2018 Annual Conference

The purpose of this tutorial is to train attendees on measuring digital healthcare promotional campaigns. We will cover methodologies and examples from both direct-to-consumer campaigns and non-personal promotion to healthcare professionals. Instruction will consist of lecture-style presentations and multiple hands-on exercises that use anonymized actual digital promotional data and results.

After the presentation, attendees will:

• Understand the current digital landscape and the evolution of key metrics
• Learn best practices for digital campaign optimization
• See benchmarks across the pharmaceutical industry
• Analyze actual (anonymized) digital media and website data and apply it to real-world campaigns

Students should bring a laptop with Microsoft Excel for measurement exercises.

Preliminary Agenda

• Motivation for Digital Campaign Measurement
• Rise of digital advertising spend in pharmaceuticals
• Need for accountability and return on investment
• Mid-campaign media optimization
• Measuring within Digital Activity
• Types of digital channels: search, display, email, social, etc.
• Engagement rates: open, click-through, views
• Website analytics, including key onsite activities (e.g. doctor discussion guides, sign-ups)
• Tools: Google Analytics, Site Catalyst, Social Media Platforms
• Benefits and limitations
• Practical exercise with measurement and brand insights
• Measuring Impact of Digital Exposure to Offline Health Outcomes
• Methodology: tagging, registrations, log file analysis
• Audience quality: are you reaching the right health targets?
• Match back analysis to post-media behaviors
• Calculating Rx and campaign impact
• Practical exercise with measurement and brand insights
• Advanced Topics
• Cross-channel media analyses
• Attribution in health and other industries
• Integrated measurement of offline and online promotions

As companies ask their employees to take more risks, innovate at every turn and anticipate the future, courage is becoming an ever more critical leadership and job skill. Organizations comprised of courageous individuals and leaders are able to make difficult decisions quickly, take advantage of market opportunities and grow their organizations during even the most challenging of times.

In this presentation, Cindy shares the learnings of over 5,000 interviews about courage, how to build it and how to use it. She helps audiences understand and identify their unique type of courage, and shares what every leader and individual needs to know to ‘build their courage’ and create teams that communicate, innovate and learn from every experience.

Today, 90% of data created are unstructured, and it is growing faster than structured data. Among the unstructured data in pharma, a large portion is in the form of images and PDF files, such as scanned EMR/EHR records. As the volume of these PDFs and images grows, it becomes increasingly difficult to turn them into valuable information quickly. The main opportunity costs of such dilemma are costly overburdened data warehouse, slow access of other structured data, and inability to learn from valuable information, especially in the field of precision medicine and cancer research. In this research, we intend to address the following three questions:

• How to utilize and combine big data and distributed technologies to manage unstructured data efficiently?
• When can machine learning and natural language processing be used for analyzing unstructured data, ranging from images, text documents, to semi-structured data?
• How to experiment with and test different machine learning tools and algorithms for production and scalability?

The promise of predictive, real time multi/omni-marketing has been around for years, yet most - even sophisticated - pharmas/biotechs have not yet been able to implement a model that actually works. Yes, there are databases and trackers that tell you customer channel preferences or whether they have engaged with a particular content. There are suggestion engines that put forward next best actions based on simple rules or on what your best reps are doing. But what good is preference or engagement when we are measured by dollars/Rxs? What good does an understanding what the best reps are doing when most reps are - by definition - not your best reps and therefore cannot execute like a ‘best rep’? What good does it do when you optimize one channel at a time when the true value of your commercial footprint is only realized when all channels play well together?

Well, the honest answer is: Very little. What most companies still struggle with is to implement a system of technology, people, data, analytics, processes that looks across the usual silos of sales, marketing, and services, really understand how they all work together, and execute against these insights in real time.

If you have a great mechanic you completely trust, you’re likely to stay loyal, even despite the occasional mistake—that’s the bottom-line impact of a good customer experience (CX). That impact is why CX is a vital pursuit for nearly all industries. But what about health care providers? Could the objective decision-making be influenced by the emotional benefits of a pleasant experience?

While pharmaceutical companies do make efforts to improve CX, without proof that CX will have an impact on performance, changing a busy rep’s behavior can be challenging. But it’s not just reps who would need to change behavior. Because pharma has so many customer touch points, a comprehensive CX push would need to go well beyond rep calls and would require a significant cultural change within the organization.

This leads to the fundamental questions the presentation will aim to address: How can pharma companies measure customer experience? What role does customer experience play in influencing the customer decision making and how can it be quantified? What can pharma companies learn from the linkage of customer experience with performance to effect a behavior change across their functions?

The terms “Dashboard” and “KPI” have lost their meaning. “Dashboard” invokes the image of an instrument panel of a high performance car. True KPIs focus the user on a small number of metrics accounting for most of business success. However, today’s dashboards boast 70 Powerpoint slides or 50 Excel tabs and hundreds of KPIs, making clear that there is a lack of understanding about what's important.

Today’s marketing environment is more complicated than ever with channels multiplied almost beyond manageability. Someone needs to help the business sort out what is truly “key.” Enter Analytics. Our discipline is uniquely poised to provide the data-driven direction needed. Basic experimental design and mix models can indicate which promotions are driving the business both from a contribution and ROI perspective, the blocking and tackling of our trade. The greater challenge is to put analytics to work driving decisions.

We will describe an ongoing project that uses Management Science tools to identify the factors that predict brand success. The goal is to create a strategic, multi-channel dashboard which focuses decision makers on the factors affecting performance, screening out noise, resulting in better, faster resource allocation.

Join us in thinking through these issues together.

Studies have shown that 40% of launches across industries fail. In bio-pharma, tracking and monitoring launch progress are one of the top 3 launch capabilities that correlate with success. Tracking launch progress is crucial for achieving launch goals.

Traditional launch trackers have been focused mainly on sales performance, that provide limited insights on the overall product/brand performance. A comprehensive cross-functional launch tracker, on the other hand, serves to be a better option to propel the launch by unlocking issues and opportunities early on.

In this session, we will share our collective experience across multiple launches both in biopharma and other industries in the form of 6 best practices in designing an effective launch tracker. We will use two recent launch case studies to illustrate these best practices. The best practices range from “How to choose metrics” to “How to motivate cross-functional teams” and “Measuring success even before sales”.

For anyone working or planning to work on a product launch of any size in any therapeutic area, this presentation will provide actionable insights to enhance launch planning.

April Todd, Senior Vice President, Avalere, leads the Data Analytics practice, which provides creative data-driven insights to inform strategic solutions for complex business challenges and policy issues.

Prior to joining Avalere, April established MNsure, Minnesota’s health insurance exchange, and served as its Executive Director. Prior to that, as Minnesota’s State Health Economist, she led a team of researchers and analysts tasked with monitoring, analyzing, and advising on health policy issues related to cost, quality, and access. April also served as the Vice President of Strategic Analysis and Communications for Government Affairs at UnitedHealth Group and as the Director of Competitive Intelligence for United Healthcare.

April holds an MA in Public Health from the University of Minnesota and a BA from Beloit College.

Today it’s common to see images of Kiva robots running Amazon fulfillment centers, or robots in an automobile assembly line. We’ve come to accept that robots and automation co-exist with human beings. We’ve also seen advances in technology, image and voice recognition, computing capabilities and artificial intelligence that have propelled new applications in our everyday lives.

In the pharmaceutical industry, companies are still grappling with how to leverage AI to help drive commercial excellence, but it’s increasingly clear that, in pharma and throughout life sciences, AI has the power to differentiate organizational performance and can be a strategic advantage. In this session, we will cover:

• What are the examples of AI in commercial pharma today?
• What can pharma companies learn from other industries’ successes with AI?
• Where can AI make a significant impact within pharma’s commercial teams and how can teams get started?
• How to demonstrate ROI, and how to change mindsets about what AI will achieve for the organization?

We will share components of a sound AI approach and we’ll discuss how to start building, and building sustainably—because as Steve Jobs said “The Journey is the reward” and this is especially true with AI.

In this session, you will learn how a leading women’s health client:

• Leveraged emerging research techniques and frameworks to uncover “hidden” customer emotions and attitudes at different stages of the journey;
• Took patient journey design to an entirely new level by introducing the emotional layer; and
• Informed physician marketing strategies with “voice of the patient” insights.

Today’s bio-pharmaceutical industry increasingly focuses on areas of very high unmet medical need. Compared with prior decades’ blockbusters targeting common conditions, today’s brands often serve smaller patient populations suffering from relatively rare and even ultra-rare or orphan diseases. In many of these less prevalent conditions, no universally-employed ICD 9 or ICD 10 codes exist to identify known patients. These realities can frustrate efforts to identify and target the physicians treating these populations, thus “stacking the chips against you”.

This presentation will describe an innovative effort to find previously unidentified physicians treating Pulmonary Arterial Hypertension, a rare condition where early diagnosis and treatment prolongs and improves lives. It is know that there are still too many patients experiencing delayed diagnosis, leading to avoidable morbidity and mortality. This presentation will show how to use a novel methodology to find previously unidentified physicians treating PAH, to expand educational outreach efforts and help more patients get faster diagnosis and treatment. The novel methodology described below represents just such as case.

As an alternative to informatics based on known PAH patients, this effort began with known PAH-associated physicians in four related tiers. These high value targets (HVT) were mapped to physician level information in the Symphony Integrated Dataverse (IDV), as well as patient attributes inherited as patient volume that flowed through their practices over time. This novel longitudinal hybrid data structure represented the “physician journey”, through the lens of inherited patient population characteristics. Approximately 8,000 variables spanning diagnostic, prescribing, and procedure data silos were summarized under each physician, and modeled by ensembles of learning agents robust to this high-dimensional data. Following model validation, every physician in the Symphony IDV received a score for similarity to the ideal target cohorts provided. Where physicians demonstrated practice patterns highly similar to the ideal, they were defined as new HVT candidates, and their profiles were then examined to help describe key drivers of this status, and to triage them for follow up. This effort resulted in the identification of thousands of new HVT candidates, providing the opportunity to efficiently expand promotion and education efforts.

HCV is a serious infectious disease and is considerably under-diagnosed. There are several new promising HCV drugs on the market today. By diagnosing those infected with HCV earlier, patients with HCV can be treated more effectively. For pharma companies with products for under-diagnosed populations, early detection algorithms can lead to substantial increases in the addressible market.

Methods: Using the digital footprint of patients diagnosed with HCV and comparing them with matched controls, we constructed a predictive model to predict which undiagnosed patients likely have HCV. Machine learning methods included gradient boosting and random forests. Analysis was based on longitudinal prescription and claims data covering 200+ million patients in the US.

Results: The gradient boosting method was the most effective at finding undiagnosed HCV patients with a Precision considerably in excess of 90%. Undiagnosed HCV patients were more likely to be male, slightly younger, and abuse IV drugs compared to HCV patients who did not become diagnosed with HCV. Predictors involving the treatment of pain were the best at differentiating those subsequently diagnosed with HCV from those without.

Discussion: Several key patterns of HCV profiles emerged from the predictive model, namely young patients with IV drug use and pain treatment. The outputs of a predictive model can be used to inform a variety of market-shaping activities, including disease awareness, clinical support tools and focused targeting of healthcare providers by field teams that likely take care of patients with undiagnosed HCV.

Which physicians to go after is an age-old problem that is patently critical when preparing for a new product launch. Going after the wrong physicians is costly not only because it's a waste of time and money but also because of the opportunity cost of not going after the right physicians. What makes the problem particularly challenging is the fact that good targets behave just like bad ones in the beginning. They do not respond to the first few promotional interventions. Of course, good targets start responding past the critical mass but before that it is easy to confuse the two. Of course, there is a whole array of approaches at our disposal ranging from product adoption to product/market segmentation. They work fairly well but things could be better to judge by the long and protracted course correction we need to go through before we hit our stride.

Machine Learning has seen spectacular success lately, across virtually all verticals. There is not a day that goes by that someone somewhere does not hail another victory for AI. The tasks are as diverse as they are intriguing: Go, lip reading, cancer detection in tissue slides, new recipe creation, speech transcription, sports articles writing, cucumber sorting, emotion detection, orchestra conducting, and the list goes on. That's because not only is there an abundance of data and computing power, it is also relatively easy to deploy Machine Learning. Thanks to the large Machine Learning community out there, free SDKs and tools such as TensorFlow, Keras, Python, and the like are just a click away.

Clearly, we should be able to leverage some of the recent advances in Machine Learning to do a better job better targeting physicians.

Customer segmentation is an important part of pharma teams’ toolkits for connecting with patients effectively. However, too often the full potential of customer segmentation is not realized, due in part to extensive reliance on attitudinal surveys with small sample sizes and difficulty in bridging the gap between what patients say and what they actually do. In this presentation, Crossix will detail a case study illustrating how behavioral data—both real-world health data and consumer data—can be used to power actionable closed-loop segmentation solutions, including the following topics:

• Building customer segmentation using behavioral health and consumer data
• Enhancing attitudinal segmentation by linking segments to behavioral data
• Targeting customer segments across media and communications channels
• Measuring program effectiveness at the segment level and optimizing based on empirical evidence of program performance by segment

Non-personal promotion to Healthcare Professionals (HCPs) has grown over the past decade as a complement to decreasing sales force sizes. A core component of this trend is digital media advertising. Such digital advertising has expanded beyond traditional physician websites, and now marketers can reach HCPs across a broad variety of display, mobile, and video platforms, via both their professional and personal devices.

While reaching these audiences has become easier, the measurement capabilities have not kept up, and challenges remain, including:

• Inability to capture HCPs across multiple devices
• Lack of standards across endemic publisher reports that are measured separately
• Infrequent, delayed analyses beyond the end of the campaign
• Unavailability of media granularity

In response to these challenges, Crossix developed DIFA HCP™, a cloud-based analytics and optimization interface specifically designed to validate that campaigns are reaching target HCPs across the full digital landscape and driving Rx impact among those HCPs’ respective patients. Based on its experience measuring HCP campaigns through its technology, Crossix will present a recent campaign case study that measures the reach of specific publishers, the behavior of HCPs’ patients, the impact on HCP prescribing patterns, and how to best optimize investments in broad-reach health portals.

Disease severity is often defined by test scores (e.g. – Stroke, certain respiratory disorders), complex treatment algorithms (NHBLI guidelines for Asthma), or mortality scores. Identifying disease severity using secondary data is often challenging due to lack of necessary information such as test results, specific biomarker information while at the same time it is increasingly important to understand market dynamics for different patient sub segments.

Companies are often limited to gathering insights through primary market research and/or registry data where available. These methods can prove to be very expensive, time intensive and challenging to collect information.

In this session we will describe how to identify patient severity in secondary data sources in an efficient and repeatable way via different case studies.

Account Managers generally call on key decision makers at the health systems / IDNs mostly to influence policies and formulary decisions. This also improves the selling environment for the “regular” sales reps calling on the various target accounts or physicians that are influenced by that health system.

The presentation will focus on the methodology used to measure the interaction effect between AM and sales reps, and use that to size the two teams. We will discuss:

• How to quantify response of an account i.e. sales that can be generated by an account based on the direct (or indirect through AM) promotion to that account
• How to measure response in presence of interaction effects between AM and regular reps
• How to size two teams simultaneously especially when the interaction between the teams will result in continuous change of response and hence, optimal for a target

The PCSK-9 Inhibitor class of cholesterol-reducing drugs has faced unprecedented payer scrutiny resulting in only about one in five prescriptions approved within 24 hours and about one-third of prescriptions that were ultimately approved, never picked up. Our hypothesis is that the information contained in the characteristics and complete medical histories of those patients that did successfully obtain payer reimbursement for a PCSK9 therapy can be identified using statistical feature identification methods and leveraged to size the addressable or “reimbursable” market, identify the geographies and possibly physicians seeing these patients that look reimbursable and help identify the resources required to get past prior authorizations, rejections and possible reversals.

The ability to perform the computationally sophisticated analytics necessary to identify patients that look just like those who are on a PCSK-9 Inhibitor, but have not yet received a prescription, is only recently possible due to the general availability of open and closed patient claims data, which include robust lab results, and the computation power afforded by cloud based analytic platforms like AWS Redshift, Google BigQuery, Apache Spark and machine learning libraries, such as H2O, that can be efficiently deployed across multiple CPUs and computational clusters.

This presentation will describe one such approach, called Look-Alike Analysis, and how we developed a predictive model on a closed claims database, which includes all interactions for a representative subset of patients, and applied it to an open claims database, which has some data missing not at random, but covers many more lives and geographic details. The results can be used for many traditional commercial activities including market sizing, forecasting, segmentation and targeting.

Firms in the bio/pharmaceutical industry are increasingly seeking innovative strategies that result in higher efficiencies, improvements in returns to existing investments and synergies in operations and product offerings. Envisioning multiple pipeline candidates and in-line brands as part of a meaningful, synergistic portfolio rather than stand-alone, siloed products holds multiple benefits to a global firm and its customers.

With a portfolio strategy, expensive developmental programs can be streamlined and interlinked. Critical go/no-go decisions and resource allocations can be made on the basis of rational, multi-product criteria. Cross-selling complimentary products can result in higher returns to marketing and sales force investments. Customers can benefit by purchase terms and risks spread over multiple products from the same firm. Partnerships with erstwhile competitors can be forged on the basis of licensing arrangements that bolster a portfolio.

Using a case study, this presentation highlights the value of cutting edge market research and marketing science in helping a firm develop an effective cardiovascular product portfolio. Creative uses of advances in marketing surveys, database compilation and marketing modeling methodologies facilitate multi-product, portfolio based, multi-objective market modeling, understanding, prediction

The goal of this research is to develop predictive models for identifying physicians with patients who have high potentials to be treated soon for drug therapy or change line of drug therapy, or physicians with patients who have been treated already for certain indications. Patient’s activities, including supportive drugs, lab visits, symptoms, co-morbidities, doctors’ diagnosis visits, and their frequency and timing are used as key predictors. Data is collected through multiple channels. Both traditional methods and innovated predicting models are developed and tested. Traditional logistic regression is built as a base line model. Random forest and deep learning are developed and compared with the base line model. Recent breakthroughs in deep neural networks have demonstrated superior model performance, especially in scenarios where data has large volume, high dimensionality, missing records, and complex data structure.

The pharmaceutical marketing landscape has changed significantly by the emergence of several types of “health systems” (e.g., ACOs, IDNs, IHNs, etc.), and there are several physicians, clinics, and hospitals affiliated to these organizations. Most companies have well established methodologies to calculate the potential of individual accounts, but since the potential overlaps within different accounts, it is difficult to measure potential at a health system level.

With decisionmakers changing, it is becoming increasingly important to assign optimal targeting effort not only at the account level, but also at the health system level. Hence, for effective planning purposes a pharmaceutical company needs to understand patient potential at a system level.

Axtria recently developed a methodology that estimates the potential at the health system level by de-duplicating potential at accounts that are affiliated to the health system. This poster presentation will discuss in detail the methodology that was used to de-duplicate and convert account level potential to health system level potential (assuming account level potential is available).

A winning promotion strategy calls for effective segmentation and targeting in order to deliver the right message to the right audience at the right time. This underscores the importance of understanding customers from both attitudinal and behavioral aspects.

However, leveraging the available attitudinal and behavioral data to achieve an optimal segmentation solution is still a significant challenge. A common practice in the industry is to drive segmentation entirely on one data dimension, or at best, primarily on one dimension with partial guidance from the second, resulting in segments with suboptimal differentiation and actionability on the second data dimension.

As pharma brands strive for holistic promotion strategy, there is a great need to develop better methods to analyze both data in a truly integrated manner, to reliably identify the best segmentation drivers and produce optimal solution.

Here we provide an innovative and cost effective approach in which we integrate and analyze attitudinal data from available PMR/Tracking studies (Such as ATU, Demand etc.) and behavioral (including environmental) data comprehensively to obtain superior segmentation solutions. It leverages IQVIA’s vast real-world evidence data, including longitudinal patient, HCP, IDN and payer data, as well as attitudinal data from various primary market research studies. Machine learning methods and predictive modeling are applied to perform segmentation and target identification. The final segments show great differentiation on both data dimensions and uncover insightful behavior patterns and their attitudinal drivers. The results are highly actionable for segment membership assignment for the entire customer universe, target prioritization, resource allocation and message development, laying solid foundation for a successful promotion strategy.

In this applied research project, Crossix has leveraged extensive connected data sources to measure the reach overlap, Rx attribution, and ROI of a multi-channel DTC campaign combining television, digital display, and print advertising. Crossix directly calculated the overlap in reach between these channels, and measured the new patient starts from each channel alone and via the various combinations of channels. This ultimately leads to an ROI by channel, both independently and jointly.

This poster describes the methodology by which Crossix builds propensity models using the combination of health data and consumer data to predict the likelihood that a population will exhibit target health criteria, thereby presenting a solution to the problem of reaching healthcare audiences while remaining HIPAA-compliant and privacy safe. Its particular focus will be on the advantages found in using multiple transactional datasets to focus in on audiences who treat for a condition which has both prescription and over-the-counter therapeutic options.

In a world where Google customizes ads based on our web activity in real-time and Amazon makes purchase suggestions in real-time, Pharma’s targeting approach seems archaic. Dynamic Targeting aims to create weekly plans that not only tell reps who to target each week but also why and how, by leveraging a wide variety of data sources, predictive analytics, and machine learning. The solution is tightly integrated with Veeva and provides HQ tight controls to ensure that the field execution is in line with brand strategies.

Key Benefits:

• Go Beyond traditional inputs for targeting / call planning
• Maximize opportunity via agility and timely insights
• Simplify processes and communications for the field

How do we know if we are serving our customers the right number of emails, and delivering the right message in the right order? Join us as we dive into some of the challenges marketers face in our rapidly evolving direct marketing ecosystem. Emails are getting lost in the clutter and customers are oversaturated with content, becoming annoyed, opting-out, and disengaging. We see that, while email programs are showing positive ROI, open rates are decreasing, inbox placement is spotty, and clicks are virtually non-existent. Merkle’s PMSA poster presentation will introduce an innovative new approach for your portfolio email strategy, which will ensure optimal content engagement, minimal opt-out, and positive lift.

In the Life Science industry, demand studies are regarded as one of the most risky types of market research to conduct. Prototypically customers’ estimates of future utilization are too high to be believable. In turn, the results of such studies are difficult to socialize and may even be dismissed outright. Despite the consistent problem of demand study invalidity, we find little evidence that the industry has systematically attempted to examine the ways in which varying the design of studies influences the quality of the demand estimates that are returned. This interactive poster provides a critical, meta-analytic and theoretical look at four aspects of demand design framed with four basic questions.

• Do Demand Study Participants Think the We Expect Them To? Most demand study designs are predicated on the assumption that customers make relatively rational or optimal choices and evaluations of the products we test. This assumption warrants critical examination, and has implications for various aspects of study design.
• Are Accurate Demand Estimates Reliably Attainable? As an industry, we have paid minimal attention to the demand exercises we use to capture our most important variables. Typically, we rely on simplistic dependent variables that are suboptimal because they consistently result in predictably overstated numbers that overstate peak utilization. We explore how alternate approaches influence results and predictive validity.
• When it Comes to Conjoints, How Complex is Too Complex? We see little systematic examination of the assumptions that underlie widely-used experimental exercises used in demand studies (e.g., conjoint, DCM). The decision science literature strongly suggests that that people do not make choice in ways that are consistent with these design assumptions.
• Do the Contents of the Product Profile Matter? Finally, we know from research in cognitive science that how you present information can be just as important as the information itself. We examine how different styles of TPP (Target Product Profile) presentation influence demand study results.

This poster explores all of these issues from the standpoint of modern decision theory and with extensive meta-analytic evidence of hundreds of prior demand studies. We present both summary rules for how different design decisions will systematically influence demand estimate quality, and (where possible) offer specific recommendations for improving demand study designs. Pros and cons of design tradeoffs are also examined so that researchers can make sensible decisions about specific features of their studies. Visitors to this poster will have the ability to watch deep-dive video summaries of all four topical areas, and should feel free to explore the entire content overview of to focus on topics most relevant to their concerns.

Pharmaceutical companies are often faced with the tricky problem of identifying avenues to accelerate sales growth for mature products. In our experience, efforts to address this problem are somewhat narrowly focused on one of two areas: first, understanding and addressing the provider, patient or payer barriers; or second, diving deeper into field force interaction effectiveness to improve customer experience. To unlock the true potential of mature products, pharmaceutical companies will need to bring in a “surround sound” perspective, drawing learnings from different customer stakeholders, the field force, and the headquarters. We also believe that an introspective approach where pharmaceutical companies first identify their good and great customers, and subsequently crystallize the reasons for differences in product adoption across customers, is a stepping stone for accelerated sales performance. Such an approach is invariably data-driven and requires harnessing the collective power of sales and claims analytics, voice of customer from primary research, and field force profiling of good and great customers. In this session, we will present our framework for the good-to-great approach to accelerate sales growth, and share a case study where we applied this framework to a mature product for the treatment of a respiratory disease.

Our goal was to provide an overview of current published perceptions of the future of outcomes-based contracts (OBCs) between payers and providers in healthcare management, which offer a method of cost-containment and differentiation in the market to payers, while allowing them to provide optimal care to patients/customers. Structured searches identified 11 publications/white papers/presentations on the topic of OBCs. The included sources describe key hurdles of OBC implementation (e.g. data collection) and discuss possible solutions (e.g. test cases). Unless the hurdles which impair the effective implementation of OBCs are addressed, they will not deliver on their promise and will remain a niche solution in healthcare.

Medicare Part A and B claims for 50+ million beneficiaries were used to map health systems at the market level by extracting HCP affiliations to both Health Systems and Medical Groups. CareSet then assessed the amount of control that health systems have on prescriber behavior according to 8 factors, including the rate of new treatment adoption, use of EHRs, employment, etc.

Once data tables and weighted-directed graph data is mapped into an RDB (relational database) and graph database, attributes are visualized to depict geographic partnerships and variables tied to local healthcare systems dynamics. To validate key features, ANOVA and other methods are used to test where group affiliation matters, the impact on physician behaviors and patient outcomes.

A key finding is a correlation measure associating group affiliation to the rate of new treatment adoption by the group’s members. The study provides several examples based on prescriptions for specialized drugs billed as medical benefits. In conclusion, these volumes of open data sets can shorten knowledge capture rate necessary to benefit sales, marketing, training, and HEOR teams. Furthermore, this method can identify where protocol-driven testing, such as RCTs, or outcomes-based contracts have better chances for efficient execution.

Life Sciences companies spend millions of dollars on incentive compensation. Sales contests are a key part of incentive compensation and are used to motivate sales representatives during key market events, brand launches, expanding breadth, and depth of prescribing, etc. If managed properly, the sales contests can generate significant gains in terms of incremental sales revenue. Unlike non-personal channel investments, most companies don’t have a systematic mechanism of measuring the effectiveness of sales contests. A comprehensive assessment framework incorporates financial ROI, change in rep activity/behavior as well as measurement of field force motivation and sentiment. Key considerations account for strategic timing, long term vs short term impacts and controlling for other channels/market events. The ultimate objective of the assessment is to help brand and sales leadership make informed decisions about designing and rolling out future sales contests.

Targeting refers to segmentation of HCPs/accounts in the context of the opportunity they represent for a given brand. This can range from a simple to a complex process and rely on secondary data as well as primary market research. In many cases, current targeting approaches used in pharma may miss significant opportunities to distinguish between ‘Prospects’ from ‘Suspects.’ As once expressed by Lester Wunderman (considered the “Father of Direct Marketing”), ‘Prospects’ are ready, willing, and able to buy whereas ‘Suspects’ are merely eligible to do so. In this presentation, we will focus on 3 main areas for optimal targeting and highlight 2 to 3 key elements of each. This will help us differentiate ‘Prospects’ from ‘Suspects,’ and lead to a significantly improved ROI in one of the largest portions of M&S spend.

R&D departments of pharmaceutical manufacturers and clinical research organizations have historically been lagging behind in the use of real word patient data to support clinical development. This is changing rapidly and multi-center trials now often rely on globally available patient, prescription, and hospital discharge data to select appropriate countries, identify sites, and accelerate trial recruitment. Moreover, these data sets are increasingly leveraged for early clinical development to refine target product profiles, clinical development plans, and study protocols.

The development of a universal Patient identifier and integration of clinically rich data with traditional claims data now enables new insight generation and utility within the Life Sciences industry. The potential of this powerful, patient-centric data creates limitless opportunities to analyze markets and strategically deploy resources:

• Optimize ongoing clinical research and clinical trial
• Right-Size target patient populations
• Enhance patient journey and buying process development
• Drive disease/diagnostic education
• Optimize commercial strategy and execution

Problem Statement (Case Study): This case study focused on the 5th pillar: commercial execution of real-time triggers. When early treatment is paramount to treatment success, waiting for a confirmatory diagnosis claim to appear in longitudinal patient medical claims may be too late for intervention. In such cases, reaching a treating HCP while he/she is exploring possible diagnoses is critical. This is particularly impactful within rare diseases where these early signals (lab results) are still relatively infrequent (thus resulting in a very manageable / highly actionable volume of information). Two approaches were used:

• Utilizing a universal patient ID to integrate clinically rich lab results with traditional claims data to identify treating HCP.
• Coupling longitudinal claims data (possibly with confirmatory labs as well) with machine learning enables development of algorithms to identify “high likelihood” undiagnosed patients.