Efficiently marketing to sufferers of rare diseases while preserving patient privacy is not an easy task — but it’s an important one, especially when lives are on the line. Take hereditary angioedema (HAE), for example. Most people have not heard of this condition; only one in 66,000 people has it. Living with HAE is a challenge as it can cause swelling in any part of the body, including a patient’s face. If the condition affects their throat, it can even be fatal.

Fortunately, a pharmaceutical company created a free app to help HAE patients last year. This app enables patients to share information with doctors and learn more about life-changing, and even life-saving, treatment options.

However, while the HAE app has the potential to greatly improve patient lives, marketing this app to the very patients who would benefit from it is a near-impossible feat given its rarity. Patients suffering from less common conditions deserve to have helpful information. In these situations, it takes a solution that combines precision with respect for patient privacy.

Reaching Rare Disease Patients While Preserving Privacy

It is vitally important that patients are not targeted based on their individual health information. At the same time, this is challenging because the very nature of marketing to a smaller population means traditional channels, built for scale and broad awareness, become less effective and efficient.

Rare diseases are typically defined as conditions affecting fewer than 200,000 people. The challenge with marketing to rare condition sufferers is that campaigns can’t afford to rely on mass marketing to reach these patients. The costs are often too high while the condition prevalence is too low to justify the budget. These limits put a particularly severe strain on the ability of companies, like the one behind the HAE app, to educate people about life-changing options.

The math behind this challenge is simple: the U.S. is home to 330 million people, roughly 5,000 of whom are HAE patients. That means that only 0.0015% of the U.S. population has this condition. Just because a patient’s condition is rare doesn’t mean they should feel alone or lack access to resources that could improve their lives.

How can a healthcare marketer possibly reach these patients and improve their lives while preserving patient privacy?

Thankfully for healthcare marketers, precision and patient privacy are not mutually exclusive. With DeepIntent’s patented Patient Modeled Audiences technology, we are able to create custom (and privacy-safe) audiences comprised of people with a higher likelihood than targeting the U.S. population to have HAE without targeting based on their health data or other protected personal information. We ultimately helped the company with the HAE app exceed its download goal by 45% and most importantly, helped a group of patients manage their condition with a closer link to their healthcare provider (HCP).

Patient Modeled Audiences Combines Precision and Patient Privacy

In 2020, DeepIntent launched Patient Modeled Audiences, which was later named one of the year’s “most innovative products” by PM360. Patient Modeled Audiences is a custom modeling solution that creates unique, HIPAA-compliant audience segments for programmatic activation.

The Patient Modeled Audiences process uses data science to determine the correlation between demographics and de-identified data related to diseases to create campaign-specific modeled audiences. These models are precise while still protecting patient privacy. Once custom models based on these correlations are built, they’re brought into our healthcare demand side platform, where users are scored based on demographic attributes — not health data or patient status.

Patient Modeled Audiences targets users based on these demographic correlations rather than patient or health data. As a result, the solution allows healthcare marketers, in a privacy-safe way, to increase the rate at which they reach qualified audiences for patients with both common conditions and rare diseases. This is especially important for the latter. With Patient Modeled Audiences, marketers can raise awareness of conditions, treatments, and clinical trials, 37% of which usually fail to meet their enrollment goals.

Differential Privacy Ensures Patient Privacy is Protected

In developing Patient Modeled Audiences, DeepIntent partnered with Mirador Analytics, an independent third-party consultancy that evaluates the process and ensures the use of de-identified data meets HIPAA requirements. However, with differential privacy as a core aspect of our technology, we’re able to go above and beyond HIPAA requirements while preserving the power of our targeting models.

Differential privacy is technology founded on cybersecurity principles of preserving privacy. This approach to privacy can be applied to machine learning algorithms to extract critical insights and generalized learnings from datasets without tying a generalization to a specific person.

Using differential privacy, we intentionally include predictive error or “noise” to make sure our model isn’t learning to identify individual patients. Differential privacy improves the anonymity of datasets already stripped of any personally identifiable information (“PII”) from our model building process.

Machine learning algorithms relying on differential privacy can identify patterns, such as that between demographics and health conditions. Importantly, differential privacy is uniquely known to preserve and improve privacy while actually improving the ability to recognize such correlations.

This means that when healthcare marketers build Patient Modeled Audiences, not only is their targeting solely demographic-based, but differential privacy ensures that PII or health data (“PHI”) is not used within their marketing campaign. By helping machine learning algorithms better recognize correlations, differential privacy ultimately improves their ability to target those that could most benefit from specialized treatments or clinical trials that these marketers have to offer.

Patients Prefer Relevant Advertising

Health advertising improves patient health outcomes while filling a gap in public health awareness. Last year, we surveyed patients to better understand how they feel about pharmaceutical advertising.

More than half (51%) of respondents find ads more memorable when they’re relevant to their medical conditions. This highlights the incredible opportunity healthcare marketers have to improve patient education by personalizing the information they receive, helping them make more informed healthcare decisions.

At the same time, this greater awareness empowers patients to have more informed conversations with their healthcare providers, building more trust between the two. Through an earlier survey, we learned that patients are also more likely to adhere to treatment recommendations for something they recognize from advertising.

Gone are the days when precision and patient privacy were mutually exclusive.

Thanks to emerging technology such as differential privacy, DeepIntent is able to go beyond HIPAA compliance and work toward our mission to improve patient outcomes through the artful use of advertising, data science, and de-identified real-world clinical data. As industry standards have continually evolved to keep up with new technology and opportunities to improve privacy, Patient Modeled Audiences allows DeepIntent to lead the path to balancing patient privacy and improving patient outcomes.

Companies that deal with health data should be certified under rigorous industry privacy and security standards. And at the same time, our industry needs standards that are actively maintained and support innovation. Delivering relevant health information to healthcare providers and patients is vital for improving patient education and patient lives.

Patients deserve to benefit from the advances in data science and privacy technology that make that possible without compromising their privacy.