Adherence to medications is a major factor in determining patient outcomes. Stroke patients on direct oral anticoagulants (DOACs) are perfect examples of this adherence issue. Regular adherence to their DOACs is essential to prevent the risk of stroke and bleeding. Previously, we have looked into other creative solutions to improve adherence, including ingestible sensors. This time, we are looking at the concept of directly observed therapy that no longer relies on healthcare professionals reviewing patient videos, but has built-in artificial intelligence (AI) capabilities.
In a 12-week, randomized, parallel-group study, Labovitz et al tested the effect of an AI-enabled app, AiCure, on medication adherence in 28 ischemic stroke patients (20 patients were on DOACs including dabigatran, rivaroxaban, and apixaban; 8 were on warfarin). Not only can AiCure provide patients with reminders and dosing information, it is also capable of carrying out directly observed therapy through HIPAA-compliant facial recognition, automatic medication identification, and real-time medication ingestion confirmation using a smartphone.
Through visual confirmation, the AiCure measured the overall adherence in the AI group to be 90.5% (SD 7.5%), while standard plasma drug concentration tests indicated 100% adherence in the AiCure group, compared with 50% in the control group. The results for the sub-analysis of the DOACs patients only were similar. Mean cumulative adherence based on visual confirmation was 90.1% (SD 7.3%). Standard plasma drug concentration tests showed 100% versus 33% adherence from the AiCure and control groups, respectively.
Interestingly, the difference in cumulative adherence between the AiCure and control groups based on pill count was not as drastic as the difference observed with plasma drug concentration tests for both the entire study group and the DOACs only group (all patients: 97.2% vs 90.6%; DOACs only: 96.4% vs 90.9% for the AiCure and control groups, respectively). This reflects the importance of confirming adherence at the point of administration when regular blood tests are not feasible from both convenience and financial perspectives.
While this study provides valuable information related to ischemic stroke patients, its findings can be extended to other conditions that require high adherence to medication. In fact, the value of AiCure has been tested in a number of therapeutic areas including schizophrenia, and it has been identified as a potential tool in minimizing nonadherence in clinical trials. However, the evidence for AiCure so far seems to come from small-scale studies, and studies with larger sample sizes are warranted. Overall, it seems like a good solution for patients who want to stay well but need a little nudging.