2017 Annual Conference

Dramatic changes in the capacity, size and distribution of technology has meant an unprecedented acceleration in the generation of electronic data in recent years. This includes traditional health data, but also nontraditional sources like smartphones, shopping data and more. While the commercial sector has taken the lead in applying this big data, it’s clear that it will also be applied to both the traditional healthcare system as well as to health promotion outside that system. Dr. Joel Selanikio, a technologist and practicing physician, leads the audience through the definitions and origins of big data through examples of big data in business, and finally to the first steps—and the larger promise—of big data in healthcare.

As we enter an era of increasing transparency on patient and physician level data, marketing scientists are granted incredible access to data that can be used to improve patient outcomes in meaningful ways.

In this presentation, we share a straightforward approach for leveraging multiple sources of publicly available data to identify pockets of unmet medical need through network analysis. Taking this one step further, we then propose applying well-understood machine learning classification techniques to identify HCPs within those pockets that may have the greatest impact on addressing patient need across academic and community settings.

We believe marketing science leaders sit at the precipice of change and offer the following goals for this presentation:

• To share our belief that easy-to-access data sources can be productionalized to address practical questions on patient need;
• demonstrate the use of machine learning for predictive modeling for physician identification that can outperform traditional methods (e.g., expensive primary research);
• To make the case for data-driven, end-to-end systems that can be used to improve decision-making across a multitude of patient and physician-level signals.

The methods presented here represent a standardized approach that any team can take to model potential engagement opportunities.

In today’s world different physicians/specialties play key role in providing overall care to a patient. However, for a pharmaceutical company it is difficult to understand the exact role each of the physician is playing in the continuum of patient’s care and how those roles influence the overall treatment. It therefore becomes essential for a pharmaceutical company to establish and understand the networks of physicians and then track them over time to enhance messaging and promotional efforts and derive more value.

In this abstract we will showcase a solution that helps to identify the network of physicians and track the movement of the network with time.

The approach includes the following two pieces that work in tandem:

• Build the Network: Who are the different physicians involved in providing care, what are their different roles (diagnosis, referring, treating/decision maker and maintainer), and how are they connected among each other?
• Approach – Analyze the patient pathways using longitudinal patient level data. Identify the physician’s, associated roles and networks by tracking each patient longitudinally.
• Prioritize physicians based on volume, their roles and their connections (i.e.). Identify influential physicians and decision makers and connectivity to other physicians through network maps.
• Map the Key opinion leaders (KOLs) onto the identified network to understand their sphere of influence.
• Track the Network: How is the network evolving over time and how the physician valuation change?
• Approach – Use patient level longitudinal data to track the physician role and network pattern over time for patients of interest. Assess the performance against each of the network point built earlier. Identify how referrals to particular influential physicians are changing and the new physicians that are becoming key stakeholders in patient’s care.
• Rank each of the physician more comprehensively based on their true network connection, role, patient volume and influence.

Data Sources:
Symphony Health Solutions (SHS) patient level longitudinal claims and internal data from clients is used. The patient level claims data variables include: patient, prescriber, diagnosis or medical condition, plan / payor information, claim status (i.e. approved, rejected, and switched), procedure - surgical, medical, medically administered drug, and care facility type (i.e. emergency room, outpatient clinic, inpatient).

Conclusion/Take Away Message: This approach enables the client to:

• Identify influential players in the network
• Understand how the physicians are connected to other physicians
• Understand various roles physicians are playing in the network
• Map the KOLs to understand their sphere of influence
• Track and reallocate resources based on change in network
• Rank each of the physician more comprehensively based on their true network connection, role, patient volume and influence.

Thus to be informed, focused and optimize their promotional efforts in the right direction and enhancing their brand’s commercial potential over time.

To date, few studies have been conducted that link patient support program data to secondary claims data to measure real-world outcomes. We developed an innovative approach to evaluate a patient support program by linking multiple data sources including the IMS Pharmetrics Plus claims database to patients enrolled in a patient support program to demonstrate real world benefits of the program. The ability to work across companies to link program enrollees with secondary data is an important tool in helping to identify which programs are effective in delivering value to patients. We found that adherence was better, and that total health care and hospitalization costs were significantly lower in program participants.

Recent changes and pressures amid health care reform, industry consolidation and expanding regulations has transformed the life sciences landscape. As individual health care providers have merged into Integrated Delivery Networks or IDNs and become a more significant sales target for pharma clients, there is an increasing need for data to help navigate today’s highly complex, layered web of IDN relationships.

In this session, we will examine the evolving health care landscape, the complex and layered nature of IDNs and the related challenges, questions and demands in navigating IDNs. In addition, we will explore solutions for success with IDNs by looking at a case study that helps determine which groups are associated with particular specialties, track coordination of care across HCPs, IDNs and ACOs, build targeting strategies based on account affiliations, provider, practitioner and group relationships.

Why Important: IDNs have forced life sciences organizations/pharma clients to shift from traditional rep-to-provider selling models to an account-based, B2B selling models that focus on value, relationships and influence. This new IDN selling model demands new data solutions to understand the hierarchy of relationships to successfully execute business strategies.

Supporting Use Cases: In this presentation, we will examine how a life sciences firm recently used provider profile, affiliation and relationship data and volume-based medical claims intelligence to identify thought leaders and peer-to-peer learning networks in their analysis of the type 2 diabetes market in the United States.

The use case will cover:

• Understanding the IDN organizational structure of priority IDNs by looking at the size and structure of IDN network, the type and degree of influence/control and those with diabetes focus by leveraging strategic profiling elements and datasets.
• Identification of individuals engaging in activities related to diabetes such as inpatient/outpatient access and formulary decision making, quality of care and outcomes research initiatives, physician metrics, cost control, and risk-sharing policy making, product reviews, contracting, and distribution oversight, along with diabetes protocols and guidelines development
• Strategy development of programs, activities, and initiatives for diabetes chronic disease management
• Linkage of scientific/socio leadership with IDNs

Organizations in all industries are striving to create an analytics-driven organization like Amazon or Google. Pharma is no different. Terms like “Big Data,” “data scientists”, “predictive analytics,” and “machine learning” are becoming commonplace, though much like other disruptive innovations, there has been a healthy dose of hype.

Some pharma companies are starting to benefit from using new analytics techniques to drive business insights. A few examples include:

• Leveraging RWE data to drive commercial and brand strategy
• Leveraging digital information to segment HCPs based on content affinity
• Using traditional data, social media and text analytics to build KOL influence maps
• Building a suggestion engine to recommend the “next best action” for sales reps and digital

But most pharma companies have struggled to meet the aspiration. A recent Economist Intelligence Unit / ZS study found significant dissonance between the levels of investment in analytics and its impact on the business. Building a great analytics organization that drives how a company makes decisions is a multidimensional challenge.

Many factors have to converge to make this happen. In this presentation, we will discuss what it takes to construct a great analytics organization, and we will review an analytics maturity model that companies can use to gauge their own progress. The key dimensions of the maturity model are:

• How do we develop and hire the skills and roles (data analyst, business analyst, advanced analytics, data scientist, etc.)? How should new skills such as data science and UX integrate with existing skills? What skills should be insourced vs. outsourced?
• How can analytics organizations demonstrate value to the business?
• How should analytics organizations balance innovation and ‘doing the work’? How do we develop an agile environment that encourages experimentation? How do not get caught up in “lab mode” and scale and operationalize innovations?
• How do we drive change in our stakeholders to make them more comfortable with analytics-driven decisions and actions?
• What is the optimal structure (role of centers of excellence vs. customer-facing, global vs. local organizations)?
• How can analytics be treated as an asset?
• How do we best leverage the new Big Data and visualization technology without feeling paralyzed by the vendor choices?
• What is the role of offshoring?

This session will help the analytics professionals in the audience connect the dots between the challenges that they’re currently facing in their organizations and the strategy of building a world-class analytics organization and embedding it into commercial decisions.

Gone are the days of the simple physician-patient relationship. The rate of change is accelerating as the march towards value-based care collides with advancing technologies, computing power and connectivity. There are more market uncertainties, increasingly complicated relationships, and new, non-traditional players in healthcare data and delivery. Problems in this environment appear increasingly unsolvable as they are constantly changing in their dynamics. But the same forces driving the complexity and uncertainty are presenting novel solutions and breakthrough opportunities. Kevin will share experiences from exploring exponential technologies and organizations.

Our industry is changing faster than ever before. HCPs and patients are changing at the speed of the newest consumer technology innovations such as Google Glass and Apple iWatch. Policy and regulations are shifting. The healthcare ecosystem is consolidating, acquiring and divesting.

In this world, companies will be differentiated by their ability to innovate: to take advantage of the latest data sources, integrate them into the enterprise and drive meaningful insights.

However, many companies are impeded in their ability to innovate by several factors including:

• Data Management strategy and technology
• Organizational silos and regional variations
• Limited analytical expertise and ecosystems

We conducted a benchmarking study on client readiness to deliver analytics innovation. Our study provides insights on questions like:

• What are the emerging data sources that are being leveraged, and how?
• What are the Big Data technologies are being used, and which are prevalent?
• What are the trends with unified data management and governance?
• Which are key initiatives, focus areas and innovations that are being driven?
• Which models for analytics service delivery are used, and what is the role of partner strategy in this regard?

While immunotherapy represents a promising advancement in oncology treatment, the advent of these therapies has also heightened challenges and prompted new business questions for pharmaceutical companies. Since 2014, three anti-PD-1/PD-L1 agents have launched with indications in multiple tumors, and many others are in late stage clinical trials. These immunotherapies work across tumors, potentially disrupting the current treatment paradigm of tumor-specific treatments. Given the growth of these new multi-indicated therapies, pharmaceutical companies are tasked with building more innovative forecasts to predict product usage as well as understanding how oncologist prescribing trends can inform immunotherapy adoption.

Traditionally, pharmaceutical companies have relied on analogs, primary market research and industry benchmarks to help address these questions. However, gaps exist in these data sources. Electronic health records (EHR) serve as an alternative approach to address these questions using real-world evidence of oncology patient-level EHR data. EHR data can provide objective insights into drug utilization trends in near real time. We will examine anti-PD-1/PD-L1 uptake and biomarker testing patterns in Melanoma and NSCLC to identify trends that may predict treatment patterns for a multi-indicated product launching in a new disease area. This research will be a joint collaboration between Flatiron Health and ZS Associates.

Over the past four years and for over 50 brands, Crossix has been the leading innovator in digital campaign measurement within the pharmaceutical industry. Our work has led the pharmaceutical industry to move away from flawed proxy metrics such as customer survey and website clicks to instead directly link digital media exposure to diagnosis and treatment-based clinical metrics:

• Audience quality – percentage of those reached satisfying a diagnostic or treatment criteria
• Intent to Treat – percentage of those reached who visit physician or pharmacy
• Conversion – percentage of those reached who start treatment with the advertised brand or within a therapeutic sub-category

In this study, Crossix and Optum present a meta-analysis of 20 recent digital campaigns. These campaigns span a wide range of therapeutic categories, and form a mix of branded and unbranded campaigns. This analysis is performed at the campaign/publisher level and is focused on display and video advertising across both desktop and mobile devices.

After the meta-analysis, we will show the impact of enhanced real-world evidence clinical data on digital measurement, specifically; this data supplements mid-campaign audience quality to better predict end of campaign conversion, and it provides insights at key points along the patient journey.

Machine learning is a group of highly advanced predictive analytics methodologies which allow us tap into the full potential of the health claims data in order to gain insights into a particular market or therapeutic area. Unlike traditional methods of analytics used in the industry which takes a deductive approach through the use of business rules, machine learning takes an inductive approach to market exploration. Methodology: The health claims data is a rich data source however can perform poorly with categorical variables when derivative data structures are not leveraged in order to bring forward the key variables. Case Study: Using Machine Learning and health claims data allows to identify the diagnoses, procedures and prescriptions which are most predictive of a patient receiving a particular preventative therapy. Levering these insights to develop a model to score providers based on their propensity to treat patients in their practice which demonstrate characteristics that make them ideal treatment candidates. Conclusion: This Machine Learning Based Targeting provides valuable insights and a relative score of provide targeting value which allows for a better assignment of sale force resources verses the traditional deciling approach. This methodology is especially useful for targeting in rare disease spaces including oncology, immunology and rheumatology.

Undifferentiated messaging does not always address HCP’s information needs effectively. Nowadays, patients are more likely to seek information on their own than in the past. However, patient awareness does not always lead to symptom and treatment discussion with HCPs without effective HCP touch points with pharma representatives. A holistic approach, including impact analysis, predictive modeling, and DTC analysis can be combined to better address HCP and Patient information needs.

Impact analysis provides insights about the effectiveness of HCP touch points, and helps to optimize their frequency. Predictive Modeling output can be utilized to generate segments. Personalized messages can be designed for each HCP segment to ensure that HCPs receive tailored messages to address their information needs. A pilot can be designed and implemented to evaluate personalized messaging effectiveness.

DTC programs increase patient awareness of existing treatment options, and initiate patient dialog with HCPs on symptoms. National DTC analysis can be conducted to evaluate overall DTC effectiveness. Regional DTC impact analysis illustrates DTC effectiveness by region, therefore provides guidance on DTC effort by region.

Personalized HCP touch points and patient DTC work together and address unmet patient needs effectively.

This presentation will cover the following topics:

• Enabling better and faster insights by replacing Chart Audits with EMR for tracking KPIs
• Integrating behaviors from EMR data with physician perception research to guide brand strategy
• Understanding reasons for patient drop off by creating patient journey
• Leveraging EMR platform to reduce patient OOP burden

Patients are more engaged in their healthcare decisions as they face increasing cost sharing responsibilities and are better informed of treatment options. Considerations for brand adoption go beyond clinical evidence to include patient preference and values. Therefore understanding “patients as consumers” aspect of the healthcare ecosystem presents new opportunities for pharma companies to enhance patient care quality and achieve brand’s commercial success.

Patient’s consumer attributes such as demographic/social economical characteristics and life style have been used to inform direct-to-consumer marketing and patient support programs. To further leverage these insights and formulate an integrated marketing and sales strategy, there is a pressing need to incorporate “patients as consumers” metrics in sales force effectiveness such as resource allocation, messaging and promotional offerings to increase adoption, trial and loyalty.

In this presentation, we will share a novel approach, with a case study, to introduce ‘patients as consumers’ dimension to sales force effectiveness and highlight its importance and value in impacting HCP’s prescribing behavior. It provides readily actionable recommendations for sales force resource planning and promotion tactic mix. Adding “patients as consumers” dimension, together with other HCP prescribing influencers, allows us a more comprehensive and robust approach to drive sales force effectiveness.

Social listening data can be a viable and valuable input to sales forecasting. We are trying the address the following questions:

• Is it viable to include the social listening data in the sales forecasting equation? More specifically, what social listening data are we going to use and how do we perform quality assurance checks on it?
• Is it valuable to include the social listening data in the sale forecasting equation? More specifically, would social listening data help predict the drug sales?
• Can social listening data provide a new prospective? For instance, could pharmaceutical companies place more emphasis in the social media space to increase their drug sales?

Agent-based models (ABM) are classified as a class of computational methods for simulating the actions and interactions of autonomous agents with the aim of measuring their effects on the system as a whole. ABMs use elements of game theory, complex systems, computational sociology, and evolutionary programming as well as Monte-Carlo methods are used to introduce randomness. ABMs are simulations that explicitly represent individual agents - humans, institutions, or organisms – along with their defining traits.

Simulation results emerge from the interactions among agents such as patients, physicians, sales representatives and managed care organizations. The results of the simulation can be used for a wide variety of analyses that may include forecasting prescription volumes and patients, understanding responsiveness to marketing and sales stimuli and simulating the impact of formulary wins/losses.

Industrywide, despite often high aspirations, six out of ten product launches under-deliver against expectations. The low success rate is due to pharma/biotech’s inability to successfully adjust its go-to-market approach to the increasingly heterogeneous local health care markets. At Enginologi, we believe that this heterogeneity requires a more in-depth understanding of what drives launch success. It is not enough to ask whether a certain go-to-market approach is successfully driving a launch, but where and under what circumstances it is successful. Enginologi discusses real world (blinded) examples of how state-of-the-art analytics have helped better prepare for successful commercialization and, where necessary, turn around stalled launches.

The FDA is the gatekeeper for new product approval in the United States. The Agency has a long history of speeding the regulatory pathway for therapies that represent the first available treatments for a disease or have potential to materially improve the standard of care for serious conditions. This began in 1992 with the FDA Accelerated Approval Program, and in 1997, the FDA Modernization Act created a “fast track” designation. Ten years later, the 2007 FDA Amendments Act created “priority review.” Most recently, the FDA Safety and Innovation Act (FDASIA) of 2012 created “breakthrough therapy” designation. This research evaluates the impact of the "special designations" on product performance.

With the growing presence of social platforms and with around 40% of the world online, the Pharmaceutical industry’s stakeholders' online presence is also growing. The Pharma has traditionally looked to historical marketing engagement activity and market research to understand customers. This growth of online presence and dialogue offers Pharmaceutical companies an opportunity to learn real-time about customer behavior, preferences, reactions, influencers, and markets. Insights from these online conversations is only amplified when this data is merged with structured data.

However, it is a challenge to find Pharma-relevant conversations among the millions of conversations present online and collecting data via social listening tools is only the beginning. Conversations need to be put in context to help extract insights.

In this presentation we will share how modern technologies, such as big data and natural language processing (NLP) can be used to help put raw, online conversations in context. We will also share what insights can be gleaned from contextualized conversations and how these conversations can be joined with traditional, structured data sets for more advanced analytical applications. For example, social data can be used to validate forecast assumptions and to better forecast clinical trial enrollments. Join us to learn about how to analyze social data and the opportunities made available from doing so.

This presentation will showcase the strategy and framework utilized for a successful Data Governance Program implementation at Janssen Commercial Excellence and the business value it generated – both in terms of financial ROI and process excellence. We believe that it will be of immense value for our industry peers to learn how the Data Governance framework can enable next generation Analytics and Insight Driven Commercial Excellence.

Vast changes in the global healthcare landscape have transformed the way drugs are prescribed. The power of making decisions around treatment is moving away from individual physicians towards key decision makers within hospitals, group practices and integrated delivery networks (IDNs). These decision makers decide what drugs are prescribed and also have some influence on the drugs being placed on formularies. Purchasing decisions are seeing a shift from multiple local sites to singular centralized headquarter locations. Additionally, as high prescribing physicians relocate, it has become crucial to track their movement within or outside the network. Hence, pharmaceutical companies have to increase their sales and marketing focus on creating and managing quality relationships with these key decision makers (Key account management).

Clinical trials are the engine of pharmaceutical innovation, leading to the development of new therapies as well as extending the uses for existing ones. They are essential to a healthy pipeline ensuring the continued growth of the company. In oncology and other aggressive diseases, they also offer treatment to patients where no other option exists. Clinical trials are complicated and expensive to set up and run.

Typically, the design and management of a clinical trial falls to three broad groups; the medical teams, who design and lead trials; clinical operations teams who plan, execute and manage trials; and the medical science liaisons (MSLs) who maintain relationships with leading KOLs and principal investigators at major academic institutions and clinics.

Clinical trials involve the recruitment of patients at participating sites. A site is a hospital or clinic (academic or community), that agrees to participate in the clinical trial and commits to recruiting a certain number of patients. At a given site, the trial will be entrusted to the principal investigator, a physician who is responsible and accountable for conducting the clinical trial. The PI assumes full responsibility for the treatment and evaluation of human subjects, and for the integrity of the research data and results, although s/he may be supported by a team of sub-investigators.

At a high level, the process of setting up and completing a clinical trial is as below. Once the trial design and protocols are finalized, the MSLs and the clinical operations team will propose potential sites for the trial. This is called site identification. Following this, the clinical trial leads will shortlist the sites and backup sites to be included in the trial.

The cost of setting up each trial site is very high, and hence appropriate site selection is crucial. This presentation will focus on the “site identification” step that results in a set of high-priority sites for the trial leadership to choose from.

The Millennium century we live in now has a direct impact on the way pharmaceutical companies evolve their promotional strategies and build innovative touch points to reach consumers. Companies are taking advantage of new capabilities and adopting new promotional channels to expand their reach both physically and digitally. However, companies should also examine their traditional mindset to generate value through all channels by providing the right message at the right time. Moreover, they should evaluate and customize the context and timing of the messages. A holistic messaging approach not only offers robust efficient execution, but also more effectively satisfies customers’ needs.

Pharmaceutical companies spend billions of dollars on sales compensation plans, but plan modeling techniques continue to lag other industries in their sophistication and rigor. Across organizations of all sizes, the most common technique used is “back-testing”, a simplistic approach of running last year’s results through the new plan. Enhanced modeling uses backward looking data to generate assumptions, which can then be adjusted based on future expectations. This allows for scenario based modeling that “stress-tests” the plan under a range of possible scenarios, not just what happened last year. The primary methodology used was monte carlo simulation, which leverages iterations of simulated data in order to generate a robust set of findings.

It has become well publicized that many international regulatory agencies have pursued a multitude of methods with the intent of limiting the price that a pharmaceutical manufacturer can charge during a global launch. One of the most common practices for negotiation between companies and governments for reimbursable prices of pharmaceutical products is the application of International Reference Pricing (IRP) rules. IRP is perhaps one of the most significant challenges that the pharmaceutical industry currently faces and clearly indicates that the industry operates in a globally interconnected environment. In its simplest form, IRP is a government attempt to compare the price of a pharmaceutical agent to that of comparable countries, resulting in a benchmark price that is not substantially different from those compared. As the launch sequence progresses, referencing relationships between countries evolve into larger and more complex networks that become extremely difficult to navigate, resulting in a challenge that closely resembles the butterfly effect where a small change can have a significant impact later on. The driving motivation of this practice is to impose the effect of price erosion as companies launch their products in more countries over time. Furthermore, once global price is lost it is lost for good, forever limiting a brand’s revenue potential.

Much of the research to date has focused on industry behavior from the perspective of economic policy. Studies have, for example, taken a statistical approach to explain market behavior as it reacts to IRP. However, very little has been done to address optimal launch policies despite the significant financial, political, and reputational implications of pricing. Our experience shows that the industry profoundly needs more rigorous capabilities to compute, address, and implement solutions to this problem. We present a methodology, centered on Mixed Integer Linear Programming, which has been implemented in practice by SAS to address the challenge of determining the optimal launch sequence of pharmaceutical products in a global portfolio. We have found that by implementing this framework for launch sequencing, better launch decisions can be made resulting in reduced price erosion and more revenue generation for the rest of the product lifecycle.

A Pharmaceutical company is launching a new therapeutic treatment in a complex disease state characterized by a lack of clear diagnostic criteria, but generally associated with multiple co-morbidities (including obesity, diabetes, and hypertension).

Targeting physicians likely to have eligible patients in this space is made difficult by the lack of diagnostic criteria (or even an unambiguous ICD9 code specific to the disease). To develop a robust targeting and SF sizing plan, Axtria and the company developed an innovative use of APLD to assess patient likelihood of being at risk for the disease, and assigning them to the doctor most likely to initiate therapy (if the disease is actually present).

The approach depended on scoring patients based on multiple attributes associated with the disease state, including age, comorbidities, other related diagnoses, and overall engagement with the healthcare system. The highest-scoring patients were tracked to the physician most likely to be able to properly diagnose and initiate treatment.

This presentation will outline the process and discuss the decisions / tradeoffs required.

High value physicians are those that are associated with cohorts of patients who have key diagnostic and/or prescribing attributes. Patients that acquire one or more of these key attributes can then be mapped to their associated physician, who are in turn ‘triggered’ for a sales call. The likelihood that a patient will be prescribed a drug is often strongest when a combination of key attributes are present. The space for even a modest number of triggers can be large. For example, testing 10 triggers with 2 states (True/False) represents a search space of 2036 unique binary patterns, each of which needs to be enumerated and tested. Here we present an efficient machine learning approach based on TREE algorithms, to search this space for optimal trigger combinations by building supervised learning classification models, and then extracting the TREE node-specific information as database query rules for the best patient cohorts within each model. These rules can then be used to monitor and trigger physicians. While the search is not exhaustive, it can be conducted in a fraction of the time it takes to test all possible combinations, and the process becomes more efficient as the space grows.

Our primary goal as business analysts is to answer business questions. This is of paramount importance as our answers inform and shape a slew of questions which in turn spawn questions that our answers inform and shape. That's how we ultimately define the trajectory and destiny of our company, for better or for worse. Answering a business question starts with choosing the right database. Unfortunately, this first step is too frequently the first misstep. But it does not have to be this way. We'll present a 2-step approach that ensures we'll choose the right database for the job.

We should not celebrate just yet. There are two more hurdles to clear. They look deceptively simple but have tripped even the most seasoned of analysts. The first one is what we call the Principal-Agent problem and it has to do with the challenges of describing a task to an analyst in a way that the results of the analysis are just as good or even better than if we had performed the analysis ourselves. The second hurdle which we refer to as the Galileo's Dilemma arises when we have to present findings that fly in the face of the company's beliefs, whatever those beliefs may be. We'll provide insights to increase the odds of clearing these hurdles.

With more data than ever, life sciences organizations continue to invest heavily in using this data to enhance their commercial innovation. Such investments are leading to data-driven insights about how they might be able to alter their sales force and marketing strategies: new sales roles, key account models (KAM), innovative non-personal promotion, multi-channel DTC campaigns, and more. These types of commercial initiatives have the potential to increase outcomes – however, many investments will not generate the desired results. To reduce the risk inherent in innovation and to understand how to optimally deploy resources, organizations are embracing rapid and statistically-robust business experiments to accurately determine the cause-and-effect impact of each commercial strategy before taking on the risk and expense of rollout, as well to understand the adjustments necessary to generate maximum ROI.

One of the challenges for many pharmaceutic companies is to reach potential customers who are not yet aware of their products. In this presentation, we will illustrate how we use look-alike models to identify the most valuable prospects and significantly increase our clients’ marketing reach by targeting these prospects. We will discuss the steps we take to build and implement the look-alike models, and show how we used the model to support online display campaigns to acquire leads into CRM programs.