The next era of digital therapeutics

Hogan Lovells

Hogan Lovells

Products incorporating artificial intelligence/machine learning (AI/ML) algorithms have enormous potential for health care. The power of AI can enhance the capabilities of software that has the potential to benefit more patients. However, the introduction of AI to the health care landscape raises unique regulatory questions, such as whether and how an AI product should be regulated as a medical device and what sort of data will be necessary to demonstrate that the product performs as intended. Digital therapeutics (DTx), which often incorporate AI algorithms, have incredible promise, but because they are in many ways revolutionary – especially if they are “continually learning” – they have struggled to fit within existing frameworks for delivering and regulating care.

On Thursday, 18 May, Hogan Lovells is hosting its annual Health Care AI Law and Policy Summit, an informative and interactive program where our thought leaders and industry guests will address a variety of topics including new and emerging health care AI policies and regulatory considerations, implications for ethics and consumer safety, developments in the U.S., UK, and EU, and more.


Promise and pitfalls of the current DTx framework

A subset of digital health, DTx are software-driven therapeutic applications that are by themselves the therapy or are coupled with other therapeutic modalities to increase the therapeutic effect. The promise of DTx is the breadth of their possible applications. For example, they can provide individualized therapies for the treatment of conditions as varied as mental health, palliative care, substance use disorders, and chronic conditions such as irritable bowel syndrome (IBS) and diabetes. This merging of technologies, however, can also create tricky issues in their implementation and commercialization. While the COVID-19 pandemic accelerated the interest in these technologies, many challenges to their adoption remain, not the least of which is a clear path to commercialization and payment.

Although the U.S. Food and Drug Administration’s (FDA) approach continues to evolve, most DTx receive their first FDA market authorization (MA) as prescription software-as-a-medical device (SaMD). Obtaining this premarket authorization typically requires the submission of both nonclinical and clinical data to validate the algorithm and demonstrate that the device achieves its claimed intended use, in a manner that ensures that benefits to patients outweigh potential risks. Where the DTx was developed using AI or ML or seeks to incorporate AI itself, FDA requires that the sponsor provide detailed data and information about the source of the data, the methods by which the algorithm was trained, and how the resulting algorithms were validated. The Agency not infrequently engages in detailed inquiries about the source data (on issues including bias, potential confounding factors, demographic representation, and others) and the sponsor’s methods. For that reason, it is advisable to begin discussions with FDA early (through the pre-submission process) in order to obtain input on proposed study plans and related details before completing product validation.

Once they have received FDA authorization at the federal levels, prescription medical devices are subject to complex and varied state licensing requirements that can attach to the activities of manufacturers and their distribution partners, most of which are set up for the distribution of traditional medicinal products. Licensing rules vary by state as well as between prescription and over the counter (OTC) drug or device types, recipient, and facility/entity type. Adding to the complexity, most state licensing paradigms were developed to handle prescription drugs and controlled substances and, consequently, are not always well suited for the distribution models used for prescription medical devices and even less so for DTx. Further still, while distribution models are well developed for distributing tangible things, they are often not well suited for DTx for which there is no physical product, merely software downloads of applications and provision of access codes. While some states do not have any licensure requirements for prescription medical devices, many states do and those requirement apply equally to DTx and may extend from manufacturers to their distributors and third party logistics providers.

Prescription DTx (PDTx) require a health care provider to issue a prescription or medical order before a patient can access the product. Often, companies will rely on partnerships with telehealth providers who can evaluate the patients, and if appropriate, issue a prescription for the product. This raises important considerations around the agreements between these parties, in order to avoid pitfalls with issues such as state regulation of telehealth, requirements for prescriptions provided via telehealth, applicability of state pharmacy and medical device regulations, possible Sunshine disclosure obligations, and also the corporate practice of medicine. While it is possible to set up these arrangements in a way that avoids triggering issues, it requires careful consideration of the various law and issues at play.

As a practical matter, lack of payer coverage for DTx also provides a key barrier to adoption. DTx do not readily fall under any Medicare or other payer benefit category. While some DTx companies seek to fit them into existing drug or device paradigms, others liken them more to durable medical equipment (DME). While this approach may allow new devices to access the market faster, it may also result in lower reimbursement. At the moment, most DTx companies are negotiating with payers and pharmacy benefit managers one at a time and in the meantime bearing much of the costs or adopting a cash pay model in favor of market adoption.

Regulating AI

FDA and other regulators continue to refine their approach to regulatory requirements and also the data burden that is needed for DTx devices for marketing. In late 2022, FDA announced its greatly anticipated Clinical Decision Support (CDS) Software guidance. Notably, this updated approach appears to position more software products and AI tools within the realm of FDA regulatory authority than was previously the case, as software tools which issue a singular output or one that is tied to time-critical decision-making are now considered to generally be subject to active FDA regulation as a medical device. The final guidance also reinforces FDA’s position that CDS tools based on AI algorithms are unlikely to qualify for the non-device category, because it is nearly impossible to enable the user to independently review the inputs, outputs, and methods to get from one to the other for a “black box” algorithm.

To date, FDA has not approved any SaMD with evolving AI, which means that manufacturers need to think carefully about how they plan to evolve their algorithms as new data and real world experience is gathered. Many such iterations under normal circumstances may well require a new pre-market authorization. As one option, manufacturers can consider submitting for FDA review a Predetermined Change Control Plan (PCCP), which allows them to obtain pre-approval for changes to AI devices by pre-defining for FDA how specific types of changes can be adequately validated in-house without the need for further FDA review. This approach has been accepted on a case-by-case basis to date and we have assisted clients in working through the process with the Agency. As we have described here, FDA recently published long-awaited draft guidance outlining its views on PCCPs. We expect the guidance, once finalized, to further encourage innovation and delivery of AI/ML-enabled medical devices by promoting a general approach that enables manufacturers to make certain updates to their devices without re-engaging FDA prior to their implementation.  However, a PCCP requires manufacturers to plan carefully as its approval will be based on particular circumstances and will be limited to specific types of changes that are predefined by the company and agreed to by FDA.

Advocating for value based payments

Additionally, digital health companies are looking to drive policy change in how digital therapeutics and other AI-enabled products are regulated globally, and closer to home, to encourage payment based on driving efficacy as demonstrated by patient outcomes and real world evidence (RWE). Further, there are initiatives to encourage market adoption by incentivizing payer coverage. Legislation has been introduced in Congress, which would expand Medicare and Medicaid payer coverage for PDTx meeting certain criteria, such as FDA approval.

In the meantime, DTx companies must continue to advocate for the value of their therapies, which can overcome barriers to in-person access and the availability of health care providers (HCPs) and may well offer a treatment modality that does not bring a side effect profile seen with many pharmaceuticals, while providing personalized and cost-effective treatment options to deliver care to patients.

[View source.]

DISCLAIMER: Because of the generality of this update, the information provided herein may not be applicable in all situations and should not be acted upon without specific legal advice based on particular situations.

© Hogan Lovells | Attorney Advertising

Written by:

Hogan Lovells

Hogan Lovells on:

Reporters on Deadline

"My best business intelligence, in one easy email…"

Your first step to building a free, personalized, morning email brief covering pertinent authors and topics on JD Supra:
*By using the service, you signify your acceptance of JD Supra's Privacy Policy.
Custom Email Digest
- hide
- hide