PM360 asked experts in data analytics about how the process behind collecting and analyzing data is changing, including these four key questions:
- How is the advancement of artificial intelligence, natural language processing, machine learning, and deep learning impacting the data analytics landscape?
- How has the rise of concern for data privacy and the implementation of the EU’s General Data Protection Regulation (GDPR) affected pharma’s ability to collect or use data?
- What can organizations do to help make their employees more data literate? Ultimately, how do you make everyone comfortable using data analytics?
- What is the future of data analytics? How else is data collection, analyzation, visualization, implementation, etc., likely to change in 2019 and beyond?
The dramatic improvements in computing capacity, both in terms of raw computational power as well as architectural advancements have profoundly altered the data analytics landscape. Although it may not be directly evident, the effects of this shift manifest themselves in various aspects of our day-to-day lives.
Netflix employs recommendation engines that analyze minute details of viewing habits to make suggestions that improve the user experience. Blind-spot assist sensors in cars can detect objects as small as a cell phone and provide collision warnings. Ads on webpages are selected about 100 milliseconds before the webpage loads based on the user’s history in a highly competitive real-time bidding marketplace.
In the end, the ideal objective for many of these initiatives is to advance living standards. In this regard, pharma has a unique opportunity to support and contribute to the emerging analytics-based ecosystem. Longitudinal patient analysis using advanced data mining tools can reveal complex patterns that cause adverse outcomes. Machine learning methods such as unsupervised learning can detect biologically similar patients across large volumes of genomic profiles. Commercial market research can use neural network models such as LSTM to improve forecasting models.
Pharma marketers can use machine learning to measure and optimize for the outcomes that truly matter—engaging the right consumers with messages that matter to them. We have consistently seen pharma marketers who leverage machine learning with their digital video marketing achieve higher engagement, lower bounce rates on websites, and greater pages per session. These are the metrics that matter—not traditional marketing metrics such as CPM, clicks, or impressions.
What gets measured gets managed, and with machine learning marketers can achieve both regulatory compliance and true engagement by consumers. In pharma, we often use what we call engagement experiences to qualify viewers as potential patients or caregivers. The AI uses that self-reported feedback to find more qualified audiences as the campaign goes on, thus reducing waste and resulting in more and higher quality site traffic.
Four principles apply: 1) Start by measuring the outcomes that matter; 2) incorporate customer listening right in the ad experience to leverage real-time human feedback; 3) leverage machine learning to measure and optimize for engagement in real time; and 4) apply a smart regulatory framework to creative, engagement, and customer understanding and retargeting.
At the beginning of the decade, you couldn’t have a conversation with anyone in the orbit of healthcare marketing without hearing about “big data.” Inevitably, marketers’ inability to transform all of this data into actionable insights sent the industry tumbling into Gartner’s “trough of disillusionment.” More than anything else, the rise of AI has provided pharma companies with the marketing and analytics tools they need to operationalize their expanding datasets.
Doing so effectively, however, requires a carefully considered approach, as even machine learning algorithms don’t run themselves. The processing power of these algorithms is indisputable, but the business case for machine learning hinges on whether it’s used to answer the right questions.
As such, it’s important for pharma marketers to recognize the limitations of their AI-based analytics tools. For instance, even the most sophisticated predictive tool can only assess the likelihood of various outcomes—not foretell the future—and should be treated accordingly. Ultimately, AI will be as adept at answering some key business questions as it is incapable of answering others, and understanding which questions fall into which category is the first step toward making AI a reliable addition to our marketing and analytics toolboxes.
The GDPR requires organizations that process the personal information of EU data subjects to comply with a statutory framework. The most prevalent change is a new definition of consent, which must be informed, specific, and freely given as well as the much discussed “right to be forgotten.”
Pharma companies usually process data on a legal basis known as “legitimate interest.” Under GDPR, this means the company has a reason to hold an individual’s information but has not obtained their consent. In this case, it is advisable to send out an information notice to the individual, giving them the opportunity to opt out. Previously, industry regulations allowed pharma companies to send educational and regulatory communication, but consent was required for promotional emails. Pharma companies were hence already operating on a similar premise. Having acted in a highly regulated data environment for many years, the pharma industry is hence better poised than others to handle recent regulations.
In fact, people taking more control over and ownership of their personal data actually provides an opportunity for pharma to change the narratives around data sharing. The benefits of patients sharing data to advance medical research are immense, and technologies such as distributed ledgers will make that safer and easier.
Protecting data has been a significant concern for healthcare and pharma long before GDPR. Certainly, the EU legislation has put in place more stringent requirements that will require additional diligence on the part of pharma, but also that of the broader healthcare community. But, pharma companies in general—because of their global footprint—are particularly at risk if there is a failure to get consent or comply with the requirements of GDPR.
All of this is happening at a time when accessing the right data is becoming even more important for drug discovery and other use cases. Despite these new challenges, we are seeing pharma companies adopt newer technology to meet these requirements, but also to reduce the cost of data acquisition.
As an example, we’re helping companies use CADEx, which is a blockchain-based platform for securely sharing encrypted datasets. This supports the protection of personal data and when combined with machine learning it supports the ability to de-identify data, mitigate usage conflicts, and make the usage of data more efficient. Through diligent legal and compliance processes and newer technology that can automate and secure the management of data, we’re expecting that most pharma companies will continue to successfully navigate the data security landscape.
Healthcare market researchers face many challenges today in protecting patient privacy. These challenges include: Keeping patient data anonymous and secure; the possibility of discovering patient information of which patients themselves are unaware; and the industry working on innovations that will change diagnosis, treatment, and monitoring of patients’ conditions.
The healthcare industry has numerous rules and regulations in place to protect patient privacy and healthcare information. In Europe, this information is protected by the GDPR, which took effect in May.
But, questions remain for healthcare market research, business insight research, or research sponsored by corporate businesses, such as pharma companies, whose objectives can be brand measurement, examination of unmet needs, patient preference, product improvements, general satisfaction, and marketing optimization. This research is different from research conducted for the advancement, discovery, or development of knowledge in the medical field. Industry associations, such as EphMRA and BHBIA, are publishing guidelines on their positions on such research in relation to GDPR. As a general rule, we recommend always securing consent if there is a direct interaction with the data subject. If not, then consider legitimate interest or public interest as a legal basis for data processing.
Access to massive amounts of raw data and analysis tools can quickly inundate users with too much information. Additionally, complex data transformations and processes can intimidate and lead to data distrust. A proven method to grow an organization’s use and acceptance of data analysis is applying a series of steps to build users knowledge and understanding, focusing on purpose and not process.
Visualization or using charts and graphs to represent the raw data is the simplest form of analysis. Presenting visualizations rather than raw numerical summaries demonstrates how to use data to make decisions and answer questions quickly and with more confidence. Beyond visualization is the application of actual analytics, deriving information through algorithms or other processes that generates more details.
The key to organizations successfully helping their employees to become more data literate is explaining how the transformations take place using non-scientific terms and without requiring a calculus refresher. Organizations can drive data acceptance and generate analytical interest by using examples that resonate with the employees’ backgrounds or functions. Replacing mystery with understanding leads to team confidence and readiness to embrace analytics.
The best way to drive data literacy starts by fundamentally transforming the role of organizations’ internal analytics teams from data doers to data leaders. Rather than concentrating expertise in a single department, modern analytics teams need to drive a culture of data-driven decisions from strategic to routine and tactical.
Literacy starts with access. Democratization of data is about delivering the right data to people throughout the organization. This requires an open, consistent data architecture that includes self-service tools along with sufficient training to make sense of the fundamentals behind the metrics: The context, the data sources, and the calculations.
Organizations need to invest in data expertise by bringing best-in-class thinking and staying on top of current trends in data science, data engineering, statistics, data strategy, and data platforms. Analytics teams can expand marketers’ decision comfort zone by supporting a process of identifying key business problems, developing a learning agenda, and backing up choices with predicted success and measured results. Successful use of data in decision making requires seeing it through from opportunity to case study, playing the role of data partner, and delivering on the promise of better decisions through better use of data.
Data analytics and AI are changing the pharma industry. Many organizations have already realized the urgency to promote data literacy and try to effectively turn tremendous amounts of data into insights. They are investing in data collection activities and are building analytical platforms. However, even with sufficient data and well-established IT infrastructure, many employees still hesitate to use them. One reason: The overall lack of the right skill set to transform data into business insights.
In order to alleviate this knowledge gap, rather than relying on courses or systematic training, working with a knowledge-sharing and inquiry platform encompassing data, tools, and technology allows users to search and study in a targeted manner. This approach could significantly improve learning efficiency and show great advantages in this fast-pace industry.
Moreover, instead of repeatedly responding to and stressing the importance of using data, management needs to develop a series of impactful business cases to demonstrate data’s true value. Successfully solving business problems by leveraging data analytics would build confidence within the organization and clearly demonstrate the value in using it. In the meantime, organizations should establish a set of standards and best practices to help employees use data efficiently.
One of the biggest challenges for data collection and analytics is the siloed nature of data storage, coupled with an inability, or unwillingness, to share.
New technologies, such as distributed ledgers, ease concerns with data-sharing through permissioned access, while allowing for a more holistic view of the patient experience, and product lifecycle. These technologies also bring with them a lessened technical commitment for integration, providing a greater opportunity for both breadth and depth of data.
As the digital landscape expands, the increased volume of data will allow for a deeper understanding of contributing factors in analytics. This leads to more insightful statistics that can be applied to tailor patients’ consultation and care.
Detailed statistics on prescription adherence and outcomes will also provide drug companies with a better understanding of where their products are being distributed, how much was charged, and under what circumstances. The full manifestation of this wider and deeper data-set has yet to fully reveal itself, but big things are on the horizon.
While enhanced predictive models and faster insight generation garner substantial attention, the future in analytics will likely be centered on creativity.
Mining existing data and leveraging the exponential growth of processing power will form a foundation of analytic thought, but those thinking about what it will take to gather additional data information and how it might be applied in new and nuanced ways will blaze the path forward.
Think about the importance of capturing not just swipes on an app, but the velocity and pressure in which that swipe is made: Does it tell us about the user’s day? His or her growing frustrations with the app? A level of urgency that might require a different level of intervention?
Said another way, the model of the past of gathering, organizing, and assessing data will eventually lead to the realization that all data is incomplete or potentially out of context. Those organizations who can think beyond the current availabilities and limitations, and then make the meaningful investment to find, create, or connect the pertinent data, will be ahead of the curve.
The future of data analytics will be nothing short of a revolution in the healthcare business model itself. With rare disease and specialty micro-markets becoming the focus, many companies will embed data analytics into the heart of their business models and use it to create differentiated and tailored treatment wellness plans for each patient. These efforts could achieve significant advances in the quality of health outcomes and reductions in the cost or burden of care, even if only modest advances are made in the therapeutic value of drug treatments.
Some of this revolution is already beginning, with the collection of personal genomics data as well as new frontiers in patient behavioral health management, which can offer game-changing improvements in real-world evidence generation and a new data science of behavioral health. To unlock these opportunities, pharma companies will turn their data collection, analysis, and visualization capabilities to face and better serve the end-consumer, while still considering the varied and dynamic needs of stakeholders throughout the treatment journey. Real-world genomics and health data will provide increased insight into genetic phenotypes, so companies can better predict patient-specific health outcomes and side effects as well as provide real-time interventions to improve treatment plans over the course of chronic illnesses.