Digital health innovation continues moving full-force in transforming the business of healthcare. For pharma and medtech companies in particular, this ongoing shift has pushed them to identify ways to create value for patients beyond the drugs themselves. From new partnerships between digital health and life science companies to revamped commercial models, collecting and extracting insights from data is at the core of these growth opportunities. But navigating the rapidly evolving terrain is no simple task.
To help these companies effectively incorporate and utilize digital health tools, Rock Health partner ZS Associates draws on over 30 years of industry expertise to guide them through the complex digital health landscape. We chatted with Principal Pete Masloski to discuss how he works with clients to help identify, develop, and commercialize digital health solutions within their core businesses—and where he sees patients benefiting the most as a result.
Note: This interview has been lightly edited for clarity.
Where does ZS see the promise of data- and analytics-enabled digital health tools leading to in the next five years, 10 years, and beyond?
Data and analytics will play a central role in the digital health industry’s growth over the next five to ten years. Startups are able to capture larger, novel sets of data in a way that large life science companies historically have not been able to. As a result, consumers will be better informed about their health choices; physicians will have more visibility into what treatment options work best for whom under what circumstances; health plans will have a better understanding of treatment choices; and pharmaceutical and medical device companies will be able to strategically determine which products and services to build.
We see personalized medicine, driven by genomics and targeted therapies, rapidly expanding over the next few years. Pharmaceutical discovery and development will also transition to become more digitally enabled. The ability to match patients with clinical trials and improve the patient experience will result in lower costs, faster completion, and more targeted therapies. The increase in real-world evidence will be used to demonstrate the efficacy of therapeutics and devices in different populations, which assures payers and providers that outcomes from studies can be replicated in the real world.
How is digital health helping life sciences companies innovate their commercial models? What is the role of data and analytics in these new models?
The pharmaceutical industry continues to face a number of challenges, including the increasingly competitive markets, growing biosimilar competition, and overall scrutiny on pricing. We’ve seen excitement around solutions that integrate drugs with meaningful outcomes and solutions that address gaps in care delivery and promote medication adherence.
Solving these problems creates new business model opportunities for the industry through fresh revenue sources and ways of structuring agreements with customers. For example, risk-based contracts with health plans, employers, or integrated delivery networks (IDNs) become more feasible when you can demonstrate increased likelihood of better outcomes for more patients. We see this coming to fruition when pharma companies integrate comprehensive digital adherence solutions focused on patient behavior change around a specific drug, as in Healthprize’s partnership with Boehringer Ingelheim. In medtech, digital health tools can both differentiate core products and create new profitable software or services businesses. Integrating data collection technology and connectivity into devices and adding software-enabled services can support a move from traditional equipment sales to pay-per-use. This allows customers to access the new equipment technology without paying a large sum up front—and ensures manufacturers will have a more predictable ongoing source of revenue.
That said, data and analytics remain at the core of these new models. In some cases, such as remote monitoring, the data itself is the heart of the solution; in others, the data collected helps establish effectiveness and value as a baseline for measuring impact. Digital ambulatory blood pressure monitors capture an individual’s complete blood pressure profile throughout the day, which provides a previously unavailable and reliable “baseline.” Because in-office only readings may be skewed by “white coat hypertension,” or stress-induced blood pressure readings, having a more comprehensive look at this data can lead to deeper understandings of user behaviors or conditions. Continuous blood pressure readings can help with diagnoses of stress-related drivers of blood pressure spikes, for example. These insights become the catalyst for life science companies’ new product offerings and go-to-market strategies.
What are some examples of how data sets gathered from partnerships with digital health companies can be leveraged to uncover new value for patients and address their unmet needs?
As digital health companies achieve a certain degree of scale, their expansive data sets become more valuable because of the insights that can be harnessed to improve outcomes and business decisions. Companies like 23andMe, for example, have focused on leveraging their data for research into targeted therapies. Flatiron Health is another great example of a startup that created a foundational platform (EMR) whose clinical data from diverse sources (e.g., laboratories, research repositories, and payer networks) became so highly valued in cancer therapy development that Roche acquired it earlier this year for close to $2B.
It’s exciting to think about the wide array of digital health solutions and the actionable insight that can be gleaned from them. One reason partnerships are important for the industry is few innovators who are collecting data have the capabilities and resources to fully capitalize on its use on their own. Pharma companies and startups must work together to achieve all of this at scale. Earlier this year, Fitbit announced a new partnership with Google to make the data collected from its devices available to doctors. Google’s API can directly link heart rate and fitness activity to the EMR, allowing doctors to easily review and analyze larger amounts of data. This increase in visibility provides physicians with more insight into how patients are doing in between visits, and therefore can also help with decision pathways.
Another example announced earlier this year is a partnership between Evidation Health and Tidepool, who are conducting a new research study, called the T1D Sleep Pilot, to study real-world data from Type 1 diabetics. With Evidation’s data platform and Tidepool’s device-agnostic consumer software, the goal is to better understand the dynamics of sleep and diabetes by studying data from glucose monitors, insulin pumps, and sleep and activity trackers. The data collected from sleep and activity trackers in particular allows us to better understand possible correlations between specific chronic conditions, like diabetes, and the impact of sleep—which in the past has been challenging to monitor. These additional insights provide a more comprehensive understanding of a patient’s condition and can lead to changes in treatment decisions—and ultimately, better outcomes.
How do you assess the quality and reliability of the data generated by digital health companies? What standards are you measuring them against?
Data quality management (DQM) is the way in which leading companies evaluate the quality and reliability of data sources. ISO 9000’s definition of quality is “the degree to which a set of inherent characteristics fulfills requirements.” At ZS, we have a very robust DQM methodology, and our definition goes beyond the basics to include both the accuracy and the value of the data. Factors such as accuracy and absence of errors, and fulfilling specifications (business rules, designs, etc.), are foundational, but in our experience it’s most important to also include an assessment of value, completeness, and lack of bias because often these factors can lead to misleading or inaccurate insights from analysis of that data.
However, it’s not easy assessing the value of a new data source, which presents an entirely different set of challenges. One very important one is the actual interpretation of the data that’s being collected. How do you know when someone is shaking their phone or Fitbit to inflate their steps, or how do you interpret that the device has been taken off and it’s not tracking activity? How do you account for that and go beyond the data to understand what is really happening? As we get more experience with IOT devices and algorithms get smarter, we will get better at interpreting what these devices are collecting and be more forgiving of underlying data quality.
What are the ethical implications or issues (such as data ownership, privacy, and bias) you’ve encountered thus far, or anticipate encountering in the near future?
The ethical stewardship and protection of personal health data are just as essential for the long-term sustainability of the digital health industry as the data itself. The key question is, how can the industry realize the full value from this data without crossing the line? Protecting personal data in an increasingly digitized world—where we’ve largely become apathetic to the ubiquitous “terms and conditions” agreements—is a non-negotiable. How digital health and life science companies collect, manage, and protect users’ information will remain a big concern.
There are also ethical issues around what the data that is captured is used for. Companies need to carefully establish how to appropriately leverage the data without crossing the line. For example, using de-identified data for research purposes with the goal of improving products or services is aligned with creating a better experience for the patient, as opposed to leveraging the data for targeted marketing purposes.
Companies also face the issue of potential biases that may emerge when introducing AI and machine learning into decision-making processes around treatment or access to care. Statistical models are only as good as the data that are used to train them. Companies introducing these models need to test datasets and their AI model outputs to ensure gaps are eliminated from training data, the algorithms don’t learn to introduce bias, and they establish a process for evaluating bias as the models continue to learn and evolve.