Chapter 6: Health Platform Businesses – From Investigations to Platforms

Healthcare is a sector struggling to reduce the cost of care while improving health outcomes, improving providers’ productivity, and meeting patients’ expectations. This quadruple aim has been relevant for many years and remains unattainable. The consequences of COVID-19, plus the politically, socially, and economically complex environment in Europe, are putting additional pressure on already fragile health systems.

Platform business models have gained significant traction in recent years, bringing scale, affordability, and other advantages that could benefit the healthcare sector as well. However, these new models are present more frequently in certain industries such as e-commerce (e.g., eBay, Craigslist, and Alibaba), travel (e.g., Airbnb and Uber), and communication (e.g., Zoom), to name but a few. They have shaped consumer expectations (e.g., same-day delivery) and triggered traditional businesses to rethink their operating model (e.g., production is done outside of the company via a platform business model). However, platform businesses have had limited traction in healthcare, and it is difficult to identify flagship examples as they exist in other industries.

In this chapter, we will first define what a platform business in healthcare is. We explore how the current state of the art in platform business theory applies to healthcare, including aspects such as open architecture, rules of governance, and core interactions. Subsequently, we test this theory with field observation done around the 12 BigMedilytics study projects (the pilots), leading to a set of requirements to successfully move from a set of innovative products to platform businesses. Finally, considering the theory presented and practical observations, we discuss the potential for a health platform business to be a solution for the quadruple aim in healthcare, namely (1) to reduce the cost of care while (2) improving health outcomes, (3) increasing provider productivity, and (4) meeting patient expectations.

Platform business models have gained significant traction in recent years, bringing scale, affordability, and other advantages that the healthcare sector has yet to benefit from. In Europe, there are more than 200 companies operating as platforms across 28 countries [1], mostly from France and the UK, and competing against more well-known and established US players. They facilitate transportation (goods and people), offline and online services; employ all skill levels; range from non-profit to for profit (commissioning model, membership fee, flat rate, etc.); vary in size – as small as under e1 million turnover and 10,000 clients and service providers, to as large as e100 million or more, respectively, 1 million clients and service providers. These platforms have shaped consumer expectations (e.g., same-day delivery) and triggered traditional businesses to rethink their operating models. These models are more prevalent in certain industries such as e-commerce (e.g., eBay, Craigslist, and Alibaba), travel (e.g., Airbnb and Uber), and communication (e.g., Zoom), and less so in healthcare, which remains yet to be disrupted by platform models, with fewer representative examples limited geographically and in scope.

Let us first see what a platform is and then focus on health. There is extensive research available on platform models, and in general terms, a platform is an (1) open architecture, [2] with a (2) governance model (that sets the rules of the game – for participation, monetization, sanctions, etc.), facilitating (3) core interactions between parties.

Platforms are typically designed around a core interaction, where participants are put in contact by a matching algorithm to exchange a value unit, [3] the equivalent of a product (be it tangible or intangible) for a traditional business. Typically, a product is the result of a pre-defined set of ingredients, combined by a company’s processes (often its intellectual property), before it reaches the end client for consumption. The value unit of a platform is the result of interactions between producers and consumers enabled by the platform.

With the core interaction clearly defined, the platform owner designs what is called an open architecture. This is a technical piece of work that is proprietary to the platform owner. Built in-house or with partners, the architecture enables the exchange of information, goods, services, or currency. The platform has built-in functions to pull in both sides of the platform (value producers and value consumers), to match them at the right time with the right value, and to facilitate such exchange. It is modular, meaning it is made of independent parts designed to function as a whole, with visible design rules and hidden design parameters [4]. It has a stable core and highly customizable and diverse features, allowing for personalization and matching in many more ways than a traditional business can [5]. This is also possible thanks to data and advanced analytics, which cannot be dissociated from the platform, being fully embedded in the way it operates. For example, real-time voting, ranking, and feedback loops give information about the performance of current products and inform decisions about future ones, creating an intrinsic value for the platform itself.

A platform would not be complete if, in addition to the open architecture and the core interactions, there were no governance models. These are the laws (European, national, platform-specific, and so forth), norms, architecture, and markets that ensure the appropriate distribution of value between producers, [3] consumers, and platform owner and incentivize future interactions. These are also meant to discourage or penalize misbehavior and create a general environment of trust, as a platform creates value with resources it does not typically own or control. This phenomenon was phrased as the inverted firm, [3] where production does not happen in the company but outside through the company’s partners. For example, Uber does not own the cars, but the drivers (producers of value) do. Through the governance rules, the platform locks in value exchange, making it more attractive to do business on the platform than outside.

After this succinct review of platform definition and its three components, we move on to understand its application in a domain such as healthcare, considering that such an industry that relies on information, generates massive amounts of data, and is highly fragmented is said to be ready to be disrupted by platform models [3]. Despite extensive research, [6] it is complex to apply it to healthcare given the variety of stakeholders involved with misaligned incentives, the sensitive nature of health data, the regulatory framework, unequal technological adoption, etc. The adoption speed of a platform model has been slow in healthcare, yet it is believed to become the “new normal” [7].

To start with, there are multiple interpretations of what a health platform is. For example, there are platform-enabled ecosystems that leverage technology to connect an existing portfolio of partners and their respective goods and services [8]. Building such an asset is seen to be a strategic choice comparable with Mergers and Acquisitions, yet less risky because it is less capital intensive, but it requires pre-existing core operations, a mature customer base, and enough partners available. There are with health information exchange system where parties exchange services, information, and other resources to create smart and sustainable healthcare ecosystems [9]. For the purpose of our analysis, we have simply considered the platform definition presented previously as an open architecture with rules of governance designed to drive interactions within the health domain.

First, as platforms are designed around a core interaction, should this be related to health delivery, health financing, health enablement, or something else? The first is provided by medical professionals with various degrees of specialization (e.g., medical assistants, doctors, nurses, nutritionists, etc.) engaging with patients in different care settings (e.g., primary care office, ambulatory care clinic, digitally, etc.). The second is delivered by the government, insurers, employers, or individuals themselves, who pay fully or partially for the care delivered. Related to the first and second are numerous adjacent interactions for the enablement of care, such as appointment booking, drug delivery, data sharing, etc.

Second, the architecture designed to be open in platform terms to facilitate interaction is generally closed in healthcare (except perhaps for research purposes). This is driven by the national and international regulatory framework from the General Data Protection Regulation (GDPR) and is a consequence of technological planned or unplanned obsolescence, causing significant interoperability issues. For example, a hospital cannot share data with another hospital about a patient because they are using different providers or versions of electronic medical record (EMR) systems. The Social Security institutions of European countries hold the largest volume of health data, and legitimately so. They receive data from medical providers responsible for care and dispatch it to other players for payment purposes, research, etc. Medical providers, be they a solo practice or hospital group, have a view limited to their own patient interactions, which can be limiting, particularly when working with patients with multiple or complex conditions.

Last but not least, governance is mandatory in health and for health platform models, for example, to define the roles and responsibilities of all parties, the operating terms and conditions, the rules of funding, the alignment of incentives, and to penalize in case of misbehavior, malfunctioning, etc. In a traditional health setting, the doctor has the responsibility of ensuring the right diagnosis and treatment. The payment of care is done based on value, or most frequently, on a fee for service defined by local governments and funded by taxes in most European countries. In health platform models, the rules of governance for care delivery and care funding ought to be given the highest attention, building on current best practices and innovating to overcome challenges regarding inefficiency, [10] workforce shortage, health inequalities, [11] or reimbursement, [12] to name but a few.

The BigMedilytics initiative encompassed 12 study projects spread across three themes – Population Health & Chronic Disease Management, Oncology, and the Industrialization of Healthcare Services – with a common goal to prove the positive impact of data analytics on healthcare systems in Europe [13].

The five studies in the Population Health & Chronic Disease Management section aim at demonstrating that big data analytics can reduce the burden on the secondary care institutions through a better triage and orientation of patients to the most appropriate place of care. These studies focus on specific conditions that drive mortality and morbidity in Europe – kidney, diabetes, chronic obstructive pulmonary disease, asthma, and heart failure – and on populations with comorbidities. The three oncology studies aimed at demonstrating how analytics can enable better and personalized treatment with lower complication rates for patients and higher productivity rates for the medical practitioners. With the focus on breast, prostate, and lung cancers, they represent the most frequent cancer types and involve the highest burden in Europe. The four healthcare service industrialization studies aimed at leveraging analytics to improve hospital workflows, critical for productivity (radiology; asset management), and quality of care (sepsis management; stroke management). Studies targeted specific customers, had different value propositions, and required tailored activities, resources, or partners. The geographical coverage, clinical focus, or maturity stage of the big data innovation differed as well, with some being in the minimum product development phase while others were preparing a product for commercialization.

To understand the potential of becoming health platforms, the following analysis was carried out on the core components of a platform: architecture, participants, core interaction, and governance.

First, it is important to identify who the participants are who have the potential to become platform owners? In the BigMedilytics studies, participants were representative of the healthcare sector: medical professionals, medical institutions, health techs, patients, etc. Each study had a lead organization in charge of the overall coordination between study partners and who closely managed the timeline, scope, and budget for its own organization. It brought key capabilities, such as data, technology, or clinical operations, necessary for the study to achieve its goals. In platform theory, this lead organization might be a platform owner, value producer, or consumer of value. This central role in the BigMedilytics initiative does not automatically imply that it is best suited for platform owners.

Research and medical institutions (Incliva VLC Biomedical Research Institute; Charité Medical University; University of Southampton; Erasmus MC Medical University; Karolinska University Hospital; National Centre of Scientific Research Demokritos; ETZ Elisabeth-TweeSteden Ziekenhuis Medical Center) come with the advantage of innovative, modular technology, applicable to diverse sectors, and a commercially driven strategy. They are by nature institutions operating at the inter section of science and healthcare, relying to some degree on legacy technology, and trying to balance research with the delivery of care.

More often than not, the platform owner comes with the technology that enables interactions. In the case of BigMedilytics, although all partners have some degree of technology, yet best equipped to be platform owners would be pure technology companies (e.g., Huawei, IBM, Contextflow, Philips).

This being said, more complex set-ups, including dedicated spin-offs, could overcome some of the above-mentioned challenges. In the end, the purpose of a health platform model – for public good, commercial good, or both – will strongly influence, in the case of healthcare, who the platform owner, producer, and consumers are.

Second, can there be a single unifying core interaction for all BigMedilytics studies that could justify the convergence of all studies, leading to a greater impact? Or can an individual study evolve the product around a core interaction? By analyzing the studies, we have identified three main interactions:

• Patients and medical providers exchange information for the purpose of managing the individual’s health and ensuring a high quality of care.

• Hospitals’ mobile asset providers (or a dedicated service) share data that medical providers use to find and manage assets more efficiently.

• Algorithms generate insights that medical providers consume to help reduce or improve diagnosis and treatment.

  Hence, an all-encompassing core interaction could be framed as an exchange of information between people and physical assets in quasi real time for the purpose of faster and better diagnosis and treatment and the remote management of people’s health and assets.

Third, what does an open architecture mean in healthcare? The fact that healthcare is highly regulated can seem to be in contradiction with having an open architecture, like the Apple Store has for app developers, Airbnb for house owners, or Craigslist for anyone. Appointment booking solutions have platform characteristics, where medical professionals create an account, pay a monthly fee, and manage their appointments online, while patients search for medical professionals in many specialties nationwide for in-person and online consultations. The more practitioners on the platform, the more patients will use it, and in turn, this will attract more practitioners, generating what is known in platform businesses as the network effect [14].

Interestingly, those BigMedilytics studies focused on location characteristics that might readily evolve into a platform. In fact, the underlying architecture is open enough to include new and more interactions. For instance, if patients or medical professionals are connected in addition to mobile assets, this could provide an opportunity for interactions, both medical and non-medical. For example, a patient in the hospital can see doctors’ schedules for meetings during the day; similarly, the patient can be visible in real time to doctors while moving between examination rooms. If the platform is available across hospitals, it could also help to keep track of patients transferred from one to the other easily. For these examples to materialize, data privacy and security measures need to be built into the architecture to evolve and adapt in tandem.

The BigMedilytics studies focusing on a particular disease could further explore the possibility of being the go-to platform for that particular disease for patients, medical providers, funders, and others. For example, a solution for patients with a chronic condition is to interact with all the relevant medical personnel (e.g., doctor, nurse, midwife, etc.) and non-medical experts (e.g., nutritionist, physical exercise trainer) to help better manage the condition. This could have higher relevance for patients with comorbidities, who are often left to navigate between different doctors and miss a coordinated approach. In these examples, in addition to data privacy considerations, the incentivization mechanism and the staffing model ought to be analyzed, so the solution is an integral part of care delivery, enhancing it rather than operating alongside it.

Fourth, what are the rules of governance to be used in a health platform model? Laws, norms, architecture, and markets are meant to ensure transparency and instill trust in the platform, reward the good, disincentivize the bad, enable feedback, etc. The purpose of a platform – for example, commercial, public interest, and so forth – is a determining factor for the nature of the rules of governance. In addition to the architecture, the rules of governance dictate how open or closed (and for what) a platform is. The roles and responsibilities of participants – owner, value producer, and consumer – will be covered by these rules both for business as usual as well as for unexpected situations (e.g., platform is down, unable to provide a service, malfunction, and misbehavior). For the 12 studies, the foundation for governance was represented by the consortium’s partnership agreement, where roles and responsibilities were clearly defined for each partner. Additionally, European and national data privacy regulations on the management of health data were referenced through study implementation.

The BigMedilytics studies began before the COVID-19 pandemic. This unprecedented event made the quadruple aim even more challenging to attain. The cost of care increased considerably across Europe, by 6.3% in Germany, 3.9% in France, and 15.7% in the UK, mainly due to the acquisition costs of masks and tests and bonuses for the health workforce [15]. Accelerated development of mental health issues, in particular among the young and the poor, deferred diagnosis and treatment, and waiting times as long as 3 months for an appointment impacted the health outcome [16,17]. The productivity of the health workforce has been negatively impacted by the shortage of skilled personnel, who were unevenly distributed around territories and whose mental health suffered as well because of the COVID19 pandemic. Does a health platform model have the potential to attain the quadruple healthcare aim: (1) to reduce cost of care, while (2) improving health outcomes,(3)increasing provider productivity, and (4) meeting patient expectations under these circumstances?

First, as per the cost of care across all European countries, hospital services, followed by outpatient services, represent the largest government expenditure on health as a percentage of gross domestic product [18]. In-patient services – those that require the patient to stay in the facility – are expensive and promise a higher success rate though disruptive to the patient’s daily life. Outpatient services – those which can be provided at home or in non-hospital settings – are less expensive and less intrusive to the patient’s everyday life. Thus, to address the first quadruple aim – reduce cost of care – and have a large addressable market, a platform would best tackle these two service areas.

Second, health outcomes are changes in physical and mental health that result from measures of specific healthcare investments or interventions [19]. Standard measures are usually general mortality, infant mortality, and life expectancy, but they can be health-related quality of life, functional status, symptoms and symptom burden, health behaviors, or patient experience of care [20]. There are direct and indirect investments and interventions, medical and non-medical, that impact health outcomes, which makes the spectrum of possibilities for a platform very broad. For example, appointment booking solutions that enable quick access to care can count indirectly towards a positive health outcome just like a doctor’s intervention, yet the nature (i.e., frequency and magnitude) of the impact is significantly different.

Third, the productivity of the medical workforce has been part of the quadruple aim in healthcare for some time now, with more and more demands on a shrinking workforce (15 million more health workers needed by 2030 [21]), which was also highlighted during the COVID-19 pandemic. Technology aimed to automate manual processes to reduce the burden on highly skilled medical providers so they can spend more time with the patients, to remotely monitor patients with severe conditions at home, AI-enabled tools that predict people at risk for a specific condition, to name but a few examples, has yet to prove a positive impact on productivity. Healthcare workforce concerns are related to the ability to meet demand [22], adapt to new operating models, support workforce wellness, and recruit and retain new talent. Health workers and patients are directly impacted by the lack of interoperability among IT systems, so a platform would be expected to communicate with legacy systems with modern technology to reduce the barriers to usage as much as possible and positively amplify the workforce’s impact.

Lastly, meeting patient expectations is possibly the most challenging objective in healthcare. People have different experiences and degrees of health literacy, which impact how empowered they feel to take care of their own health versus being taken care of. For example, the Patient Activation Measurement is a survey that assesses an individual’s knowledge, skills, and confidence in managing their own health and healthcare [23]. The Patient Activation Measurement shows that a person with a level 1 score feels disengaged and overwhelmed, transferring full control to the medical professional to make decisions about their health. At the other end of the scale, a person with a level 4 is actively pushing further to adopt better behaviors, keeps informed, and discusses treatment with medical professionals. Moreover, patients expect to have access to care when they need it within a reasonable amount of time. For example, waiting times for common surgeries vary from less than a month to more than 8 months [17]. In the UK, in August 2022, 7 million people were waiting for treatment, and around 390,000 had been waiting for over a year (which is 375 times the pre-pandemic data in July 2019) [24]. Patients expect doctors to communicate well, to be empathic and make them feel cared for and regarded, [25] and to be trustworthy, knowledgeable, and loyal [26]. Lastly, patients expect better digital experiences when scheduling and interacting with medical professionals [27].

The quadruple aim in health, namely (1) to reduce the cost of care, while (2) improving health outcomes, (3) increasing provider productivity, and (4) meeting patient expectations, is more relevant today than ever before. The gaps in cost, outcomes, productivity, and patient expectations have deepened with the COVID19 pandemic, and new and more efficient ways of delivering and paying for healthcare are expected to emerge.

In this context, the 12 BigMedilytics studies and associated innovations are very promising. Transforming them into business, and more largely to the extent of having an impact across Europe, requires – as observed in the field – significant investments and time.

From business theory, we know that platform models have the potential to alleviate some of the challenges, in particular when scaling the business. Moreover, healthcare platforms can bring a positive contribution by enabling a large number of interactions for a diverse group of people and assets in a timely, efficient, and productive manner.

This is why in this chapter, we not only explored, based on business theory, what it would take to have a successful platform in a healthcare environment, but also analyzed what it would take to evolve BigMedilytics’ studies, or as a whole, into a platform business. Beyond our study, there are considerations we did not address since we believe that they require a more intense use-case-specific analysis. The socalled network effect is most notable, but there are also regulatory, technological, cultural, and financial barriers. In any case, what has proven essential to the success of BigMedilytics and will be necessary for any healthcare platform, in particular in Europe, is and will be the cooperation between cross-sector and cross-border parties.