Why Your Doctor Doesn’t Care What Your Smartwatch Says (and how to change it) 

By Ryan Neely, Ph.D. 

Harping on the inefficiencies of the American healthcare system is almost cliché at this point. Costs are rising, outcomes are lagging, and misaligned incentives prioritize treatment over prevention. At the same time, millions of Americans happily pay out of pocket for wearable health trackers that record their physiology 24/7. Undoubtedly, terabytes of longitudinal health data are uploaded every day that capture sleep patterns and exercise statistics for each individual user. However, equally without doubt is the magnitude of the eyeroll you are likely to get if you email your physician a screenshot of your health tracker app. At best, they will humor you and bring you into the clinic for a real diagnostic test. How do we bridge the gap between consumer-grade wearables and point-of-care diagnostics that healthcare providers rely on to make treatment decisions? 

More Data Isn’t Always Better 

Your doctor isn’t ChatGPT – feeding her greater volumes of data won’t automatically make her better at her job. Doctors are already overwhelmed with the ever-increasing amounts of data generated in healthcare systems today. Dealing with the increased data load is a leading reason cited by doctors as a barrier to adopting remote monitoring technology¹. A survey found that 92% of doctors felt interactions with electronic health records to be excessive, while 73% felt like this burden impacted patient care². While I might find it interesting to parse through my heart rate variability over the course of a typical night, this level of granularity simply isn’t meaningful or practical for a physician to review.  Continuously collected data may be a boon for training AI, but this type of data on its own only creates more work for overloaded clinical staff.  

Quality over Quantity 

If you’ve ever gone running with a watch that measures your heart rate, you may have noticed some questionable values. Optical sensors at the wrist have a hard time distinguishing between changes in light caused by your pulse and changes in light due to the motion of your body, especially if you have a bony wrist or the watch is worn loose. In the context of fitness tracking, these errors aren’t a big deal – but in the context of clinical diagnostics, they’re a nightmare. Did the patient experience tachycardia in the middle of his run? Or did the watch just lock on to his stride rate as he went down a hill? Most consumer-focused wearables aren’t subject to any standards that regulate their accuracy and performance, which can vary widely between devices, individuals, and the context of use. Additionally, wearable devices prioritize comfort by collecting signals from convenient locations like the wrist, rather than where the signals are strongest. This limits the range and quality of data they can collect (it’s hard to measure the activity of the brain or gut from a smartwatch). Doctors know this too – data quality is another primary reason for hesitance in adopting wearable monitoring tools¹.  

Bridging the Gap 

The enthusiasm for health tracking among consumers is a significant opportunity to expand preventative care and engage patients in their own health – two factors that are known to improve outcomes and lower costs. The key to unlocking these benefits is to find a way to make wearable devices play nice with the health system without losing the sleek convenience that makes them attractive to consumers. At Skribe, we believe that there are 4 key factors that are necessary to make this possible: 

1. Data quality and clinical relevance. The bedrock of any successful approach has to be the quality of data. Even the most powerful AI models are only as good as the data used to train them. For healthcare to take wearables seriously, they need to reliably measure physiological signals directly. In many cases, this probably means sensors need to be placed on the body directly over the target organ, such as the heart or brain. These signals should be meaningful in a clinical context – ideally, they would be sufficient to trigger clinical action rather than another point-of-care diagnostic test. 

2. Multiple levels of analysis. Clinicians today are overwhelmed with data, but they also aren’t ready to fully trust black-box AI systems with the health of their patients. A successful wearable monitor should have the power to churn through all of the data it collects to provide a simple, at-a-glance summary of patient health across the dimensions that matter for clinical action. At the same time, clinicians need the ability to drill down into the details and review raw data to verify predictions generated by machine learning models. As models improve and gain credibility, greater levels of abstraction may be acceptable, but until then the ability to peek behind the curtain will be essential to gain the trust of providers. 

3. Delivering insights (not data) the right way. Apps are convenient for consumers, but healthcare runs on EHR systems like Epic. For wearables to work with healthcare, they should deliver meaningful insights in a way that fits existing clinical workflows – which probably means plugging into the EHR rather than a standalone software solution. That’s not to say patients should be locked out of their own data – but what the patient sees in their app and what the doctor reviews in an EHR will probably be different. Clinical-grade wearables should carefully consider how information is presented to doctors and patients to maximize relevance but also be careful not to alarm patients with alerts that cause them to seek unnecessary care.  

4. Comfort and convenience. There is a growing trend towards “consumerization” of healthcare – and wearables have their role to play in this transition. A wearable device that captures the highest quality data and delivers life-saving insights to doctors is useless if no one wants to wear it. Just because something is a medical device doesn’t mean it can’t be sleek, comfortable, and convenient. Compliance with remote monitoring systems declines exponentially with time – to overcome this tendency, we need new solutions that balance user needs and comfort without sacrificing doctor’s need for actionable data. 

AI is revolutionizing many industries – healthcare included. Wearable devices have the potential to supercharge predictive and preventative care by providing the huge volumes of data required to feed these new AI models. However, fighting the inertia of the healthcare system requires a targeted approach. At Skribe, we’re focusing on a new technology built to maximize data quality and insight delivery in a way that serves doctors’ needs without losing the convenience that makes consumer devices appealing. We believe that this approach can have a significant impact on improving care delivery and outcomes while reducing costs and physician workload.   

References 

1. Serrano, L.P., Maita, K.C., Avila, F.R., Torres-Guzman, R.A., Garcia, J.P., Eldaly, A.S., Haider, C.R., Felton, C.L., Paulson, M.R., Maniaci, M.J. and Forte, A.J., 2023. Benefits and challenges of remote patient monitoring as perceived by health care practitioners: a systematic review. The Permanente Journal, 27(4), p.100. 

2. Siegler, J.E., Patel, N.N. and Dine, C.J., 2015. Prioritizing paperwork over patient care: why can’t we do both?. Journal of Graduate Medical Education, 7(1), pp.16-18.