Predicting Disclosure Among People With Opioid Use Disorder: A Machine Learning Approach

The abuse of opioids is a serious and prevalent problem and research is needed on factors that may place individuals at risk for misusing opioids. Depression is a common co-morbid mental health problem among opioid users. Theory and research suggest that early maladaptive schemas may underlie mental health problems including depression and substance abuse. The current study sought to determine whether early maladaptive schemas were associated with depression among a treatment seeking sample of male and female opioid users (n = 194). We also examined whether depression, as assessed by the Minnesota Multiphasic Personality Inventory, Second Edition, varied by gender and whether gender differences in depression remained after controlling for early maladaptive schemas. Results showed that women scored significantly higher than men on three of the five early maladaptive schema domains and that gender did not predict depression after controlling for schema domains. Early maladaptive schemas were also more strongly associated with depression for men than women. Implications of these findings for interventions and future research are discussed. Individuals with opioid dependence have a number of early maladaptive schemas that may be contributing to the onset and maintenance of substance use. Although there are generally broad gender differences in major depression, findings from the current study suggest that early maladaptive schemas are a better predictor of depressive symptoms than gender among opioid dependent adults. The treatment of opioid dependence, with or without co-morbid depressive symptoms, should target early maladaptive schemas.

Purpose of ReviewArtificial intelligence (AI) offers a new frontier for aiding in the management of both acute and chronic pain, which may potentially transform opioid prescribing practices and addiction prevention strategies. In this review paper, not only do we discuss some of the current literature around predicting various opioid-related outcomes, but we also briefly point out the next steps to improve trustworthiness of these AI models prior to real-time use in clinical workflow.Recent FindingsMachine learning-based predictive models for identifying risk for persistent postoperative opioid use have been reported for spine surgery, knee arthroplasty, hip arthroplasty, arthroscopic joint surgery, outpatient surgery, and mixed surgical populations. Several machine learning-based models have been described to predict an individual’s propensity for opioid use disorder and opioid overdose. Natural language processing and large language model approaches have been described to detect opioid use disorder and persistent postsurgical opioid use from clinical notes. Summary AI holds significant promise in enhancing the management of acute and chronic opioids, which may offer tools to help optimize dosing, predict addiction risks, and personalize pain management strategies. By harnessing the power of AI, healthcare providers can potentially improve patient outcomes, reduce the burden of opioid addiction, and contribute to solving the opioid crisis.

Opioid use during pregnancy is a critical public health concern that has given rise to significant treatment needs throughout the perinatal period. The purpose of this study was to gather perspectives of pregnant and parenting individuals in recovery from opioid use disorder (OUD) to better understand their needs and identify ways to improve care and support parenting. We conducted semi-structured interviews with 18 pregnant and postpartum people receiving medication for opioid use disorder (MOUD) at an outpatient buprenorphine clinic. The coding team developed a codebook using both inductive and deductive methods that was applied to all interviews. Key topics and trends were identified using thematic analysis. Four themes emerged from participants' reflections on their experiences with health care as pregnant and parenting people in recovery: (1) need for support with parenting, (2) concerns about MOUD, (3) difficulty accessing resources, and (4) value of peer support and options for receiving information. There is a need to prioritize support with parenting, understanding MOUD, and accessing basic needs to better prepare pregnant and early postpartum individuals for parenting in recovery from OUD. It is also important to incorporate peer support into services to better support pregnant and parenting people in recovery.

A large body of research demonstrates the safety and efficacy of medication for opioid use disorder to reduce opioid misuse and related harms. However, stigma is a substantial barrier to broader use of these medications. This endpage reviews the impact of stigma toward medication for opioid use disorder on treatment access, uptake, retention, and outcomes. I then suggests areas for future research and social change. As one of the primary service providers for people with opioid use disorder, social workers should be at the forefront of efforts to destigmatize medication treatments. Social workers are well positioned to address the consequences of stigma toward medication for opioid use disorder through research, education, practice, and systems change. VIDEO ABSTRACT SCRIPT Methadone, buprenorphine, and extended-release naltrexone, the three FDA-approved medications for opioid use disorder, are among the most powerful tools we have in our fight against the rising drug overdose crisis. A large body of research demonstrates the safety and efficacy of these medications to reduce opioid cravings, improve treatment outcomes, and prevent overdose death. However, stigma is a major barrier to their use. Despite a strong evidence base, medications for opioid use disorder are commonly viewed as ineffective and illegitimate treatments that simply ‘substitute one drug for another.’ Stigma is encoded in stringent laws and regulations which limit access to medication. For people with opioid use disorder, stigma can influence beliefs about medications and affect the likelihood that one will initiate and maintain treatment. Stigma also erodes self-esteem and social support, two important predictors for recovery outcomes. In this way, stigma toward medications for opioid use disorder has become a pressing social justice issue. As one of the primary service providers for people with opioid use disorder, social workers should be at the forefront of efforts to destigmatize medication treatments. This endpage is a call to action for social workers to address stigma toward medication treatments through research, education, practice, and systems change.

Most screening tests for T2DM in use today were developed using multivariate regression methods that are often further simplified to allow transformation into a scoring formula. The increasing volume of electronically collected data opened the opportunity to develop more complex, accurate prediction models that can be continuously updated using machine learning approaches. This study compares machine learning-based prediction models (i.e. Glmnet, RF, XGBoost, LightGBM) to commonly used regression models for prediction of undiagnosed T2DM. The performance in prediction of fasting plasma glucose level was measured using 100 bootstrap iterations in different subsets of data simulating new incoming data in 6-month batches. With 6 months of data available, simple regression model performed with the lowest average RMSE of 0.838, followed by RF (0.842), LightGBM (0.846), Glmnet (0.859) and XGBoost (0.881). When more data were added, Glmnet improved with the highest rate (+ 3.4%). The highest level of variable selection stability over time was observed with LightGBM models. Our results show no clinically relevant improvement when more sophisticated prediction models were used. Since higher stability of selected variables over time contributes to simpler interpretation of the models, interpretability and model calibration should also be considered in development of clinical prediction models.

BackgroundA brief gonadotropin-releasing hormone analogues (GnRHa) stimulation test which solely focused on LH 30-minute post-stimulation was considered to identify girls with central precocious puberty (CPP). However, it was tested using traditional statistical methods. With advanced computer science, we aimed to develop a machine learning-based diagnostic model that processed baseline CPP-related variables and a brief GnRHa stimulation test for CPP diagnosis.MethodsWe recruited girls suspected of precocious puberty and underwent a GnRHa stimulation test at Children Hospital 2, Vietnam, and Cathay General Hospital, Taiwan. Clinical data, bone age measurement, and 30-min post-stimulation blood test were used to build up the predictive model. The candidate model was developed by different machine learning algorithms that were mainly evaluated by sensitivity, specificity, the area under the receiver operator characteristic curve (AUC), and F1-score in internal and external validation data to classify girls as CPP and non-CPP at different time-points (0-min, 30-min, 60-min, and 120-min post-stimulation).ResultsAmong the 614 girls diagnosed with PP, 524 (85.3%) had CPP. The random forest algorithm yielded the highest value of F1-score (0.976), specificity (0.893), positive predicted value (0.987), and relatively high value of AUC (0.972) that contributed to high probability to identify CPP. The performance metrics of the 30-min post-stimulation diagnostic model including sensitivity and specificity surpassed those of the 0-minute model (0-min) and were equivalent to those of the model obtained 60-min and 120-min post-stimulation. Hence, our machine learning-based model helps shorten the stimulation test to 30 minutes after GnRHa injection, in general, it requires 120 minutes for a completed GnRHa stimulation test.ConclusionsWe developed a diagnostic model based on clinical features and a single sample 30-minute post-stimulation to identify CPP in girls that can reduce distress for children caused by multiple blood samplings.

Background: Cirrhosis is responsible for a large proportion of mortality worldwide. Despite having multiple scoring systems, organ allocation for end-stage liver disease remains a major problem. Since anthropometric indices play important roles in predicting the prognosis of patients with cirrhosis, these variables were used in establishment of a novel scoring system. Methods: In order to evaluate a machine learning approach for predicting the probability of three-month mortality in cirrhotic patients awaiting liver transplantation, the clinical and anthropometric information of 64 patients referred to Abu-Ali-Sina Transplantation Center were collected and followed for three months. A LASSO logistic regression model was used to devise and validate a new machine learning approach and compare it to the Model for End-Stage Liver Disease (MELD) regarding the three-month mortality of cirrhotic patients. Hand grip, skeletal muscle mass index, average mean arterial pressure, serum sodium, and total bilirubin were assessed with this new machine learning approach to predict the prognosis of patients with cirrhosis, which we named the Sina score. Results: Sixty-four patients were enrolled, with a mean age of 46.50 ± 12.871 years. Like the MELD score, the Sina score is a precise prognostic tool for predicting the three-month mortality probability in cirrhotic patients [area under the curve (AUC) = 0.753 and p = 0.005 vs. AUC = 0.607 and p = 0.238]. Our machine learning approach, the Sina score, was shown to be a precise prognostic tool, like the MELD, for the prediction of the three-month mortality probability of cirrhotic patients awaiting liver transplantation. Conclusions: The Sina score, given that its level of precision is on par with the MELD, can be recommended for the prediction of three-month mortality in cirrhotic patients awaiting liver transplantation.

Machine learning provides a flexible technique to predict the survival of patients who are admitted to hospital as emergency admissions. Mortality prediction is a central component of emergency patient quality of care and this can act as an indicator of severity to determine who needs prioritized care. Machine learning-based models, as opposed to human-crafted severity score systems, allow for much more complex and updateable models to be developed based on a larger set of input data attributes. While various studies of machine learning-based predictive models for predicting inpatient mortality have been carried out there is little literature on performance maintenance of these models. Determining the performance maintenance of these models over time determines how reliably they can be utilized into the future and for how long. The best performing model in this study achieve’s an AUC of 0.86 upon training and is able to maintain a similarly high AUC of 0.845 as of the end of the period of performance maintenance evaluation nine months later. This is the first paper that the authors are aware of to consider and measure relative performance maintenance of machine learning-based models for emergency admission mortality prediction.

<h3>Abstract</h3> <h3>Objective</h3> Stress urinary incontinence (SUI) and pelvic organ prolapse (POP) are treated with surgical mesh devices; evidence of their long-term complications is lacking. To examine long-term complications in women with SUI and/or POP, with and without surgical mesh implants. <h3>Design</h3> Longitudinal open cohort study from April 01 2006 to November 30 2018 <h3>Setting</h3> The Clinical Practice Research Datalink (CPRD) Gold database, linked to Hospital Episodes Statistics (HES) inpatient data, Office for National Statistics mortality data, and Index of Multiple Deprivation socioeconomic status data. <h3>Participants</h3> Women aged ≥18 years with a diagnostic SUI/POP code. <h3>Exposure</h3> Mesh surgery coded in HES or CPRD data, compared to no mesh surgery. <h3>Main Outcomes measures</h3> Rates of diagnoses of depression, anxiety or self-harm (composite measure) and sexual dysfunction, using Cox proportional hazards regression, and rates of prescriptions for antibiotics and opioids, using negative binomial regression. <h3>Results</h3> There were 220,544 women eligible for inclusion; 74% (n=162,687) had SUI, 37% (n=82,123) had POP and 11% (n=24,266) had both. Women undergoing mesh surgery for SUI or POP had higher rates of antibiotic use (SUI: IRR 1.15 (95% CI 1.13 to 1.18; p&lt;0.001); POP: IRR 1.09 (95% CI 1.04 to 1.14; p&lt;0.001)). Women with no previous history of the outcome, who underwent mesh surgery for SUI or POP, had higher rates of depression, anxiety, or self-harm (SUI: HR 2.43 (95% CI 2.19 to 2.70; p&lt;0.001; POP: HR 1.47 (95% CI 1.19 to 1.81; p&lt;0.001)), sexual dysfunction (SUI: HR 1.88 (1.50 to 2.36; p&lt;0.001); POP: HR 1.64 (95% CI 1.02 to 2.63; p=0.04)) and opioid use (SUI: IRR 1.40 (95% CI 1.26 to 1.56, p&lt;0.001); POP: IRR 1.23 (95% CI 1.01 to 1.49; p=0.04)). Women with a history of depression, anxiety and self-harm had lower rates of these outcomes with SUI or POP mesh surgery (SUI: HR 0.70 (95% CI 0.67 to 0.73; p&lt;0.001), POP: HR 0.72 (95% CI 0.65 to 0.79; p&lt;0.001). Women with a history of opioid use who had POP mesh surgery had lower rates of prescriptions (IRR 0.91 95% CI (0.86 to 0.96); p=0.001). <h3>Conclusions</h3> Mesh surgery was associated with poor mental and sexual health outcomes, alongside increased opioid and antibiotic use, in women with no history of these outcomes and improved mental health, and lower opioid use, in women with a previous history of these outcomes. Careful consideration of the benefits and risk of mesh surgery for women with SUI or POP on an individual basis is required. <h3>What is known on this topic</h3> There are well publicised concerns about high complication rates and harms associated with surgical mesh, a treatment that has been used in urogynaecological procedures to treat stress urinary incontinence (SUI) and pelvic organ prolapse (POP) for the past 20 years. Yet, there is a consistent lack of long-term evidence over the use of surgical mesh beyond surgical readmission, with NICE, Cochrane, and the FDA all reporting a lack of long-term outcome data. This study reports long-term complications in women with stress urinary incontinence (SUI) and/or pelvic organ prolapse (POP), with and without surgical mesh implants. <h3>What this study adds</h3> In this cohort of 162,687 women with SUI and 82,123 women with POP, rates of depression, anxiety, self-harm, sexual dysfunction and opioid prescriptions were higher in women with mesh-surgery who had no previous history of these outcomes. Lower rates of depression, anxiety and self-harm were seen in women with a history of these outcomes and SUI or POP mesh surgery. Opioid prescribing was lower in women with a previous history of prescriptions and POP mesh surgery. Careful consideration of the benefits and risk of mesh surgery for women with SUI or POP on an individual basis is required.

Opioid use disorder and opioid deaths have increased dramatically in young adults in the US, but the age-related course or precursors to opioid use among young people are not fully understood. To document age-related changes in opioid use and study the childhood antecedents of opioid use by age 30 years in 6 domains of childhood risk: sociodemographic characteristics; school or peer problems; parental mental illness, drug problems, or legal involvement; substance use; psychiatric illness; and physical health. This community-representative prospective longitudinal cohort study assessed 1252 non-Hispanic White individuals and American Indian individuals in rural counties in the central Appalachia region of North Carolina from January 1993 to December 2015. Data were analyzed from January 2019 to January 2020. Between ages 9 and 16 years, participants and their parents were interviewed up to 7 times using the Child and Adolescent Psychiatric Assessment and reported risk factors in 6 risk domains. Participants were assessed again at ages 19, 21, 25, and 30 years for nonheroin opioid use (any and weekly) and heroin use using the structured Young Adult Psychiatric Assessment. Of 1252 participants, 342 (27%) were American Indian. By age 30 years, 322 participants had used a nonheroin opioid (24.2%; 95% CI, 21.8-26.5), 155 had used a nonheroin opioid weekly (8.8%; 95% CI, 7.2-10.3), and 95 had used heroin (6.6%; 95% CI, 5.2-7.9). Childhood risk markers for later opioid use included male sex, tobacco use, depression, conduct disorder, cannabis use, having peers exhibiting social deviance, parents with legal involvement, and elevated systemic inflammation. In final models, childhood tobacco use, depression, and cannabis use were most robustly associated with opioid use in young adulthood (ages 19 to 30 years). Chronic depression and dysthymia were strongly associated with any nonheroin opioid use (OR. 5.43; 95% CI, 2.35-12.55 and OR, 7.13; 95% CI, 1.99-25.60, respectively) and with weekly nonheroin opioid use (OR, 8.89; 95% CI, 3.61-21.93 and OR, 11.51; 95% CI, 3.05-42.72, respectively). Among young adults with opioid use, those with heroin use had the highest rates of childhood psychiatric disorders and comorbidities. Childhood tobacco use and chronic depression may be associated with impaired reward system functioning, which may increase young adults' vulnerability to opioid-associated euphoria. Preventing and treating early substance use and childhood mental illness may help prevent later opioid use.

Risk Factors for Chronic Prescription Opioid Use in Multiple Sclerosis

Chapter 5 - Opioid use disorder and HIV

A quantitative understanding of cropland nitrogen (N) runoff loss is critical for developing efficient N pollution control strategies. Using correlation analysis, a structural equation model, variance decomposition, and machine learning methods, this study identified the primary influencing factors of total N (TN) runoff loss from uplands (n=570) and paddy (n=434) fields in the Yangtze River Basin (YRB) and then developed a machine learning-based prediction model to quantify cropland N runoff loss load. The results indicated that runoff depth, soil N content, and fertilizer addition rate were the major influencing factors of TN runoff loss from uplands, whereas TN runoff loss rate from paddy fields was mainly regulated by runoff depth and fertilizer addition rate. Among the four used machine learning methods, the prediction models based on the random forest algorithm presented the highest accuracy (R2=0.65-0.94) for predicting upland and paddy field TN runoff loss rates. The random forest algorithm based model estimated a total cropland TN loss load in the YRB of 0.47 Tg·a-1 (upland:0.25 Tg·a-1; paddy field:0.22 Tg·a-1) in 2013, with 58% of TN runoff loss load derived from the midstream and downstream regions. The models predicted that TN runoff loss loads from croplands in YRB would decrease by 2.4%-9.3% for five scenarios, with higher TN load reductions occurring from scenarios with decreased runoff amounts. To mitigate cropland N nonpoint source pollution in YRB, it is essential to integrate efficient water, fertilizer, and soil nutrient managements as well as to consider the midstream and downstream regions as the high priority area. The machine learning-based modeling method developed in this study overcame the difficulty of identifying the functional relationships between cropland TN loss rate and multiple influencing factors in developing relevant prediction models, providing a reliable method for estimating regional and watershed cropland TN loss load.

Customers are one of the most valuable assets within an organization. Customer churn, the phenomenon of losing existing customers to competitors, has a significant impact on business. The predictive customer churn models provide the insights needed to pre-emptively address potential issues, enhance customer satisfaction, and develop effective retention strategies. These machine learning-based churn prediction models accurately categorize customers into two groups: those likely to remain loyal and those at risk of leaving. Given the complicated relationship among characteristics that impact customer turnover, it is essential to analyze the reasons behind high churn rates. Moreover, evaluating the effectiveness of various predictive models through hyperparameter optimization and cross-validation can help organizations fine-tune their approaches to forecasting churn and ultimately bolster their retention efforts. It is found in the present study that the boosted versions of the models are clearly superior to the ordinary (non-boosted) counterparts. By utilizing these advanced techniques, companies can effectively lower churn rates while fostering a more loyal customer base.

Introduction: The risk of mortality associated with the co-prescribing of benzodiazepines and opioids has been explored in a number of papers mainly focusing on strong opioids. The mortality risk associated with the use of weak opioids has not been dealt with to a similar extent. Objective: To assess the mortality risk in primary care patients with consistent 3-year co-prescribing of benzodiazepine/Z-drugs (benzodiazepine receptor modulators) and mainly weak opioids (codeine, tramadol). Methods: Of 221,804 patients contacting the primary healthcare centres, 124,436 were selected for further analysis, 88,832 participants fulfilled the inclusion criteria, aged 10–69 years and were divided into four groups with neither any use of benzodiazepines/Z-drugs nor opioids as Group 1, 3 years’ use of opioids and no/minimal benzodiazepines/Z-drugs as Group 2, with benzodiazepines/Z-drugs and no/minimal opioids as Group 3, and finally both benzodiazepines/Z-drugs and opioids as Group 4. Hazard ratios were calculated with the no-drug group as a reference, using Cox proportional hazards regression model adjusted for age, sex, number of chronic conditions and cancer patients excluded (n = 87,314). Results: Hazard ratios for mortality increased both in Group 3 where it was 2.66 (95% CI 2.25–3.09) and in Group 4 where it was 5.12 (95% CI 4.25–6.17), with increased dose and higher number of chronic conditions. In Group 4 an opioid dose-dependent increase in mortality among persons using >1000 DDDs benzodiazepines/Z-drugs was observed when those on less than ≤300 DDDs of opioids with HR 4.94 (95% CI 3.54–6.88) were compared to those on >300 DDDs with HR 7.61/95% CI 6.08–9.55). This increase in mortality was not observed among patients on <1000 DDDs of benzodiazepines/Z-drugs. Conclusion: The study supports evidence suggesting that mortality increases in a dose-dependent manner in patients co-prescribed benzodiazepines/Z-drugs and weak opioids (codeine, tramadol). An association between the number of chronic conditions and a rise in mortality was found. Long-term use of these drugs should preferably be avoided. Non-pharmacological therapy should be seriously considered instead of long-term use of benzodiazepines/Z-drugs, and deprescribing implemented for chronic users of these drugs when possible.