As an angel investor and patent attorney, I come across an increasing number of digital health startups that apply AI to solve a host of healthcare needs. This is unsurprising as AI makes people’s lives more efficient, powering many programs and services that help them do everyday things, such as connecting with friends and using an email program or ride share service. The best examples of AI in daily life are travel navigation, smart home devices, smartphones, drones and smart cars. Healthcare simply is another use of AI to help humans. I will discuss the interactions of the patent system and FDA regulatory system and how a digital healthcare company can navigate both systems to arrive at a doubly protected market place with patents and FDA approval.
Potentials for AI and digital healthcare
AI and digital healthcare applications have the potential to revolutionize the healthcare industry by improving the accuracy and efficiency of medical diagnoses, treatments, and procedures. Some examples of how AI and digital healthcare applications are being used include:
- Medical imaging: AI algorithms can be used to analyze medical images such as X-rays, CT scans, and MRI’s, to help detect and diagnose conditions such as cancer and heart disease.
- Electronic health records: AI can be used to analyze electronic health records (EHRs) to identify potential health risks, track the progression of diseases, and develop personalized treatment plans.
- Clinical decision support: AI can be used to provide doctors and other healthcare professionals with real-time guidance during patient care, helping them make more informed decisions.
- Remote monitoring: AI-powered digital healthcare applications can be used to remotely monitor patients with chronic conditions, such as diabetes, heart disease, and COPD, to help manage their symptoms and prevent complications.
- Robotics and automation: AI-powered robots and automation can be used in surgeries and other medical procedures to improve precision and accuracy, and reduce the risk of complications.
- Drug discovery: AI can be used to analyze large amounts of data in order to identify new drug targets, predict the efficacy of new compounds, and accelerate the drug discovery process.
- Wearables and mobile health: AI-powered wearables and mobile health applications can be used to track vital signs, monitor physical activity, and provide reminders for medication and other healthcare tasks.
- Virtual health assistants: AI-powered virtual health assistants can help patients with scheduling appointments, managing medication, and answering healthcare-related questions.
Google recently announced that it will be partnering with iCAD, an organization that specializes in cancer detection, to bring its AI-enabled mammography program into the world for breast cancer detection. iCAD, through the licensing partnership with Google, will integrate Google’s AI platform and its suite of AI options for breast imaging. This could improve the company’s ability to screen for breast carcinoma earlier in the process.
Artificial intelligence has been a key component of health care in recent years. It promises to solve the greatest problem, finding the disease. These digital healthcare improvements have far-reaching consequences. Patients could have faster access to the health care system. It could also mean fewer medical professionals need to be involved, which could reduce health care spending.
Machine learning (ML) and artificial intelligence (AI), both based on machine learning, have the potential to transform healthcare by generating new and valuable insights from the large amount of data generated. Examples of high-value applications are earlier diseases identification of new patterns or observations on the human body, improved diagnosis and detection physiology and the development of personalized diagnostics.
One of the greatest advantages is AI/ML is a software feature that allows it to learn from real-world experience and application. It can be used to enhance its performance, and offers the ability for AI/ML software (training) to learn from real-world feedback. These technologies are unique among other software because they can improve the performance of software (adaptation).
As a medical device (SaMD1) and as a rapidly growing area of research-and-development. With the right regulatory oversight, AI/ML-based SaMD can be safe and effective with software functionality that enhances the quality of patient care.
AI promises great things. Many people don’t realize they have health problems until it is too late. The screening process is a major reason. Specialists are rare and hard to find, so booking an appointment just to have a screening done is not always the best use of either their time or that of the patient.
What if AI-enabled software could be installed at every doctor’s office and pharmacy for quick breast cancer screenings? Imagine if similar technology was available to detect diabetes. Or cavities. Imagine if it were easier to get a doctor’s appointment because there were fewer people coming in for checkups. Imagine doctors being able to see only those patients who need treatment.
The technology is still very young. The Food and Drug Administration struggles to regulate the data sets that these platforms are trained from. It is clear that no health technology can succeed without the support of doctors who are concerned about black box technology. Although AI in diagnosis is a well-known concept, startups that use it to diagnose have only recently received serious funding. Between 2020 and 2021 funding reached the billions for only the second time in over a decade. Crunchbase data shows that funding for 2022 was more than $883 million.
Twin Pillars of Digital Healthcare Exclusivity
Investors often have a favorable view of medical device companies due to the exclusivity provided by FDA approval and patent approval.
- FDA approval: Medical devices must go through a rigorous regulatory process before they can be marketed and sold in the United States. This process, called premarket clearance or approval (PMA) or 510(k) clearance, is designed to ensure that the device is safe and effective for its intended use. The exclusivity provided by FDA approval can give investors confidence that the device has a clear path to market and that the company has a competitive advantage over other companies.
- Patent protection: Medical devices are often protected by patents, which give the patent holder the exclusive right to make, use, and sell the invention for a certain period of time. Patent protection can provide medical device companies with a competitive advantage by preventing other companies from copying the device and can also provide a revenue stream through licensing or selling the patent to other companies.
- Potential for High returns: Medical device companies have the potential to generate high returns for investors due to the high costs of research and development and the significant barriers to entry in the market. As a result, these companies can charge premium prices for their products, which can lead to strong revenue growth and high profit margins.
- Growing demand: The aging population, increasing incidence of chronic diseases, and the need for innovative and cost-effective healthcare solutions are driving the demand for medical devices. This growing demand for medical devices provides a tailwind for the industry and can provide investors with a stable source of returns.
All of these factors can make medical device companies an attractive investment opportunity for investors, however, it’s important to consider the risks associated with these companies as well, such as regulatory changes, intense competition, and the cost of research and development.
Software and FDA Regulatory Issues
The FDA has approved or cleared several AI/ML-based SaMDs to date. These have typically only included algorithms that were “locked7” prior marketing. Any algorithm changes will likely require FDA premarket approval for any changes beyond the initial market authorization. Some algorithms, however, can be adapted over time. These AI/ML-based SaMD have the power to learn continuously.
The adaptation or modification to the algorithm can be realized once the SaMD has been distributed and has “learned” from actual-world experience. These types of adaptive AI/ML algorithms that are continuously learning may produce a different output than the one originally cleared. This is because the algorithm can adapt to changing inputs and learn from real-world experience. These tools are highly iterative, autonomous and adaptive. A new total product lifecycle (TPLC), regulatory approach is required to facilitate a rapid product improvement cycle where the digital healthcare devices can be improved continuously while still providing effective protections.
The FDA will review premarket submissions to evaluate the safety and effectiveness device software functions, including software in a medical device (SiMD) and software as a medical device (SaMD). The FDA uses a risk-based approach in determining when a premarket submission is required. The guidance on software modifications under 510(k) process focuses on the risks to patients and users resulting from software changes. Premarket submissions may be required for certain software modifications that could result in patient harm such as:
- Changes to risk controls to avoid significant harm.
- Any change that has a significant impact on the clinical functionality or performance specifications device.
The above approach is applicable to AI/ML-based SaMD. It requires a premarket submission from the FDA if the AI/ML software modifications significantly affect device performance or safety and effectiveness6; if the modification is to the device’s intended use; or a major modification to the SaMD algorithm.
A supplemental application is required for a PMA-approved SaMD where changes that could affect safety or effectiveness are required, such as new indications, new clinical effects or significant technology modifications that impact performance characteristics.
After you file premarket submissions for FDA’s evaluation of the safety and effectiveness of device software functions that meet the definition of a device under section 201(h) of the Federal Food, Drug, and Cosmetic Act (FD&C Act).1 The FDA also provides its Guidance for the Content of Premarket Submissions for Software Contained in Medical Devices which provides the FDA’s thinking related to the documentation FDA recommends your company include for the review of device software functions in premarket submissions.
Healthcare AI and patents
We’ve seen more innovation in medical device design over the past few years, especially with the rise of mobile medical gadgets, a trend that was further accelerated by the COVID pandemic. Medical device designers and manufacturers are in a strong position to create innovative products that improve patient care, diagnostics, treatment and save lives. Protecting your invention is essential as you innovate to keep pace with this growing trend and better serve patients, doctors and other healthcare professionals. A medical device patent is the answer.
Patents for medical devices prevent others from using, making, or importing your invention. Patents give you an advantage on the market and allow you to reap the financial benefits of your new, innovative medical device. The world of patents can be confusing and complex. We are here to help you understand the process and provide guidance.
Types and applications of patents for medical devices include:
- Utility patent. Also known as a nonprovisional patent. The utility patent primarily focuses upon how the device works. The Utility patent accounts for 90% of all patents issued by U.S. governments. It covers invention of a useful product, process or technology.
- Provisional Patent: Patent protection will be granted to the first person to file. A provisional patent allows you to hold your spot in the line until you are ready to file for a utility patent. You have one year to file for a provisional medical device patent. Your provisional medical device patent will expire after the year.
- Design patent: This type of patent covers the design of your medical device’s exterior. A design patent covers an original, novel, or ornamental design for a medical device. This could include the device’s shape, user interface, and even its touchscreen design.
AI Patenting and Subject-Matter Eligibility Issues
Software healthcare patents tend not to run into patent eligibility issues raised in the Supreme Court’s Alice decision which caused rejections in numerous software patent applications. The Alice test is a legal test used in the United States to determine whether a software patent is valid. The test takes its name from the 2014 U.S. Supreme Court case Alice Corp. v. CLS Bank International.
The Alice test is used to determine whether a software patent claim is directed to a patent-eligible abstract idea, and if so, whether there is an “inventive concept” that transforms the abstract idea into a patent-eligible invention. A claim that is directed to a patent-eligible abstract idea but lacks an inventive concept is considered to be patent ineligible under 35 U.S.C. § 101.
In the context of digital healthcare, there have been many patent applications filed for software and systems related to healthcare. Some of these patents have been challenged as being directed to abstract ideas, and thus not patent eligible.
In general, to defeat Alice, you should document how the AI software improve the functioning of the computer itself. For example, “[C]laims ‘purporting to improve the functioning of the computer itself,’ or ‘improving an existing technological process’ might not succumb to the abstract idea exception.” Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1335 (Fed. Cir. 2016). The claims themselves must recite “how the purported invention improved the functionality of a computer.” The claimed “improvement” cannot be “result-oriented generality” that amounts to “a mere implementation of an abstract idea on a computer, not the specific way to improve the functionality of a computer.” See Koninklijke KPN N.V. v. Gemalto M2M GmbH, 942 F.3d 1143, 1152 (Fed. Cir. 2019). For example, in the Gracenote, Inc. v. Free Stream Media Corp., 2019 U.S. Dist. LEXIS 190397 (D. Del. 2019), the patent claim at issue was claim 11, and this claim was upheld by the court:
11. A method comprising:
playing back multimedia content on a multimedia playback device, including providing at least some of the multimedia content on a display associated with the multimedia playback device;
during the playback of the multimedia content by the multimedia playback device, repeatedly deriving, by the multimedia playback device, fingerprints from respective segments of the multimedia content;
comparing the derived fingerprints to reference fingerprints representing features of the multimedia content, each reference fingerprint associated with one or more actions;
determining that one of the derived fingerprints matches one of the reference fingerprints; and
in response to the determining that the one of the derived fingerprints matches the one of the reference fingerprints, causing execution of an action associated with the one of the reference fingerprints, the action being associated with a time point indicating when, in the multimedia content, the action is to be performed.
The court upheld the patent noting that it discussed specific improvement over the prior art: the claimed invention “enable[s] detection of trigger actions without modifying the multimedia signal” to avoid the disadvantage of watermarking, which “necessarily changes the video/audio.” The invention also noted that it improved the functioning of the device: “the fingerprint matching ensures that the trigger actions appear at the correct corresponding moment in the broadcast since the invention is time-independent but content-dependent.” The court also pointed to unconventional use of existing technology with “known fingerprints in an unconventional manner to improve the accuracy of a trigger’s position within a multimedia stream.”
It’s important to note that this test is not the only one to check if a patent is valid and other factors are also taken into account, such as novelty, usefulness, and non-obviousness.
If you have specific patent application in mind, it would be best to consult a patent attorney who can give you a more specific analysis on patentability issues of that particular application.