Do I need a tokenization engine capability?
Authors: Asheesh Shukla , Sankha Bhattacharya and Kirti Prakash
One of the questions we deal with from our clients is — do I need tokenization and hence a tokenization engine?
A key reason for this question is that there are some myths and lack of clarity about what a patient token is. There are also instances when tokenization and anonymization are used interchangeably.
In this post, we will break down when tokenization is needed and, in the process, provide clarity on what tokenization is.
Thinking of the patient dataverse
Imagine yourself as the lead of the patient dataverse within a Pharmaceutical organization. Your patient dataverse consolidates data available within your organization and from external data partners.
Think about the value of these data sources in silos and in an integrated fashion. The journey for integrated patient data across the patient dataverse inevitably leads to the question on tokenization.
Making choices for integration
If you are considering integrating data within the Type 1 archetype — within the data sets available in your organization — a robust patient MDM process can help you achieve the same. You can use the MDM techniques and stewardship processes to integrate data within these data sets and you don’t need a tokenization capability (though tokenization engines also provide this capability. The challenge for you will be to think about how to integrate the same with other Type 2 and Type 3 archetypes shown above. But more on that later
If you are looking for integration of patient data within the Type 2 archetype mentioned above, then you have the option of engaging data aggregation partners who will do this for you. Another approach is to sign agreements with each of the data providers — such as SPs or hubs — to receive compliant PHI data, integrate it and convert it to a Type 1 archetype by building relationships and compliant data sharing ecosystems.
When considering how to integrate data sets within the Type 3 archetype, one method is to work with integrated data set providers, like Healthverity or Symphony, who can provide such data sets, such as integrated claims or labs data sets.
As you can see, patient data tokenization capabilities aren’t necessary if you are considering integrating data within one of the three archetypes. There are several accepted practices and robust capabilities to buy or build that solve this problem within these archetypes in silos.
Consider tokenization capabilities if you are seeking to integrate patient data across the three archetypes.
A tokenization engine helps create unique encrypted identifiers for patients across health systems in a way that:
- Maintains patient privacy and makes it impossible to reverse engineer a token to identify a patient.
- Ensures generation of the same token across the data sets for the same patient to enable integration.
So, what’s the takeaway?
If you are considering integrating data within any of the archetypes mentioned, you probably don’t need a tokenization engine or capability. However, if you want to integrate data across the archetypes, then you need to think about implementing a tokenization engine.
In our experience, the value of patient data is leveraged far more effectively when it is integrated across the three archetypes. Integrating within the archetypes only will help, but may result in a fragmented patient experience, a lack of richer insights into the end-to-end patient journey and a data ecosystem that does not fully align with your patient-centric objectives.
This raises some new questions
Data integration raises a few questions, but one of the key follow-ups from this discussion that we have seen from our clients is how to align tokenization engines and capabilities with partners in Type 2 and Type 3 archetypes.
Let’s dive into that topic in the next blog post.
