JMIR Publications

ABSTRACT

Background:

With the capability to render pre-diagnosis, consumer wearables have the potential to affect subsequent diagnosis and the level of care in the healthcare delivery setting. Despite this, post-market surveillance of consumer wearables has been hindered by the lack of codified terms in EHR to capture wearable use.

Objective:

We sought to develop a weak supervision-based approach to demonstrate the feasibility and efficacy of EHR-based post-market surveillance on consumer wearables that render AF pre-diagnosis.

Methods:

We applied data programming where labeling heuristics are expressed as code-based labeling functions to detect AF pre-diagnosis incidents. A labeler model was then learned from the labeling function predictions using the Snorkel framework. Running the labeler model on the clinical notes probabilistically labeled the notes, which were then used as a training set to fine-tune a classifier called Clinical-Longformer. Resulting classifier identified patients with AF pre-diagnosis mentions. 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 pre-diagnosis.

Results:

Labeler model learned from labeling functions showed high accuracy (0.92, F1-score 0.77) on the training set. The classifier learned on the probabilistically labeled notes accurately identified patients with AF pre-diagnosis (0.95, F1-score 0.83). Cohort study conducted using the constructed system carried enough statistical power to verify the key findings of the Apple Heart Study that enrolled a much larger number of participants, where those patients who received pre-diagnosis tended to be older, male, and White with higher CHA2DS2-VASc scores (P < .001). We also made a novel discovery that patients with pre-diagnosis are enriched for anticoagulation (50.63% vs. 35.85%) and eventual diagnosis of AF (13.38% vs. 1.38%). At the index diagnosis, existence of pre-diagnosis did not stratify patients 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 pre-diagnosis. Further work is necessary to generalize these findings for patient populations at other sites.