Machine Learning Applied to Clinical Laboratory Data Predicts Patient-Specific, Near-Term Relapse in Patients in Medication for Opioid Use Disorder Treatment

Pyzowski, Paul; Herbert, Barbara; Malik, W.Q.

doi:10.1101/2020.08.10.20163881

Cited by 1 publication

(4 citation statements)

References 8 publications

(7 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, it is difficult to compare our model performance directly with that of other models. The study most similar to ours used urine drug data combined with basic demographic data from 139 OUD patients in a hospital-based outpatient buprenorphine program to predict relapse risk after 4 weeks of urine-validated opioid abstinence 20 . That model achieved overall accuracy of 89.9%, sensitivity of 0.65, specificity of 0.94, and a positive predictive value of 0.65.…”

Section: Discussionmentioning

confidence: 92%

“…The study most similar to ours used urine drug data combined with basic demographic data from 139 OUD patients in a hospital-based outpatient buprenorphine program to predict relapse risk after 4 weeks of urine-validated opioid abstinence. 20 That model achieved overall accuracy of 89.9%, sensitivity of 0.65, specificity of 0.94, and a positive predictive value of 0.65. Our models were lower in specificity and slightly lower in accuracy, but higher in sensitivity and positive predictive value.…”

Section: Discussionmentioning

confidence: 93%

“…Risk-prediction models have been widely implemented in other health fields (eg, cardiovascular disease prognostic and diagnostic models: Khera et al 16 and Wessler et al 17 ); however, to our knowledge, only 3 risk models have been developed to predict MOUD outcomes. [18][19][20] Furthermore, these models focus only on predicting return to use or treatment discharge/retention in specific subpopulations (eg, patients with a prolonged period of opioid abstinence, first MOUD treatment episode, or patients of Hispanic/Latino ethnicity). The modal pattern of MOUD outcomes, at least in the first few months of MOUD, is neither abstinence or full-blown return to use, but rather an intermediate state of sporadic use, which may or may not be a harbinger of unwanted outcomes.…”

mentioning

confidence: 99%

“…One logical way to forecast a patient's future status is by developing a statistical or mathematical risk-prediction model. Risk-prediction models have been widely implemented in other health fields (eg, cardiovascular disease prognostic and diagnostic models: Khera et al 16 and Wessler et al 17 ); however, to our knowledge, only 3 risk models have been developed to predict MOUD outcomes 18–20 . Furthermore, these models focus only on predicting return to use or treatment discharge/retention in specific subpopulations (eg, patients with a prolonged period of opioid abstinence, first MOUD treatment episode, or patients of Hispanic/Latino ethnicity).…”

mentioning

confidence: 99%

See 3 more Smart Citations

Using Machine Learning to Predict Treatment Adherence in Patients on Medication for Opioid Use Disorder

et al. 2022

View full text Add to dashboard Cite

Objective: Patients receiving medication for opioid use disorder (MOUD) may continue using nonprescribed drugs or have trouble with medication adherence, and it is difficult to predict which patients will continue to do so. In this study, we develop and validate an automated risk-modeling framework to predict opioid abstinence and medication adherence at a patient's next attended appointment and evaluate the predictive performance of machine-learning algorithms versus logistic regression. Methods: Urine drug screen and attendance records from 40,005 appointments drawn from 2742 patients at a multilocation office-based MOUD program were used to train logistic regression, logistic ridge regression, and XGBoost models to predict a composite indicator of treatment adherence (opioid-negative and norbuprenorphine-positive urine, no evidence of urine adulteration) at next attended appointment.Results: The XGBoost model had similar accuracy and discriminative ability (accuracy, 88%; area under the receiver operating curve, 0.87) to the two logistic regression models (accuracy, 88%; area under the receiver operating curve, 0.87). The XGBoost model had nearly perfect calibration in independent validation data; the logistic and ridge regression models slightly overestimated adherence likelihood. Historical treatment adherence, attendance rate, and fentanyl-positive urine at current appointment were the strongest contributors to treatment adherence at next attended appointment. Discussion: There is a need for risk prediction tools to improve delivery of MOUD. This study presents an automated and portable risk-modeling framework to predict treatment adherence at each patient's next attended appointment. The XGBoost algorithm appears to provide similar classification accuracy to logistic regression models; however, XGBoost may offer improved calibration of risk estimates compared with logistic regression.

show abstract

Section: Discussionmentioning

confidence: 92%

Section: Discussionmentioning

confidence: 93%