The conversation around whether a healthcare startup should build a data platform or buy one isn’t anything new. In fact, this has been a debate since cloud-based technologies rose to popularity—we’re just now seeing more startups enter the space and opt to build their platforms internally.
Startups choose to build their data platforms for several compelling reasons. For example, it could give them greater control over their data security and architecture or they feel they know their data best and can better manage it internally. Whatever the reason, if a startup is looking to develop its own health data platform, there are five factors they must always consider.
Performance and Data Storage
The first question about data storage is, “where should we store it?” Is it better if it’s stored in the cloud? What about the old-school way of storing it on-premise with a solid-state drive? This answer can depend greatly on the type of data and amount, but for most, cloud storage is the preferred method.
Once the startup has narrowed down where they want to store the data comes the tricky part of choosing which cloud to use. Do they go with Microsoft? What about Amazon or Google? What third party can they trust with this sensitive information? Startups also need to take into account the platform’s performance. Is it going to accurately accomplish all that they need it to?
Finally, finding the ideal “shape” for data storage oftentimes takes a lot of trial and error. The way the data is organized can have dramatic effects on the performance of queries or the interoperability with external systems. It can take many iterations, with lots of time and development effort, to find the right way to store data.
My recommendation is to choose a platform that is tried and trusted. Use one that has been thinking specifically about the unique consideration for storing health data, not just any data. Startups can skip the learning phase by using a system that has already been tested with varying subsets and spectrums of data and is designed to scale for both storage and query performance, such as the Precision Health Cloud.
Security and Compliance
Managing security and compliance is one of the hardest parts of building a system involving health data. You must balance protecting patient privacy while delivering quality patient care and meeting the strict regulatory requirements set forth by the Health Insurance Portability and Accountability Act (HIPAA) and other regulations, such as the EU’s General Data Protection Regulation (GDPR). The number of data breaches affecting healthcare companies also continues to rise year over year. With this all in mind, startups must ensure their platform is continuously audited against multiple robust frameworks and regulations such as HITRUST and SOC2 and remains fully compliant at all times, which can be a lengthy and expensive effort.
In addition to the platform handling its own security, a platform for health data should be prepared to give startups the frameworks and tools to make their own compliance easier. A health data cloud can give them APIs to make implementation easier, such as consent tracking, frameworks for supporting GDPR and auditing capabilities.
Authentication and Authorization
Startups operating in the healthcare space have even greater needs when it comes to authentication and authorization.
The platform needs to be designed to handle authentication and authorization requirements that span from a hospital system, for example, to a customer-facing application. This means they need to create a very flexible system. Large enterprises like hospital systems and schools demand single sign-on and consumers want simple systems like passwordless logins and social account integration. While these may seem less of a priority for developers building the platform, it’s an important factors when getting user buy-in.
Once users are logged in, controlling access to the data also requires a high level of configurability. To make this as easy as possible, I recommend building with an attribute-based access control system. This way, developers can easily enable detailed access control policies.
Interoperability
Interoperability is something startups looking to build their own health data platform need to pay close attention to. In order to be interoperable, the platform must be able to access, exchange, integrate and cooperatively use data in a coordinated manner, within and across organizational, regional and national boundaries.
If partnering with large organizations, startups need to be mindful that the majority will want electronic medical record (EMR) integration while consumers will want integration with Apple HealthKit or Google Fit. In order to make these integrations easier, I recommend startups ensure their platform has a well-defined data storage system like Fast Healthcare Interoperability Resources (FHIR). Not only does FHIR make these integrations easier, but it also helps avoid data being trapped in the platform.
Analytics
Analytics is the fifth and final thing startups must take into account when building their own health data platform. Once they have data in their system, they’ll want to start analyzing it. How they do so again comes down to the types of partners or users they are targeting.
There are many types of people who will want to look at the data and each one will want their own type of tools. For example, doctors or coaches will want tools that can help visualize and explore data for small groups of people to find the right way to help. Whereas individuals will want to see their own data and trends over time.
As the data set gets larger and in order to provide the most value to users, startups need to have a powerful data analysis tool like Jupyter Notebooks or artificial intelligence algorithms like TensorFlow to help them understand the data.
There is a lot to consider when a startup is looking to build its own health data platform. From where to build the platform to how to analyze the data coming in, startups need to ensure each step is taken care of. As a software architect, I know firsthand what it takes to build a high-functioning and intelligent health data platform, and these are the five things I believe all startups must consider when developing their own platforms.
About Matt Lavin
Matt Lavin is a software architect at LifeOmic, a software company that leverages the cloud, machine learning and mobile devices to power precision health and wellness solutions.LinkedIn, Twitter
