The Role of Artificial Intelligence In Access

Artificial intelligence (AI) is dominating today’s world. Great expectations of progress and deep fears of what it means for individual jobs keep our society very busy. Corporations invest heavily in AI solutions. As a consumer, you can notice the vastly growing number of AI-driven customer service solutions. The quality of service is often still questionable, but the savings are substantial, and the AI train seems unstoppable.

How can AI play a role in prescription drug pricing and access?

AI and the Access Journey

To identify areas for AI application in pricing and access, we should consider the various steps that are involved in the access process. These steps are described in the “access journey” that is illustrated in Figure 1 below.

After a drug’s regulatory approval by the FDA or similar authorities in other countries, payers, medical associations, provider organizations, prescribers, and patients each play a role in enabling patients to access or receive a prescription drug. What role can AI play in each of those steps?

Pricing and Payer Access

Government and private payers place a high value on randomized controlled clinical trials in which a new treatment is directly compared with the standard of care. Payers are unlikely to relax these requirements. They often see unblinded data and placebo-controlled or single-arm trials as subject to bias and thus unacceptable. For the drug industry, AI can help better understand past payer decisions, depending on the types of clinical improvement claims within each therapy area. In the future, this may help predict the success of new drugs before pricing or reimbursement dossier submission in countries where the process is sufficiently transparent. AI may also help with trial design by more precisely defining the patients most likely to benefit from the treatment.

Clinical Guidelines and Value Frameworks

Medical societies have a critical advisory role in treatment practices and drug prescribing. AI can be of great assistance in determining the best treatment for specific patient types. Harvesting existing data for each disease area can help solidify a drug’s place in therapy after it has been used in many patients. Still, it can form a hurdle for new drugs, where such experience is not yet available.

Provider Formularies

Particularly in the United States, we have seen a lot of provider consolidation over the past decade. Most practicing physicians are now employees of integrated delivery networks (IDNs), rather than individual entrepreneurs. IDNs can use AI to ensure that their treatment guidelines and pathways deliver optimal treatment outcomes (quality metrics) and improve practice economics. They use incentives and individual performance metrics to ensure physician adherence.

Physician Prescribing

Physician prescribing behavior will be increasingly driven by medical societies’ and provider guidelines, as well as by individual physician performance benchmarks optimized with AI. Social media ratings for physicians, the “Tripadvisor for Patients,” are mainly driven by patients' feelings about a physician's bedside manners, as they typically don’t see any outcome statistics.

Treatment Fulfillment and Patient Adherence

Improved information systems and AI can provide the prescribing physician with a better dashboard to understand each patient's insurance drug formulary, help them avoid or circumvent prior authorizations, and find affordable options.

What Is the Largest AI Impact?

Since AI relies on large volumes of data, it is likely to have a particular impact on the use of existing drugs through medical societies’ clinical treatment guidelines and providers' practices and guidelines. The emergence of greater clarity on the best treatment for each patient may further increase the access hurdle for new drug treatments, as they become relatively more uncertain than experience-driven solutions.

What Are the Implications for the Drug Industry?

A better understanding of the value of existing treatments is likely to place a higher burden of proof on new treatments. It will entice payers to further clamp down on long-term outcomes data requirements unless there are expectations of breakthrough outcomes based on surrogate data and a strong rationale based on mechanism of action. A demonstrated commitment to postlaunch registry data and openness to risk-sharing agreements linked to the registry data can help to cement the deal. In order to improve provider and prescriber adoption, we need to think about how we can shape our evidence program to accelerate pathway adoption by specifically targeting the place in therapy that we are pursuing.

Ed Schoonveld is a value and access adviser for Schoonveld Advisory, LLC, and author of The Price of Global Health. He can be reached at [email protected].