By Allison Proffitt
February 25, 2022 | Michael Snyder has been tracking his own health with extreme attention for 12 years now, gathering datapoints including his genomic data, blood markers, microbiome makeups, clinical measures, and much more. He’s found it extremely fruitful, and over the past two years, he’s applying what he’s learned to early diagnoses of COVID-19.
At his talk this week at the Molecular and Precision Med Tri-Con, Snyder, the Stanford Ascherman Professor and Chair of Genetics and the Director of the Center of Genomics and Personalized Medicine, outlined the issues with medicine today. Our healthcare system is reactive and much too focused on illness. Care involves infrequently gathering a few data points, and then making care recommendations based on population averages. Instead, he argues for a proactive approach that would frequently gather many more datapoints, leading to individualized healthcare. It’s a precision health model, he said, not a precision medicine approach.
Our goal, he said, should be to better understand what it means to be healthy on an individual basis. How do health patterns differ for different people? What is my resting heart rate? My normal body temperature? What do aberrations from the baseline mean for my health?
Several years ago Snyder launched a “personal omics profiling” pilot project enrolling 109 individuals with a median age of 53.4. For the past nine years, the cohort has been tracked with omics data being gathered every three months (unless ill, then more frequently) along with a host of other clinical tests, questionnaires, and bloodwork and sensor data. It is literally billions of measurements, he said.
In the first three years, the study made 49 major health discoveries including early diagnoses of cancer, infections, and serious heart issue—all pre-symptomatically. No one technology found all of these things, Snyder points out, but the combination proved very very powerful for detecting disease early. Snyder started a company, Qbio, that serves as a medical version of the study, using genomics, clinical tests, whole-body MRI, sensors, and many other tests to track health. QBio has been equally effective to the pilot, Snyder said, having found early stage leukemia, pancreatic cancer, ovarian cancer, prostate cancer, brain lesions, aneurysm, and more.
Tracker on Your Wrist
Among the findings of these studies, Snyder found that simply tracking changes in resting heart rate was proving to be very useful in detecting infections and illnesses. While tracking his own heart rate and blood oxygen levels, Snyder detected his own Lyme disease infection based on increased skin temperature, elevated resting heart rate, and low blood oxygen—all before a fever or any symptoms. The infection was later confirmed with bloodwork.
In 2017, the team published the Change-of-Heart algorithm which used wearable data to identify abnormal heart rate patterns. (DOI: 10.1371/journal.pbio.2001402).
Pandemic Opportunity
In 2020 when the COVID-19 pandemic began, Snyder had already been improving the algorithm and building the detection system at scale. He was able to quickly launch a new study to see if off-the-shelf wearables—first Fitbits, now both Fitbits and Apple Watches—could be predictive of infection. The study has, thus far, enrolled over 5,000 people.
Wearables are powerful because you wear them 24/7, he explained, and currently 20% of the US population is wearing one. Even small changes in heart rate and sometimes skin temperature can be detected and tracked over time.
Among the study participants were 32 COVID-19 positives, with both diagnosis and symptom onset dates. Snyder’s group found that they could detect COVID-19 infection in advance of symptoms with a Fitbit based on elevated heart rate. By watching heart rate trends over time, the alerting algorithm, NightSignal flagged a change 9.5 days before symptom onset.
The change isn’t dramatic—between two and seven additional beats per minute—but if you’re measuring constantly, you see this change pretty well, Snyder said. It works about 80% of the time, and the app flagged a change a median of four days before symptoms, seven days before diagnosis.
Of course the app is not diagnostic, and a change in heart rate can be attributed to things other than COVID-19 or even an infection. Workplace stress can have lasting impacts on heart rate, Snyder said, as done intense exercise, high volumes of alcohol consumption, or a vaccination. But individuals can take those details into account, Snyder pointed out. His hope is that the alerts would prompt people to take earlier COVID-19 tests if the app picked up heart rate changes without any known external cause.
The ultimate goal, Snyder said, is to build personal monitoring systems that can accurately track various markers and alert to any deviations from baseline.