September 21 is World Alzheimer’s Day. This annual event organized by Alzheimer's Disease International (ADI), is meant to raise awareness about this terrible disease and the efforts our industry is making to find a cure. The fight against Alzheimer’s Disease (AD) has been a long and frustrating journey, with many failures of disease modification strategies due in part to the focus on later stage patients whose disease was too progressed to treat.

While there is still no cure or any meaningful treatments on the market able to slow down the progression of disease, there is reason to be hopeful. Increased awareness about the impact of AD, and efforts by multiple pharma companies to develop new therapies has resulted in a rapidly expanding AD development pipeline with a number of promising disease-modifying products now in the advanced clinical trial phases.

These drugs are all targeting the early stages of AD (prodromal phase) when patients and primary care physicians are largely unaware of the patient’s condition. Evidence suggests that new treatments are more likely to be effective in early stage patients, however identifying and recruiting these patients has been a major hurdle, with screen failure rates commonly at 75-85%.

Fortunately, we believe that advances in machine learning, artificial intelligence, and big data analytics will help push progress along.

85% failure rates

It was only in the last 10 years that it became possible to diagnose early symptomatic AD patients thanks to advances in biomarker development (both “wet” and neuroimaging) in live patients allowing to confirm AD pathology prior to dementia stage. However, drug developers are quickly finding that long standing recruiting methods don’t work for these patient populations.

The traditional approach for AD patient enrollment is to identify patients through direct-to-patient outreach networks and to target healthcare professionals (HCPs) who have historically treated AD patients and can identify potential patients from their own databases. However, patients with early stages of AD rarely talk to their physicians about the risks of this disease, which means they are not yet on the radar of the HCPs who are most likely to refer patients to AD trials. In addition, targeting patients at early AD stages significantly increases the pool of potential patients, which means sponsors need to bring in more specialists in cognitive disorders, and conduct more clinical evaluations and expensive, often invasive diagnostic tests, including PET scans and lumbar punctures. These assessments are costly and time consume, and are often ultimately futile as 8-in-ten recruits are likely to see negative results.

As an industry, we need to find a better way to more efficiently identify and screen these patients so we can expedite these trials. By 2050, more than 14 million people in the US alone will be living with AD. The challenge is how to diagnose them early enough in their disease to have a meaningful impact. IQVIA believes that Human Data Science– the integration of the study of human science with advances in technology and advanced analytics - can be an important part of this solution.

Human science meets data science

The use of big data and advance analytics, when combined with clinical expertise and sophisticated technology platforms, presents new opportunities to optimize the AD clinical development process, especially in patient enrollment, by providing alternative ways to identify and connect with patients who may be at a higher risk for AD.

Through the adoption of a machine learning predictive model, trained to ingest data from various sources, including health insurance claims, prescriptions, electronic medical records, familial history, and data collected from wearable devices, we now have an unprecedented look both back in time and into the future. We can see, for the first time, patterns in data that human analysis was unable to pick up. Correlations between attributes that help identify at-risk prodromal AD patient populations – maybe even before they are considered patients.

These (de-identified) patient groups can then be linked back to their HCPs, providing sponsors with a pre-defined population to drive more targeted recruiting and referral campaigns. The result is a more precise approach. An approach that takes full advantage of the innovation in both science, data and technology to ensure no time, and no resources, are wasted in finding a path to better outcomes.

Bigger data sets can also be used proactively to select trial sites in communities that are most likely to deliver higher enrollment numbers. Once the sites are open, a predictive analytics screening tool can be embedded at the site to develop a disease risk score that can be generated and viewed directly by the physician, which helps them to determine whether patients should be referred for additional diagnostic tests through a nearby screening center or with a dementia specialist.

Cumulative learning

The additional data collected through the diagnostic tests will be of value to researchers and patients, as well as providing a feedback loop to validate and further hone the predictive algorithm. Over time the additional data will improve the accuracy of the model in identifying prodromal AD patients, steadily reducing screen failure rates, and allowing patients to be diagnosed in a more cost- and time-effective manner. The benefits of analytics for recruiting has already been proven in other trial settings, and it offers incredible promise to help accelerate AD research and reduce the time and cost of recruiting for these trials.

The Alzheimer’s Association estimates that early and accurate diagnosis of AD could save up to $7.9 trillion in medical and care costs in the United States alone. Taking a Human Data Science approach to the most critical barrier – finding patients – can help us achieve this goal while increasing the speed with which we can finally bring new AD treatments to market. As we prepare for World Alzheimer’s Day, I think that’s something to celebrate.