Charting the Course for Invention at Caltech
From a sea of scientific ingenuity and biomedical possibilities, an invention emerges. Steeped in potential, this new intellectual property can represent non-invasive methods for measuring blood and eye pressure, an innovative use for ultrasound imaging, and more. But as Caltech researchers navigate through prior arts (e.g., previously patented inventions and publicly available products), legal conundrums, and medical quirks, pinpointing an idea both unprecedented and useful can remain an elusive enterprise.
ENGenuity spoke with three current Caltech innovators and postdocs to dive deeper into the invention process, charting the progression of research performed in the lab into a discovery suitable for a patent.
Alessio Tamborini (MS '19, PhD '23, medical engineering) is a current postdoctoral research associate in the Andrew and Peggy Cherng Department of Medical Engineering. Tamborini's work focuses on utilizing fluid-dynamic principles to non-invasively measure pressure related conditions in the eye and the cardiac system, and he is a co-inventor with Mory Gharib, Hans W. Liepmann Professor of Aeronautics and Medical Engineering, and others on multiple related patents. Although Tamborini has multiple research areas within medical engineering, his work as a postdoc at Caltech centers around a contactless imaging system for eye pressure measurement. His graduate work led to an initial patent in this area, as described below.
Gunho Kim (MS '19, PhD '23, mechanical engineering) is a current postdoctoral scholar research associate in the Department of Mechanical and Civil Engineering. Kim's research investigates the mechanical properties and ultrasound characteristics of shape memory polymers, which are materials that can return to their original shape after being stretched or bent. In this research vein, Kim is a co-inventor with Chiara Daraio (G. Bradford Jones Professor of Mechanical Engineering and Applied Physics, Investigator, Heritage Medical Research Institute), and others on a series of patents involving ultrasound-responsive polymers—these are polymers with certain additives that contribute to increased ultrasound absorption capabilities.
Dominic Yurk (BS '17, physics; MS '21, PhD '23, electrical engineering) is a postdoctoral scholar research associate in the Department of Electrical Engineering. Yurk's work deals with multiple non-invasive methods of using ultrasound to measure the cardiovascular system: one method involves vibrating arteries to measure their blood pressure, and the other applies artificial intelligence (AI) to ultrasound scans of the heart to take relevant heart diagnostic measurements. On both methods, Yurk is a co-inventor with Yaser Abu-Mostafa (PhD, '83, electrical engineering, Professor of Electrical Engineering and Computer Science), and others on a series of patents. Yurk's patent involving machine learning for cardiology is his focus during the discussion below.
ENGenuity: How would you describe your patents/inventions and what are their potential impacts?
Tamborini: The work covered in this patent [a contactless imaging system for eye pressure measurement] is about measuring ocular pulsations in the eye using our imaging system to infer pressure and pulsatile flow. The core concept involves leveraging the physiological forces exerted by the cardiovascular system on the eye to extract clinically significant information about the eye. From here, we can reconstruct patient-specific eye models from which we infer different aspects and conditions of the subject's eye.
Kim: We developed a new shape memory polymer that has high sensitivity to ultrasound, allowing it to absorb ultrasound much better than the surrounding tissues. In a medical sense, this means we can remotely target and change the shape of the new material using ultrasound without causing thermal damage to any surrounding areas or tissue. The first application we thought of using this idea was for drug delivery in the human body, but this kind of ultrasound responsive material could also be used in a microfluidics context.
Yurk: My advisor, Yaser Abu-Mostafa, who is a machine learning expert, and I got involved in a collaboration with UCSF to apply machine learning to ultrasound sound scans of the heart. We use machine learning algorithms to measure heart health parameters and take diagnostics that currently require a cardiologist appointment. For example, at a primary care doctor's office or with a home health nurse, you would be able to use a commercially available portable ultrasound machine and have AI guide you in taking effective scans of the heart. The machine learning algorithm could then provide an interpretation of those scans and throw up potential warning flags.
ENGenuity: What are the main challenges of turning your research ideas into a patent?
Kim: One hurdle is finding a niche. You can't just duplicate what other people have done. You need to find something new and assert that it is useful for a future application or for future research. When I was working on this patent, I found out that there are so many prior arts that initially appeared irrelevant to our findings, yet the examiner utilized them for counterarguments. I think one of the biggest differences between writing scientific papers and working with patents and the patent office was trying to get as much coverage as possible while not overlapping with other inventions.
Tamborini: Working with the human body is an incredibly complicated problem. Unlike in controlled lab settings, where variables can be finely adjusted, the human physiology introduces numerous factors that cannot be separated, making it a very high-dimensional space. Translating science into a patentable invention hinges on our ability to encapsulate these ideas into practical solutions suitable for the real world.
Yurk: For the machine learning heart diagnostics project, the biggest challenge before deciding to file it as a non-provisional patent is getting real human testing in an actual clinical environment. In general, with anything AI-related, it is one thing to build a model and get good performance based on a database built on static things, but the real test of a machine learning model for medical purposes is taking it into the clinic, using it on patients the model has not been trained on, and seeing if the model will actually perform in the real world. Also, the patentability of AI and algorithms is an evolving area in terms of what you can protect. Even if you try to patent your specific version of an open-source algorithm, it can be easy for someone else to come in and do their own slight tweak on it that does effectively the same thing. Because of this, there is more importance around keeping the actual code and some of the specific implementation details a trade secret.
ENGenuity: What surprised you about the patent process?
Yurk: The AI for cardiology project was initiated in the latter part of my PhD, so we were drawing up the patent as I was getting ready to defend. I was advised to keep my defense closed because if it was open, it might constitute a public disclosure. I thought that inviting some of my grad student friends from other departments and my parents would be ok since they weren't going to steal the idea and try to file a competing patent. But I learned that the problem is not that someone might steal it and file their own patent, it is that even if none of them try to steal it, once an idea has been disclosed publicly, it becomes no longer patentable even if no one else tries to file on it.
Tamborini: The most surprising aspect of this patent is the iterative nature of the inventive process. You need to go through multiple iterations, maybe five, six, or seven, before you have something that addresses two questions: is it an improvement and is it usable in the field? As engineers, we can get pretty wrapped up in the technology itself. But what might be groundbreaking in a lab might not be feasible for a doctor to use in a clinic—for example, if it significantly increases procedure times. For an innovation to be valuable in a medical setting, it needs to find a good compromise between engineering, ingenuity, and practicality. In terms of innovation within the medical field, I am also amazed by the high interdisciplinarity of current inventions. Academic institutions like Caltech have placed numerous efforts into providing well-rounded education and collaborations between different fields, and this has greatly favored the understanding and tackling of medical problems. I think these changes will allow for a future where engineers and scientists address more complicated problems.
Kim: I was not familiar with the phrases they use in patents or patent lawyers, so that change of language was also a hurdle for me at first. Additionally, when I am working with scientific papers, I normally compete against previous papers or other research labs. But with a patent, I had to deal with not just previous research and scientific papers, but also other patents. Those patents can encompass many different materials and many different forms that a specific prior art didn't really go into.
ENGenuity: If you could give advice to your past self about the patent process, what would you say?
Tamborini: If I could take a step back and start the process anew, I would focus more on understanding the integration between technology we develop in the lab and medicine in the clinic. It is essential to understand clinical procedures and challenges to ensure the research we perform truly enhances medical outcomes. Meaning, how can I develop a technology that is beneficial for the field and does not completely alter current protocols used in hospitals? As engineers, we need to consider that doctors work in very high-paced and stressful environments. In my research I have primarily focused on diagnostics. In this context, I think that progress can be evaluated by considering both diagnostic accuracy and the overall measurement experience. It is about finding a compromise between the signal quality—the information that you can extract—and how fast, comfortable, or safe this measurement is for the patients and doctors.
Yurk: I would not give advice to change the path that I have taken, but I would reinforce that it is a good idea to take opportunities to pursue cool ideas. In all the patents I've been involved with, innovation started by doing research on an idea that we didn't have a guarantee of success. The blood pressure measurement using ultrasound project (not the machine learning project) started as a whacky physics idea on paper where we had no idea if it would work in squishy human anatomy. It was cool to eventually see it validated, that this idea could work in real arteries and show real effects. Trust in the process of following cool ideas to see where they lead.
Kim: I am not sure I can give advice, but I can provide some insight. OTTCP (Caltech Office of Technology Transfer and Corporate Partnerships) has extensive knowledge and advice, and they are very happy to chat with us if our idea can be realized or formed into a patent. The first thing you need to do when you have a certain idea that can be groundbreaking is to reach out to OTTCP. They know how to protect our idea; they know how to develop that idea into a patent. If you can think of an idea, it may be something that somebody has already thought of, or it may overlap with other prior arts if expressed in certain ways. Caltech has great resources for supporting patents and entrepreneurship.
ENGenuity: What resources have been beneficial to you throughout the patent process?
Yurk: It was a combination of OTTCP, Yaser, and my other mentor and advisor, Dr. Aditya Rajagopal [visiting associate in electrical engineering], who is a senior researcher in Yaser's lab. He has been through the process many more times than I have. Between him, Yaser, and the OTTCP folks, I had a lot of people who were able to advise on the process.
Tamborini: Professor Gharib has been my mentor through the entire process. He has guided me along every step of the way, from identifying the problem to applying our skills and expertise in addressing it, developing prototypes, and eventually building a device to show that it works. Additionally, the entire OTTCP office has been incredibly helpful with the documents, timelines, and formatting of the patent application and process. There are elements of the process I was not familiar with, but are crucial for consolidating and protecting your idea. I have worked closely with John Nagarah in OTTCP from the first disclosure I submitted all the way through the patent process. He has been a critical player.
Kim: Rose Kiser from OTTCP oversaw the inventions from the group of the late professor and Nobel laureate Robert Grubbs [Victor and Elizabeth Atkins Professor of Chemistry]. Grubbs, along with professors Chiara Daraio and Mikhail Shapiro [Max Delbrück Professor of Chemical Engineering and Medical Engineering; Investigator, Howard Hughes Medical Institute], was one of my collaborators and was vital in generating the idea for an ultrasound-responsive material for medical purposes. Rose was the translator between the researchers and the lawyers, and she organized the meetings. Without her, I don't think we could have been able to do this on time.
ENGenuity: What are the next steps for you to bring your patent to life?
Yurk: My work as a postdoc right now is following through on the AI for cardiology idea. We are starting up the clinical trial and moving that forward and then, if we get good clinical trial results, we will turn those results into an actual product that we can put out into the world. The goal of what I've been wanting to do on this research path is to develop products that are used in the medical world and help patients. After the clinical trials, I have a lot to learn about the FDA process of getting something approved for use. That will be a whole different follow-on learning experience of turning something from an idea into a realized product.
Kim: The current patent that we have worked on was about the material itself—how the material can have better ultrasound sensitivity or absorption. But the next patent will be about the concept—how we can apply our previous invention to solve problems in real life. The ongoing collaboration between the Daraio and Shapiro research groups is working on that concept patent.
Tamborini: The pipeline for the next months and years is to continue building our algorithm for the eye measurements on our imaging system to ensure it is universally applicable. We plan on performing internal healthy control validation through studies approved by our institutional review boards (IRBs) here at Caltech. Then, upon a successful implementation of this, we will arrange to transition our device to an ophthalmology clinic with patients. The focus is taking this idea/patent and systematically bringing it along through every testing step from bench to clinic.