Multi-task transfer learning for the prediction of entity modifiers in clinical text: application to opioid use disorder case detection.

BACKGROUND: The semantics of entities extracted from a clinical text can be dramatically altered by modifiers, including entity negation, uncertainty, conditionality, severity, and subject. Existing models for determining modifiers of clinical entities involve regular expression or features weights that are trained independently for each modifier. METHODS: We develop and evaluate a multi-task transformer architecture design where modifiers are learned and predicted jointly using the publicly available SemEval 2015 Task 14 corpus and a new Opioid Use Disorder (OUD) data set that contains modifiers shared with SemEval as well as novel modifiers specific for OUD. We evaluate the effectiveness of our multi-task learning approach versus previously published systems and assess the feasibility of transfer learning for clinical entity modifiers when only a portion of clinical modifiers are shared. RESULTS: Our approach achieved state-of-the-art results on the ShARe corpus from SemEval 2015 Task 14, showing an increase of 1.1% on weighted accuracy, 1.7% on unweighted accuracy, and 10% on micro F1 scores. CONCLUSIONS: We show that learned weights from our shared model can be effectively transferred to a new partially matched data set, validating the use of transfer learning for clinical text modifiers.

Annotation of Opioid Use Disorder Entity Modifiers in Clinical Text.

Natural Language Processing can be used to identify opioid use disorder in patients from clinical text1. We annotate a corpus of clinical text for mentions of concepts associated with unhealthy use of opiates including concept modifiers such as negation, subject, uncertainty, relation to document time and illicit use.

Emergency-department initiated buprenorphine: Impact on quality of life.

Introduction: Emergency department (ED)-initiated medications for opioid use disorder (MOUD) have emerged as an acute care strategy against the opioid epidemic. When initiated in the outpatient setting, MOUD has been demonstrated to have a positive impact on patients' quality of life (QoL). It is unclear how engagement in ED-initiated MOUD, a novel initiation setting and unique patient cohort, might impact QoL. We sought to describe QoL variables reported by patients engaged in ED-initiated MOUD. Methods: A retrospective observational study of an ED-initiated MOUD program was performed, inclusive of enrollments from July 2019 through February 2022. Participants were interviewed at intake, 3-months, and 6-months, during which QoL indices were measured via Government Performance and Results Act (GPRA) variables. Descriptive statistics and Pearson's Chi-Square analyses were utilized to assess the data. Results: Of 315 participants, majority were white (78.4 %), male (64.4 %), between the ages of 25-44 (74.6 %), and heavily burdened by lack of insurance, homelessness, and unemployment. One hundred forty participants (44.4 % eligible) completed 3-month follow-up and 90 (28.5 %) completed 6-month follow-up. There were no significant demographic differences amongst respondents at 3- and 6-months as compared to intake. Objective QoL variables significantly improved at 3- and 6-months as compared to intake (p < 0.01). Subjective QoL variables also demonstrated significant improvement at follow-up (p < 0.05). Conclusion: ED patients with OUD, eligible for MOUD, may face a number of social and interpersonal variables which heavily impact QoL. ED-initiated MOUD may positively impact subsequent QoL when measured over time.

Impact of Emergency Department-Initiated Buprenorphine on Repeat Emergency Department Utilization.

Introduction: Recent studies have demonstrated the promise of emergency department (ED)-initiated buprenorphine/naloxone (bup/nx) for improving 30-day retention in outpatient addiction care programs for patients with opioid use disorder (OUD). We investigated whether ED-initiated bup/nx for OUD also impacts repeat ED utilization. Methods: We performed a retrospective chart review of ED patients discharged with a primary diagnosis of OUD from July 2019-December 2020. Characteristics considered included age, gender, race, insurance status, domicile status, presence of comorbid Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) diagnosis, presenting chief complaint, and provision of a bup/nx prescription and/or naloxone kit. Primary outcomes included repeat ED visit (opioid or non-opioid related) within 30 days, 90 days, and one year. Statistical analyses included bivariate comparison and Poisson regression. Results: Of 169 participants, the majority were male (67.5%), White (82.8%), uninsured (72.2%), and in opioid withdrawal and/or requesting "detox" (75.7%). Ninety-one (53.8%) received ED-initiated bup/nx, which was independent of age, gender, race, insurance status, presence of comorbid DSM-5 diagnosis, or domicile status. Naloxone was more likely to be provided to patients who received bup/nx (97.8% vs 26.9%; P < 0.001), and bup/nx was more likely to be given to patients who presented with opioid withdrawal and/or requested "detox" (63.3% vs 36.7%; P < 0.001). Bup/nx provision was associated with decreased ED utilization for opioid-related visits at 30 days (P = 0.04). Homelessness and lack of insurance were associated with increased ED utilization for non-opioid-related visits at 90 days (P = 0.008 and P = 0.005, respectively), and again at one year for homelessness (P < 0.001). When controlling for age and domicile status, the adjusted incidence rate ratio for overall ED visits was 0.56 (95% confidence interval [CI] 0.33-0.96) at 30 days, 0.43 (95% CI 0.27-0.69) at 90 days, and 0.60 (95% CI 0.39-0.92) at one year, favoring bup/nx provision. Conclusion: Initiation of bup/nx in the ED setting was associated with decreased subsequent ED utilization. Socioeconomic factors, specifically health insurance and domicile status, significantly impacted non-opioid-related ED reuse. These findings demonstrate the ED's potential as an initiation point for bup/nx and highlight the importance of considering the social risk and social need for OUD patients.