As a leader in the pharmaceutical industry, you know that collaborating on life science and healthcare data is essential to use machine learning to its full potential.
By embracing collaborative data ecosystems, you open the door to faster drug discovery, more efficient clinical trials, and better use of real-world data to improve treatments.
With technologies like federated learning, you can work with data across partners securely, driving innovation and bringing new treatments to patients more quickly while protecting your data and model’s IP.
But collaboration on data among multiple stakeholders is hard due to the common issue of data silos. Before we dive deeper, first let’s talk about what collaborative ecosystems and data ecosystems are.
Collaborative ecosystem and data ecosystem definitions
A collaborative ecosystem is a network of organizations working together to create shared value by pooling resources, expertise, and innovation, often with a focus on solving common challenges. The collaboration typically goes beyond individual organizational goals to achieve broader, collective outcomes.
A data ecosystem, on the other hand, is a framework where different entities collect, process, and manage data, often for specific purposes like analytics or machine learning. In a data ecosystem, the focus is on training or fine-tuning an AI model or running analytics on data that is distributed across departments or organizations.
The key difference is that a collaborative ecosystem emphasizes partnerships and shared value creation across different areas (not limited to data), while a data ecosystem focuses specifically on the lifecycle and governance of data.
A collaborative data ecosystem merges both concepts, facilitating secure collaboration on data-driven projects without the need for data sharing, for example using technology like federated learning.
Specifically in the pharmaceutical industry, a collaborative data ecosystem connects pharma companies, healthcare providers, and sites, allowing them to work on shared projects like drug development and clinical trials without compromising sensitive data.
For example, the AISB Consortium brings together biopharmaceutical companies and federated learning technology to tap into the shared knowledge and data of industry experts, helping to advance AI-driven drug discovery.
Spotlight: Healthcare Data Ecosystem
In the category of "health data" alone, there are countless data sources, data formats, and various technologies, locked in silos under different data sovereignty and ownership.
Also, secure data collaboration and ecosystems aren’t about data sharing which is a distinctive concept.
Check-out the AISB webpage
Difference Between Data collaboration and Data Sharing
Data sharing involves providing access to data from one entity to another, primarily for independent use, while data collaboration requires multiple parties working together on the same or combined datasets to achieve a common goal.
The key difference lies in the level of interaction—data sharing is typically one-way, whereas data collaboration is a continuous, interactive process with shared outcomes.
Data Ecosystems Biggest Challenge: Data Silos
According to the World Economic Forum, 97% of all data remains completely unused – data is siloed and inaccessible for analytics & AI.
Organizations don’t share their data due to regulation, security and commercial sensitivity. As a result data is distributed among organizations and never gets used.
What’s more, publicly available datasets, which most of the available pharma machine learning models are trained on, aren’t diverse enough to reliably perform on other research areas that pharmas are working on.
Let’s zoom in to healthcare, pharmaceutical sector more.
Spotlight: Healthcare Data Ecosystem Challenges
In the pharmaceutical sector, one of your biggest challenges is that data sits in silos—clinical trial data, real-world data, and drug development data are often isolated across different organizations, sites, or universities.
This isolation, combined with privacy regulations, data ownership concerns, and varying security standards, makes collaboration difficult.
Moreover, data exists in different standards across multiple technologies, making data harmonization a challenge.
When you zoom out to the ecosystem level, the complexity of integrating this data becomes even more apparent.
Yet, healthcare and pharma organizations, sites, and universities hold vast amounts of complementary data that, if combined, could provide a much more complete picture.
By tapping into this previously siloed data, you could create far greater value—well beyond what any single organization can achieve alone.
Collaborative data ecosystems offer a solution by enabling you to work with other stakeholders securely, addressing issues of data ownership, quality, and privacy while unlocking the potential for discovering new drugs or speeding up the site selection process.
Collaborative Data Ecosystems: Creating Value from Data With AI
Collecting data and connectivity alone is far from enough.
You need to be able to co-create and innovate, analyze, and draw insights from data to derive any value from it.
This requires a new approach to collaboration that goes beyond the scope that someone would expect from internal projects. And it requires new capabilities that foster innovation and value creation across company boundaries.
While the path to becoming a data-driven enterprise is clear, and its adoption can be accelerated through the implementation of end-to-end data and MLOps platforms, the required capabilities of collaborative data ecosystems aren’t yet fully understood.
As soon as different participants in an ecosystem collaborate with each other, different worlds collide.
Let us have a look at an example:
Organization A has already made strong investments in AI and aligned its entire core business with it – not only in terms of technology but also regarding internal processes and expertise.
Organization B is at an earlier stage and is working to effectively capture, clean, and validate data. This organization does not look into building its own AI capabilities as it is not relevant to the core business.
There are a plethora of capability gaps that are exacerbated as more participants join the ecosystem. Closing these gaps is instrumental in determining the ecosystem’s success or failure. To bridge them organizations need to build ecosystem capabilities either in-house or by sourcing them from partners from within or outside.
| Ecosystem Capabilities | Why it matters | How it can be approached |
|---|---|---|
| Strategy | Managing an ecosystem is very different from managing an integrated company - defining the strategy and governance of a collaborative data ecosystem is a major success factor. | Orchestrators must establish and align all partners on a shared value proposition, set the rules, coordinate the activities of the other participants, aggregate their data and expertise, and deliver a range of products or services to the end customer. |
| Data | Data is the foundation of the ecosystem. To enable true data interoperability across organizations, there must be alignment on how the data is captured, contextualized, and made available in the ecosystem. | All participants within the ecosystem must be aligned on a Common Data Model, and privacy, as well as security, must be ensured during the entire data lifecycle. |
| Innovation | Data Science is highly experimental in its nature and requires an open-minded culture of co-creation and experimentation. | Unlike most traditional product or service businesses, collaborative data ecosystems should start with a focus on establishing their value proposition by innovation before putting too much emphasis on monetization. |
| Trust and Governance | The value and effectiveness of a collaborative data ecosystem grows over time. For this to happen, the ecosystem must be designed for sustainability and address all potential risk factors, which means that data and AI assets need to be protected and data compliance must be maintained with all applicable regulations. | Modular application of Privacy-enhancing Technologies, additional security and access policies that ensure compliance, ethics, traceability, and reproducibility. |
| Operating Model | Achieving success with data science and ML needs constant delivery, iteration, and collaboration on data and models across data scientists, data engineering and the business. Scaling these efforts needs a collaborative lifecycle methodology, interoperability across data and analytics tech stacks and ultimately the capability to operationalize data applications. | Extension of known methodologies: From DevOps and MLOps to “Federated MLOps”. Highly integrated platforms that support multiple data science tools, packages, and languages to derive insights from data with advanced analytics, AI and ML. |
Value in Collaborative Data Ecosystems is Created on Multiple Levels
It is worth taking a closer look at value creation in data ecosystems, as it emerges at different levels:
What It Takes For Collaborative Data Ecosystems to be Successful
Winning collaborative data ecosystems address certain attributes that differentiate them from others. A key question is how much impact the ecosystem can have on the rest of the industry, and how innovative it is.
From a technical perspective, it needs to be highly interoperable and scalable, to foster productivity and innovation.
Finally, it needs to be sustainable, and to proactively address privacy, security, sovereignty and to protect the IP of all parties.
Never in the history of data science and AI have so many technologies converged that could enable the most innovative among us to push the boundaries of what is possible today: Moving from the internal to the external, to establish truly secure and collaborative data ecosystems on a global scale.
Apheris is safely connecting distributed health and life sciences data for analytics and AI.
The Apheris Compute Gateway, connects distributed life science data via a federated computing infrastructure with additional governance, security and privacy controls, ensuring that only approved computations can be executed on the data.
During the whole process, data owners stay in physical and operational control of their data, while data and model IP stay protected and compliance requirements (such as for GDPR) are fulfilled.
If you are interested in exploring how Apheris enables your collaborative data ecosystem, contact us.
