1Life Healthcare : The Importance of Data Integration in Primary Care

Q: Why did we decide to integrate with Apple Health and Fitbit data, rather than specific devices?

A: There are a ton of options for getting patient data. Which devices, which measurements, which platforms?

To choose, we anchored on three things: what would be familiar and easy for our patients, what would be familiar and easy for our providers, and what was well supported for our technologists. Apple Health and Fitbit ticked the boxes and then some.

What we're particularly excited about with both is that we can start simple, with a few types of data our clinicians know add value and are meaningful to care. Then, as we learn more about different measurement types, we can add them in without much additional work or explanation to our members.

This way, when we have a conversation about sharing data with a patient, the chances they have a device that we already work with is very high.

Q: What steps did you take to get the data integrated with our clinical teams? Why is that process so important and unique to One Medical?

A: The biggest step was just talking to our primary care providers. What insights aren't they getting from their patients? Where does the conversation stall out? The ability to add color and context to conversations came up over and over again.

One thing we heard from providers was a need for data to provide more insight and shed more light on the information patients share verbally. Human memory isn't great, so even when patients think they are being truthful about their health, they might be experiencing regency bias and recalling only recent events rather than those that occurred further back. For example, they may say they are sleeping well or terribly based on their most recent memories, while historic data says otherwise. Or patients might have a different idea of what a "workout" entails compared to their provider. Data can offer more detail and also make up for this subjectivity.

While we fully expected every clinician to want to do fancy analysis, they mainly expressed a need for a littlebit more insight to make the conversations more meaningful and personalized.

Q: How can data sharing improve a patient's health experience?

A: Wearables and devices already have apps that analyze your data, so you don't necessarily need a clinician to know if your weight is going down or sleep is improving. By sharing that data with a provider, however, you can frame that data within the context of other conditions and your overall health.

Apps don't know if you're anxious or on a statin, but your primary care provider does. Apps don't know that you want to lose weight to be able to get a surgery that will let you walk down the aisle at your child's wedding, but your primary care provider does. The reason for caring about a particular measurement varies from person to person.

Your provider can help you holistically look at your health habits and lifestyle (i.e. diet, medication, coaching, etc. ) that an app might not understand or cannot accurately track.

The main thing is that an app will always just be comparing your measurements to the average of the general population, or the one-size-fits-all of clinical baseline ranges, while sharing data with your primary care provider means understanding what measurements mean for you specifically.

Q: How can this data be built upon for the future and why is it helpful potentially for population health or general insights to build or inform future product updates?

A: Different things motivate different people. When people talk about personalized medicine they usually mean genetic sequencing to figure out what you need. But what if we could learn what motivators work for different people using devices?

Trying to lower your blood pressure? Maybe sleep, activity, and medication all make a difference, but each is really effective for a subset of the population. With machine learning, we could detect what interventions a patient is most responsive to and focus on that, while worrying less about what's not working.

More than that, we can check to see if our programs themselves are effective, or when people tend to need an extra push. For example, we might learn that sleep takes a few months to improve after starting an anxiety medication, but with medication and exercise, it gets better in weeks. Understanding how difficult or easy milestones can be to achieve per patient, affords us the real time ability to be there and support our patient through those most difficult moments and champion them to better health.