With the ever-growing adaptation of software in all realms of health care, the market for software for medical devices (SaMD) is forecasted to grow 16.7% per year over the next decade and surpass $5 billion by 2032. Innovations in biomedical applications are increasingly relying on software, which are often developed by software engineers in close concert with clinicians at medical institutions and biomedical researchers. Springing from the synergy produced by this unique collaboration, these technologies have the potential to vastly enhance the quality of care provided by doctors and hospital staff, and improve outcomes, and benefit the health and lives of patients.

3 SaMD Categories to Know

Software for biomedical applications is largely grouped in three categories. The first is software as a medical device in which the software itself is used for a medical purpose without being tied to a medical device hardware. One example of SaMDs are digital therapeutics applications used to prevent, diagnose, manage, or treat a patient’s physical ailment, mental disorder, or behavioral condition. 

The second category is software in a medical device (SiMD), in which software is integral to the functioning of a medical device hardware. SiMDs often tie specialized medical hardware (e.g., an insulin pump, an imaging device, blood pressure monitor, or drug delivery mechanism) with a non-medical computing device (e.g., a smartphone, tablet, wearable device, laptop, or desktop). An example of a SiMD is diagnostic imaging from a computed tomography (CT) scan used to detect the presence of tumorous growth in an organ.

The last category is software used in the manufacture or maintenance of a medical device. This includes software tied to administration of medical devices at a hospital or institution. An example of software in this category are platforms used to manage, process, and evaluate patient data.

IP Strategies Used to Protect Biomedical Software Innovations

Sophisticated intellectual property (IP) strategies can be used to protect innovations involving software for biomedical applications. Such strategies include clinical trial collaboration and sponsored research agreements, as well as filing patent applications or pursuing other forms of protection, such as trade secrets. Patents can provide a competitive advantage in the field by granting a right to exclude others from making, using, or selling the invention. Trade secrets may be used to protect its owners against someone (e.g., a former employee) misappropriating, misusing or improperly disclosing the secret, but does not protect against a competitor that independently develops it. 

The Patent Office grants patents to innovations that overcome certain legal and technical hurdles, but there are also strategic considerations in pursuing worthwhile patents. For example, an innovator may consider whether the software-based invention is detectable from the software itself or other publicly available sources. For instance, an innovative algorithm for processing data structures running on a backend server might be difficult to detect, thus weighing against seeking patent protection. In contrast, a novel dashboard interface to visualize patient data for clinicians may be easier to detect, thereby weighing in favor of seeking patent protection.

In seeking patent protection, existing relationships and licensing agreements with partners should be considered. Software for biomedical applications are often developed by computer programmers at a software company, in close cooperation with clinicians at hospital institutions or pharmaceutical companies. Within this context, IP ownership between the parties should be documented. Even if the invention might be difficult to detect by an outside entity, it may beneficial for either party to seek patent protection, given such collaborations. The filing of patents over various IP assets should be done in a manner consistent with the agreement as well as in furtherance of the owner’s interests. 

Once the decision is made to seek patent coverage, a patent application must overcome the myriad of challenges to satisfy requirements of patentability. Of particular concern with patenting software-based inventions is subject matter eligibility, which essentially prevents patents directed to an abstract idea, a mere automation of a human process or even a process that can be performed by a human. For software-based inventions, this requirement can be satisfying by framing the invention in terms of a technical problem and a technical solution – not a business problem or solution – from the perspective of the computer. The patent application itself should explain how particular components or individual steps improve the functionality of a computer, in comparison to previous techniques. 

Furthermore, the patent application should describe the invention so that the boundaries of the invention (written support) are clear and the invention can be practiced (enablement). Special care should be taken when the invention is related to a method of treatment using the software (i.e., SaMD) as opposed to just the algorithm, because the patent application should explain both the invention and its purported therapeutic effects in detail. These explanations may include: (1) prophetic examples describing future and anticipated results of an experiment or (2) working examples describing the real results of an actually conducted experiment (e.g., clinical trial data). (While outside the scope of this discussion, those who wish to release products making use of their inventions should also be aware of relevant Federal and Drug Administration  regulations.)

When the invention is related to just the computer algorithm itself, the patent application should sufficiently describe how the computer function is to be achieved. This is opposed to simply declaring the desired end result of the invention. For example, for an artificial intelligence (AI) invention, instead of describing what the model outputs at a high level, it may be desirable to include details of all pertinent aspects, such as: the training dataset, the architecture of the model itself, the precise procedure for training the model (e.g., the inputs and outputs), the pre-processing steps for new data, or the application of the model.

These and other considerations should be factored when formulating IP strategies to protect software-based inventions for biomedical applications.