MétaCan
Menu
Back to cohort
Record W3044306984 · doi:10.2196/21378

Detection of Behavioral Anomalies in Medication Adherence Patterns Among Patients With Serious Mental Illness Engaged With a Digital Medicine System

2020· article· en· W3044306984 on OpenAlexvenueno aff
Jonathan Knights, Zahra Heidary, Jeffrey M. Cochran

Bibliographic record

VenueJMIR Mental Health · 2020
Typearticle
Languageen
FieldPsychology
TopicDigital Mental Health Interventions
Canadian institutionsnot available
FundersOtsuka Pharmaceutical Development and CommercializationOtsuka Pharmaceutical
KeywordsConstruct (python library)Anomaly detectionEntropy (arrow of time)Mental illnessComputer scienceMedicinePsychologyMental healthEconometricsPsychiatryData miningMathematics

Abstract

fetched live from OpenAlex

BACKGROUND: Adherence to medication is often represented in the form of a success percentage over a period of time. Although noticeable changes to aggregate adherence levels may be indicative of unstable medication behavior, a lack of noticeable changes in aggregate levels over time does not necessarily indicate stability. The ability to detect developing changes in medication-taking behavior under such conditions in real time would allow patients and care teams to make more timely and informed decisions. OBJECTIVE: This study aims to develop a method capable of identifying shifts in behavioral (medication) patterns at the individual level and subsequently assess the presence of such shifts in retrospective clinical trial data from patients with serious mental illness. METHODS: We defined the term adherence volatility as "the degree to which medication ingestion behavior fits expected behavior based on historically observed data" and defined a contextual anomaly system around this concept, leveraging the empirical entropy rate of a stochastic process as the basis for formulating anomaly detection. For the presented methodology, each patient's evolving behavior is used to dynamically construct the expectation bounds for each future interval, eliminating the need to rely on model training or a static reference sequence. RESULTS: Simulations demonstrated that the presented methodology identifies anomalous behavior patterns even when aggregate adherence levels remain constant and highlight the temporal dependence inherent in these anomalies. Although a given sequence of events may present as anomalous during one period, that sequence should subsequently contribute to future expectations and may not be considered anomalous at a later period-this feature was demonstrated in retrospective clinical trial data. In the same clinical trial data, anomalous behavioral shifts were identified at both high- and low-adherence levels and were spread across the whole treatment regimen, with 77.1% (81/105) of the population demonstrating at least one behavioral anomaly at some point in their treatment. CONCLUSIONS: Digital medicine systems offer new opportunities to inform treatment decisions and provide complementary information about medication adherence. This paper introduces the concept of adherence volatility and develops a new type of contextual anomaly detection, which does not require an a priori definition of normal and allows expectations to evolve with shifting behavior, removing the need to rely on training data or static reference sequences. Retrospective analysis from clinical trial data highlights that such an approach could provide new opportunities to meaningfully engage patients about potential shifts in their ingestion behavior; however, this framework is not intended to replace clinical judgment, rather to highlight elements of data that warrant attention. The evidence provided here identifies new areas for research and seems to justify additional explorations in this area.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.010
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: Observational
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.003
Threshold uncertainty score0.007

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0010.010
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0010.001
Science and technology studies0.0000.000
Scholarly communication0.0010.000
Open science0.0000.001
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0010.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.026
GPT teacher head0.347
Teacher spread0.321 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designObservational
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations10
Published2020
Admission routes1
Has abstractyes

Explore more

Same venueJMIR Mental HealthSame topicDigital Mental Health InterventionsFrench-language works237,207