Maximum Time Between Tests: A Digital Biomarker to Detect Therapy Compliance and Assess Schedule Quality in Measurement-Based eHealth Systems for Alcohol Use Disorder

Läs hela artikeln här! 

The monitoring of alcohol use has been based mainly on self-reporting using timeline follow-back questionnaires. The introduction of biomarkers (Andresen-Streichert et al., 2018) has provided more objective data on alcohol consumption. In current healthcare of patients with AUD, the new golden standard for biomarker-based monitoring of alcohol usage is phosphatidyl ethanol (PEth). This due to excellent specificity, sensitivity and a relatively long half-life (~1 week) in blood. The major drawbacks of current disease state monitoring methods are poor time-resolution and problems with the quality of self-reported data (Midanik, 1988).

Methods based on self-reporting using smartphone and Internet-connected systems have opened up new paths to personalized and continuous care for AUD patients (Beckjord and Shiffman, 2014; Quanbeck et al., 2014). These electronic methods address the recall problem and improve time resolution of self-reporting (e.g. daily, weekly), but reports on clinical efficacy are scarce (Gustafson et al., 2014; Rose et al., 2017). The latest sensor-based techniques enable objective measurement-based care using either remote frequent (Gordon et al., 2017; Hämäläinen et al., 2018) or continuous (Leffingwell et al., 2013) real-time monitoring of sobriety.

Digital biomarkers (DBs) are increasingly used to monitor the state of different diseases

Digital biomarkers (DBs) are increasingly used to monitor the state of different diseases (Snyder et al., 2018). We define a DB as patient-generated physiological and behavioural measures collected through sensors and other connected digital tools that can be used to monitor, predict and/or influence health-related outcomes. The first published DB for AUD was the addiction monitoring index (AMI). AMI is based on data from an eHealth system with a cloud-connected pocket-size breathalyzer, and combines the breath alcohol content test results with the pattern of omitted tests (Hämäläinen et al., 2018).

During clinical trials of the eHealth system, we received indications that suboptimal test schedules led to secret drinking and drop-out from the study. Therefore, we hypothesized that time-based parameters extracted from the database might be useful for estimating test compliance. One particular metric – maximum time between tests (MTBT) – could be redesigned as a DB that relies on how a task is conducted rather than what the task produces as a result. Here we describe the development and verification of MTBT as a new DB useful for estimating therapy compliance.