This article was originally published by Law 360 and is republished with permission.
The information age continues to bring new uses of data, and even with the increase in computing power, the data becomes more expansive, integrated in our daily routines, and able to transform our lives.
With the leadership of the U.S. Patent and Trademark Office in transition and the law of patent eligibility constantly evolving, it can be difficult for companies to predict whether their inventions are currently patentable, what will be patentable in two or three years when their patent applications will be examined, and what will be enforceable in five to 10 years.
The previous director at the USPTO, Andrei Iancu, called for patent eligibility reform in his farewell speech, and it is unclear how Kathi Vidal, the newly confirmed USPTO director, will act to affect eligibility guidance.
Further, the U.S. Supreme Court recently asked for a recommendation from the solicitor general as to whether to grant certiorari in the patent eligibility case American Axle & Manufacturing Inc. v. Neapco Holdings LLC.
Regardless of the outcome of any of these changes and events, drafters and inventors can take steps to future-proof their patent applications.
One area that may be most affected by events and changes to patent eligibility law is big data processing. As companies integrate their systems into the online world, the companies continue to receive and process more and more data. When faced with this vast amount of data, companies look for ways to monetize this data with new products and services.
But the amount of data presents a challenge to easily analyzing and sorting this data. Solutions to these big data problems often involve a significant amount of research and development resources.
The various approaches to processing this data, whether in health care, fintech or some other industry, may be a patentable innovation. This article discusses protecting innovations for various approaches and solutions to big data.
Among the various definitions, big data generally involves processing tens to hundreds of thousands of data points and producing meaningful results from the data. These large datasets may be referred to as big data because traditional software cannot handle such complexity.
Examples of big data include sales data — e.g., sales data for a supermarket or for an online store — electronic health care records, bank and credit card company financial records, and any other context that involves a large number of users.
Big data is also likely to play a role in the upcoming metaverse and may include customer data, virtual environment data and transaction data, and the data may be structured and unstructured. The ability to efficiently process this data can give a meaningful advantage.
Although big data is involved in a large umbrella of different contexts and technologies, each company often has its own technological hurdles in handling and processing the data they gather.
For example, a credit card company may desire to organize data by user when every record of data it collects is associated with a different anonymous transaction identifier. Handling this process can be computationally complex, and not all companies may have the processing power to handle such complex computations. In the information age, companies will continue collecting data and seeking new ways to use it.
The patent laws include a statute direct to patent eligibility, and this particular statute is often a significant hurdle for obtaining patents from the USPTO or enforcing patents in court. Patent eligibility essentially requires that the claimed invention must not be directed to an abstract idea unless it includes other limitations that amount to significantly more than that abstract idea. A few layman approaches to understanding this complex body of law to consider include asking:
The framework used by the USPTO is not particularly favorable to examining big data innovations. Some USPTO examiners may not appreciate the big aspect to the data, whereby the solution can process a large number of data points that could not be processed by a typical computer in a reasonable time.
Instead, the examiners may suggest that if the dataset was quite small, then such a process could actually be performed on a typical computer or even in a human's mind. Such a perspective disregards the concept of big data that, as described above, is too large for traditional software.
In some cases, the claims of the patent application may be written to address the size of the dataset, if that approach is desirable. In other instances, the claims may be so broad that the examiner's approach is appropriate.
In attempting to overcome this patent eligibility hurdle during examination for big data innovations, shrewd applicants should take steps when drafting applications to avoid giving examiners an opportunity to make these analogies.
When attempting to patent solutions to big data problems, it may be beneficial to draft the patent applications with a problem-solution idea in mind. In other words, the patent application should describe a particular technical problem involving this data and an inventive technical solution, which should be reflected in the claims.
The solution may be an improvement over manual approaches, traditional software approaches, or it may even be an improved big data solution. The examiner may be more likely to appreciate the solution when the application describes this problem-solution rather than a generic recitation of a computer that can process this data. Examiners often look for the details that identify the problem and describe how the technology solves that problem.
Companies often use technologies such as blockchain, artificial intelligence, cloud computing and quantum computing to solve their big data problems. Approaches for patenting these big data solutions are described below.
Patenting Blockchain-Based Solutions
When attempting to patent using blockchain to solve a big data problem, such as using non-fungible tokens to improve the accuracy and reliability of stored data for a supply chain, it may be useful to include the details about how blockchain may solve the problem.
If merely reciting the use of blockchain technology as data storage mechanism, an examiner may react that the application does not have the specificity that is necessary to obtain a patent. As a proactive measure to overcome this issue, a patent application may benefit from a description as to:
These details may be helpful to overcome any concerns that the invention merely implements conventional data storage or overcome reservations by examiners regarding the functionality of blockchain technology.
Further, when patenting solutions to big data that involve blockchain, the drafter should explain the technical problem involving the processing of the data. For instance, if the solution involves a unique method of distributing data on blocks of a blockchain, the application may benefit from an explanation as to why this method is more technically efficient than storing the data in a single block.
In another example, if a solution involves a novel way of communicating between the nodes that maintain the blockchain, the application may benefit from a description of the communication between nodes. Examiners often look for these implementation details when determining if a patent is eligible.
Patenting Artificial Intelligence-Based Solutions
Artificialintelligence may be used to generate conclusions or predictions from big data. When attempting to protect an artificial intelligence approach to a big data problem, such as one that improves the processing speed, it may be useful to focus on how artificial intelligence is used or is improved upon to do so.
For example, it may be insufficient to merely say that data is input into a black box algorithm that outputs a result. The solution is more likely to be patentable if it focuses on:
Emphasis on these aspects are more likely to survive a patent eligibility challenge at the USPTO and in the courts. A few approaches to protecting artificial intelligence innovations have been successful. For instance, a novel way of training a machine learning model is likely to be patent-eligible because training a machine learning model is not an abstract idea in and of itself, and thus any improvement to such training is not an improvement to an abstract idea. Using the output of a machine learning model to change operation of a real-world device is likely patent-eligible because the machine learning model is integrated into changing the state of a real world device. An improvement to the structure of a machine learning model is also likely patent-eligible because it is an improvement to a machine learning model, which is not an improvement to an abstract idea. Each of these examples of patent-eligible concepts are not directed only to using artificial technology to solve a problem, but rather an improvement to artificial intelligence systems that enables the artificial intelligence systems to better solve the problem. But a new solution may still be patentable if it improves upon a conventional manual process as well.
When applied to big data problems, these types of solutions are often patent-eligible, especially when the patent application fleshes out the story about why these solutions are better than prior attempts to solve the big data problem.
Patenting Cloud-Based Solutions
Cloud computing may be employed to solve a big data problem, such as improving data storage and the speed of processing each data point. It may be useful to focus on how cloud computing is used differently in the big data context than in other contexts.
The mere use of off-site dedicated servers to perform big data processing may not be sufficient for patentability. So a patent application should explain how the process differs from computing that is done on local servers as well as from other conventional cloud computing applications.
The processing of data may change when utilizing a cloud computing solution, and this processing may be the key to patentability.
For instance, if data is being processed locally, a server may follow the rather simple steps of receiving data, processing the data according to a defined set of rules, and generating an output. However, if the processing is performed on a remote cloud server, problems may arise that are specific to cloud computing.
For example, one aspect of processing big data is that the data generally is generated over time and is not stagnant. This can cause local servers to send fragments of data to a remote server over multiple signals instead of in a single data packet or message.
The cloud server's handling of the data in these instances to produce meaningful results or other variations of data transmission may have potential patentability.
Patenting Quantum Computing-Based Solutions
A new and burgeoning field for patenting solutions to big data problems is using quantum computing. Quantum computing is relatively new compared to the solutions above for addressing big data problems, but quantum computing still follows the same rules for patent eligibility. For example, an application that merely mentions using quantum computing to process data may not be patentable or result in a high-quality patent. However, a patent application that is directed to a quantum computing system and more adequately describes how a quantum bit or qubit is used in a solution to big data is more likely to be patentable.
Mere description of a computer system used to implement an idea may not be enough. These solutions are more likely to withstand a challenge over time when the solution involves a new configuration for quantum computing that improves the processing of big data.
Because quantum computing is still in ramping up to become a reliable and widely adopted technology, it is likely that examiners will consider inventions relying on quantum computing to solve a problem to be eligible subject matter. It is still recommended for the patent application to describe the specifics of the underlying processes of how quantum computing is key to solving the particular problem.
The desire to monetize and utilize big data is not going away. In fact, the issues with big data are only going to become more pronounced as society moves into a more digital world and more and more data continues to be collected.
To obtain protection and further incentivize innovation regarding processing big data, it has become critical to draft patent applications to adequately describe the inner workings of the technology for which protection is being sought.
In addition, companies can take the following steps to help ensure they are adequately capturing innovations for their research and development of big data solutions.
First, maintain a log of any problems that occur when attempting to come up with a better method of processing big data and any solutions to those problems. These solutions may be patentable, and, while they may be narrow solutions, any company that attempts to use the same technology may encounter the same or similar problems;
Second, prepare detailed disclosures describing the low-level details of the system and how it operates. Patent applications should enable an individual to program the solution, so providing these details will avoid guesswork in how the system operates and will likely result in a higher-quality application.
Finally, carefully review any patent applications to ensure inclusion of the fine-grained detail — but not trade secrets — that helps overcome the patent eligibility bar.
The desire to allow a patent application to be filed without adequate review may save time now, but can ultimately lead to increased costs during examination and inadequate protection.