Scalable Approach to Consumer Wearable Postmarket Surveillance: Development and Validation Study

Background

With the capability to render prediagnoses, consumer wearables have the potential to affect subsequent diagnoses and the level of care in the health care delivery setting. Despite this, postmarket surveillance of consumer wearables has been hindered by the lack of codified terms in electronic health records (EHRs) to capture wearable use.

Objective

We sought to develop a weak supervision-based approach to demonstrate the feasibility and efficacy of EHR-based postmarket surveillance on consumer wearables that render atrial fibrillation (AF) prediagnoses.

Methods

We applied data programming, where labeling heuristics are expressed as code-based labeling functions, to detect incidents of AF prediagnoses. A labeler model was then derived from the predictions of the labeling functions using the Snorkel framework. The labeler model was applied to clinical notes to probabilistically label them, and the labeled notes were then used as a training set to fine-tune a classifier called Clinical-Longformer. The resulting classifier identified patients with an AF prediagnosis. A retrospective cohort study was conducted, where the baseline characteristics and subsequent care patterns of patients identified by the classifier were compared against those who did not receive a prediagnosis.

Results

The labeler model derived from the labeling functions showed high accuracy (0.92; F1-score=0.77) on the training set. The classifier trained on the probabilistically labeled notes accurately identified patients with an AF prediagnosis (0.95; F1-score=0.83). The cohort study conducted using the constructed system carried enough statistical power to verify the key findings of the Apple Heart Study, which enrolled a much larger number of participants, where patients who received a prediagnosis tended to be older, male, and White with higher CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes, stroke, vascular disease, age 65-74 years, sex category) scores (P<.001). We also made a novel discovery that patients with a prediagnosis were more likely to use anticoagulants (525/1037, 50.63% vs 5936/16,560, 35.85%) and have an eventual AF diagnosis (305/1037, 29.41% vs 262/16,560, 1.58%). At the index diagnosis, the existence of a prediagnosis did not distinguish patients based on clinical characteristics, but did correlate with anticoagulant prescription (P=.004 for apixaban and P=.01 for rivaroxaban).

Conclusions

Our work establishes the feasibility and efficacy of an EHR-based surveillance system for consumer wearables that render AF prediagnoses. Further work is necessary to generalize these findings for patient populations at other sites.

Characterizing Real-World Implementation of Consumer Wearables for the Detection of Undiagnosed Atrial Fibrillation in Clinical Practice: Targeted Literature Review

BACKGROUND

Atrial fibrillation (AF), the most common cardiac arrhythmia, is often undiagnosed because of lack of awareness and frequent asymptomatic presentation. As AF is associated with increased risk of stroke, early detection is clinically relevant. Several consumer wearable devices (CWDs) have been cleared by the US Food and Drug Administration for irregular heart rhythm detection suggestive of AF. However, recommendations for the use of CWDs for AF detection in clinical practice, especially with regard to pathways for workflows and clinical decisions, remain lacking.

OBJECTIVE

We conducted a targeted literature review to identify articles on CWDs characterizing the current state of wearable technology for AF detection, identifying approaches to implementing CWDs into the clinical workflow, and characterizing provider and patient perspectives on CWDs for patients at risk of AF.

METHODS

PubMed, ClinicalTrials.gov, UpToDate Clinical Reference, and DynaMed were searched for articles in English published between January 2016 and July 2023. The searches used predefined Medical Subject Headings (MeSH) terms, keywords, and search strings. Articles of interest were specifically on CWDs; articles on ambulatory monitoring tools, tools available by prescription, or handheld devices were excluded. Search results were reviewed for relevancy and discussed among the authors for inclusion. A qualitative analysis was conducted and themes relevant to our study objectives were identified.

RESULTS

A total of 31 articles met inclusion criteria: 7 (23%) medical society reports or guidelines, 4 (13%) general reviews, 5 (16%) systematic reviews, 5 (16%) health care provider surveys, 7 (23%) consumer or patient surveys or interviews, and 3 (10%) analytical reports. Despite recognition of CWDs by medical societies, detailed guidelines regarding CWDs for AF detection were limited, as was the availability of clinical tools. A main theme was the lack of pragmatic studies assessing real-world implementation of CWDs for AF detection. Clinicians expressed concerns about data overload; potential for false positives; reimbursement issues; and the need for clinical tools such as care pathways and guidelines, preferably developed or endorsed by professional organizations. Patient-facing challenges included device costs and variability in digital literacy or technology acceptance.

CONCLUSIONS

This targeted literature review highlights the lack of a comprehensive body of literature guiding real-world implementation of CWDs for AF detection and provides insights for informing additional research and developing appropriate tools and resources for incorporating these devices into clinical practice. The results should also provide an impetus for the active involvement of medical societies and other health care stakeholders in developing appropriate tools and resources for guiding the real-world use of CWDs for AF detection. These resources should target clinicians, patients, and health care systems with the goal of facilitating clinician or patient engagement and using an evidence-based approach for establishing guidelines or frameworks for administrative workflows and patient care pathways.

Quantifying the Impact of Influenza Among Persons With Type 2 Diabetes Mellitus: A New Approach to Determine Medical and Physical Activity Impact

BACKGROUND

We describe the impact of influenza on medical outcomes and daily activities among people with and without type 2 diabetes mellitus (T2DM).

METHODS

Retrospective cohort analysis of a US health plan offering a digital wellness platform connecting wearable devices capable of tracking steps, sleep, and heart rate. For the 2016 to 2017 influenza season, we compared adults with T2DM to age and gender matched controls. Medical claims were used to define cohorts and identify influenza events and outcomes. Digital tracking data were aggregated at time slices of minute-, day-, week-, and year-level. A pre-post study design compared the peri-influenza period (two weeks before and four weeks after influenza diagnosis) to the six-week preceding period (baseline).

RESULTS

A total of 54 656 T2DM and 113 016 non-DM controls were used for the study. People with T2DM had more influenza claims, vaccinations, and influenza antivirals per 100 people (1.96% vs 1.37%, 34.3% vs 24.3%, and 27.1 vs 22 respectively, P < .001). A total of 1086 persons with T2DM and 1567 controls had an influenza claim (47.4% male, median age 54, 6.4% vs 7.8% trackers, respectively). Glycemic events, pneumonia, and ischemic heart disease increased over baseline during the peri-influenza period for T2DM (1.74-, 7.4-, and 1.6-fold increase respectively, P < .01). In a device wearing subcohort, we observed 10 000 fewer steps surrounding the influenza event, with the lowest (5500 steps) two days postinfluenza. Average heart rate increased significantly (+5.5 beats per minute) one day prior to influenza.

CONCLUSION

Influenza increases rates of pneumonia, heart disease, and abnormal glucose levels among people with T2DM, and negatively impacts daily activities compared to controls.

Validity of Wrist-Worn Activity Trackers for Estimating VO2max and Energy Expenditure

Activity trackers are a simple and mostly low-priced method to capture physiological parameters. Despite the high number of wrist-worn devices, there is a lack of scientific validation. The purpose of this study was to assess whether the activity trackers represent a valid alternative to gold-standard methods in terms of estimating energy expenditure (EE) and maximum oxygen uptake (VO_2max). Twenty-four healthy subjects participated in this study. In total, five commercially available wrist-worn devices were tested with regard to their validity of EE and/or VO_2max. Estimated values were compared with indirect calorimetry. Validity of the activity trackers was determined by paired sample t-tests, mean absolute percentage errors (MAPE), Intraclass Correlation Coefficient, and Bland-Altman plots. Within the tested devices, differences in scattering in VO_2max and EE could be observed. This results in a MAPE > 10% for all evaluations, except for the VO_2max-estimation of the Garmin Forerunner 920XT (7.3%). The latter significantly underestimates the VO_2max (t(23) = –2.37, p = 0.027), whereas the Garmin Vivosmart HR significantly overestimates the EE (t(23) = 2.44, p = 0.023). The tested devices did not show valid results concerning the estimation of VO_2max and EE. Hence, the current wrist-worn activity trackers are most likely not accurate enough to be used for neither purposes in sports, nor in health care applications.