell Therapy

The stem cell field has made significant advancements, such as discovering how to induce any cell into a stem cell and how to use stem cells for augmenting muscle regeneration, rebuilding heart tissue, developing brain models (organoids) for studying mental disorders, producing islets for treating diabetes, and many other promising developments as discussed in a recent Nature Cell Biology viewpoint article. Moreover, TechSci Research reported that the market for stem cell therapy is expected to grow an impressive 12.39% annually through 2028.

While the stem cell field is rapidly growing, it also faces significant hurdles that need to be overcome. Stem cell therapies require rigorous testing and regulatory approval because stem cells can behave unpredictably during manufacturing and differentiation to functional cell types as discussed in detail in a recent Cell Stem Cell article.

That being said, AI is rapidly becoming central for innovative solutions to remove these obstacles facing stem cell therapy.

Using AI to Aid Stem Cell Therapy

Private companies, academic organizations, and government institutions now incorporate AI to find solutions for using stem cells to treat diseases. We discuss below four major areas where AI has already produced promising results.

AI-Enabled Models Predicting Stem Cell Behavior

The Chan Zuckerberg Initiative is building a "virtual cell" to simulate the key features and behavior of any cell type in the human body. The hope is that scientists can use this "virtual cell" to predict how different cells, including stem cells, will respond to external stimuli. This initiative will use the vast knowledge consolidated by global consortiums such as the Whole-Organism Cell Atlas to build a reference map of every cell type in the body. Training AI on such extensive knowledge may provide a system that can accurately predict how cells are generated from stem cells, how they will interact in the human body, how diseases cause loss of cells or cellular functions, and how to use stem cells for curing such diseases.

Dr. Carl Simon and co-workers at the National Institute of Standards and Technology and the National Eye Institute reported a powerful illustration of how AI can be used to predict cellular behavior and improve stem cell differentiation. This study demonstrated that an AI system can predict the differentiation of stem cells towards eye cells, which was used to significantly improve the quality of this process.

New Tools for Identifying Cells or Cellular States

AI is also providing new tools for identifying cells or cellular states. Dr. Buggenthin and co-workers developed an AI-enabled microscopy imaging system to identify stem cells that initiate differentiation without relying on molecular markers and demonstrated that "lineage choice can be detected up to three generations before conventional molecular markers are observable." Dr. Yang and co-workers demonstrated that such AI and microscopy systems can significantly reduce the variability of stem cell differentiation.

Simultaneously, Dr. Hirose and co-workers developed an AI system called DeepACT that can identify healthy and productive skin stem cells with the same accuracy as humans. This system allows the selection and enrichment of clinically valuable cells.

AI-enabled identification of cells and prediction of their behavior in complex culture systems will help make stem cell differentiation more predictable and improve the production of clinically relevant cells with greater accuracy and efficiency.

Automated Manufacturing of Stem Cells

AI can also provide sophisticated systems for large-scale manufacturing of pure stem cell populations, which is essential for successful stem cell therapy. For example, Nabiha Saklayen (founder of Cellino Biotech) developed a system combining stem cell biology, machine learning, and laser physics to automate stem cell manufacturing. Laser editing allowed both removing unwanted cells and delivering cargo to individual cells. The automated AI-enabled manufacturing of high-quality cell batches may be critical for providing the large numbers of patient-specific cells with low batch-to-batch variability needed for clinical trials and commercialization of stem cell therapy.

Advancing Organ-on-a-Chip Technology

Researchers have previously used stem cells to produce organoids-on-a-chip that offer better clinical predictability of drug properties than conventional cell culture systems, as a study by AstraZeneca demonstrated. Organoids-on-a-chip (sometimes called patients-on-a-chip) are three-dimensional mini-organs that have structural and functional characteristics of native organs. Moreover, Hesperos developed a model system containing several interconnected organs (called human-on-a-chip), allowing an even more sophisticated evaluation of drugs on a system of organs.

While the organoids-on-a-chip technology provides a powerful tool for drug screening and validation, this technology is difficult to scale because of variability from organoid to organoid, and it is labor-intensive to make them work on a large scale due to the complexity of these systems. Dr. Bai and co-workers reviewed multiple ways of addressing this issue using AI systems. In particular, AI can aid with bio-printing of organoids to facilitate the automation of manufacturing organoids-on-a-chip with reduced variability, as reported by Dr. Lee.

Accordingly, combining organ-on-a-chip technology with AI will significantly boost stem cell-based drug discovery and validation by providing improved methods for producing high-quality organoids.

Strategies for Patent Protection of AI-Enabled Stem Cell Technology and Satisfying the Patent Eligibility Requirement Under 35 USC § 101

The increased use of AI in stem cell research and clinical development of stem cell therapies will generate many opportunities for innovation and growth, as shown above. Intellectual property rights will be particularly important for driving innovation in this area, as advances will require the collaboration of various companies, academic organizations, and government institutions, which all have different expertise and available resources. Collaboration among all these different actors often results in intellectual property concerns, in particular patent owner rights to inventions existing before the collaboration as well as inventions arising from the collaboration.

Stakeholders should carefully monitor inventions from combining AI and stem cells and look for patent-eligible concepts at every stage to avoid issues that would frustrate commercial interests. In this regard, it is important to consider 35 USC § 101 of the Patent Act because it determines what categories of inventions are patent eligible if the other requirements are met and can be a crucial hurdle to overcome for many types of inventions.

Patent Eligibility Requirement Under 35 USC § 101

Section 101 of the Patent Act states that processes, machines, manufactures, and compositions of matter, as well as improvements thereof, are eligible for patent protection. Interpreting this statute, the Supreme Court ruled in Diamond v. Chakrabarty, 447 U.S. 303 (1980) that man-made living organisms are eligible for patent protection, but also affirmed that no patents should be granted on laws of nature, natural phenomena, and abstract ideas. Later, the Supreme Court decisions in Alice Corp. Pty. Ltd. v. CLS Bank Int'l, 573 US 208 (2014) and Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 US 66 (2012) set forth a two-part test for determining patent eligibility that can pose a significant obstacle to patenting certain software and computer-implemented systems like AI. The Mayo/Alice test first determines if the claims are directed to a judicial exception, such as an abstract idea, a law of nature, or a natural phenomenon. If the claims are directed to a judicial exception, then the second step determines whether the claim recites additional elements that amount to significantly more than the judicial exception.

While living organisms and biological molecules may be patent eligible if they are man-made, the Supreme Court in Ass’n for Molecular Pathology v. Myriad Genetics, Inc., 569 US 576 (2013) determined that isolated DNA is a naturally occurring product and patent-ineligible subject matter under 35 USC § 101. Moreover, In re Roslin Institute (Edinburgh), 750 F.3d 1333 (Fed. Cir. 2014) it was concluded that the claims were directed to a cloned farm animal such as a sheep, and therefore not “markedly different” from a naturally occurring animal. Patent eligibility of living organisms and biological molecules will therefore require showing that the invention is markedly different from its naturally occurring counterpart.

Strategies for Overcoming the Patent Eligibility Requirement Under 35 USC § 101

While satisfying the patent eligibility requirement is determined on a case-by-case basis and will be very fact-dependent, the court decisions addressing these issues provide some general strategies for overcoming patent eligibility issues. Below are three approaches that may be particularly relevant when seeking to protect AI-enabled stem cell technology.

a) Does the claimed invention improve or solve a problem facing stem cell therapy?

Even if the Mayo/Alice test has created difficulties in obtaining software patents on some computer-implemented technology, an invention deemed directed to patent-ineligible subject matter has nevertheless been found patent-eligible if the claimed invention provides technological improvement. See, e.g., Enfish, L.L.C. v. Microsoft Corp., 822 F.3d 1327, 1336-37, (Fed. Cir. 2016) where the patentee successfully argued that its claimed self-referential table for a computer database was an improvement to an existing technology and thus not directed to an abstract idea. In CardioNet, L.L.C. v. InfoBionic, Inc., 955 F.3d 1358 (Fed. Cir. 2020), the Federal Circuit found the claimed cardiac monitoring system patent eligible because it was not merely an abstract idea of distinguishing between atrial fibrillation and flutter caused by irregular heartbeats, but achieved improvement over conventional monitoring devices. Demonstrating technological improvement was also successfully used to satisfy the patent eligibility requirement in, for example, Visual Memory LLC v. NVIDIA Corp., 867 F.3d 1253 (Fed. Cir. 2017), Amdocs (Israel) Ltd. v. Openet Telecom, Inc., 841 F.3d 1288 (Fed. Cir. 2016), and Rapid Litig. Mgmt. v. CellzDirect, Inc., 827 F.3d 1042 (Fed. Cir. 2016).

Therefore, AI systems that significantly improve stem cell technology could overcome patent eligibility issues by arguing for technological improvement over conventional methods or devices. Practitioners drafting applications to cover AI inventions in the stem cell field that may raise a patent eligibility issue should describe how the claimed inventions improved or solved a problem facing stem cell therapy to provide evidence of technological improvement.

b) Do the claims covering a stem cell recite features markedly different from naturally occurring counterpart cells, and can AI help identify such features?

The patent eligibility requirement under 35 USC § 101 also causes problems for claims directed to stem cells if the claimed cells are not shown to be different from the in vivo counterpart cells. See Ass’n for Molecular Pathology v. Myriad Genetics, Inc. and In re Roslin Institute (Edinburgh).

Patenting stem cells therefore relies on whether it can be shown that the claimed cells are markedly different from naturally occurring counterparts. For example, in Ex Parte Ho, Appeal No. 2016-007472 (PTAB, 2018), the patentee successfully argued that the specific stem cell culture conditions resulted in structural differences between in vivo and in vitro mesenchymal stem cells (MSCs) , and the claims recited MSCs with higher expression levels of specific markers compared to the naturally occurring counterpart cells. While Ex Parte Ho is an appeal decision at PTAB and therefore not binding case law, this decision highlights the importance of identifying features or behaviors of stem cells in culture that could distinguish them from their naturally occurring counterparts.

The rapid advances in using AI to identify and characterize stem cells may result in the discovery of features of stem cells in culture that are not present in vivo, and may provide new options for satisfying the patent eligibility requirement for stem cells.

c) Identifying combinations of steps that result in non-conventional methods even though the individual steps are well-known.

The Federal Circuit Decision Rapid Litig. Mgmt. v. CellzDirect, Inc., 827 F.3d 1042 (Fed. Cir. 2016) found a claimed combination of freezing and thawing steps patent eligible. While the individual steps of the claims were well-known, the court found that the combination of steps was "far from routine and conventional" and provided technological improvement of cell viability.

In view of the CellzDirect, Inc. decision, it is important to consider that integrating AI systems and stem cell technology may result in patent-eligible processes even though the individual steps of the newly developed methods are well-known.

Conclusion

AI has already produced promising results for the development of stem cell technology, as discussed above. Still, the full-scale deployment of these therapies will require further technological development, which can be boosted by the availability of high-quality data for training AI systems and the use of these AI-enabled technologies to produce biological insights into the properties and behavior of stem cells. Further, identification of patent eligible subject matter arising from the convergence of AI and stem cell technology is vital for safeguarding progress and facilitating collaboration among the various actors involved.