Dataset Information

Machine learning approach to predict medication overuse in migraine patients.

ABSTRACT: Machine learning (ML) is largely used to develop automatic predictors in migraine classification but automatic predictors for medication overuse (MO) in migraine are still in their infancy. Thus, to understand the benefits of ML in MO prediction, we explored an automated predictor to estimate MO risk in migraine. To achieve this objective, a study was designed to analyze the performance of a customized ML-based decision support system that combines support vector machines and Random Optimization (RO-MO). We used RO-MO to extract prognostic information from demographic, clinical and biochemical data. Using a dataset of 777 consecutive migraine patients we derived a set of predictors with discriminatory power for MO higher than that observed for baseline SVM. The best four were incorporated into the final RO-MO decision support system and risk evaluation on a five-level stratification was performed. ROC analysis resulted in a c-statistic of 0.83 with a sensitivity and specificity of 0.69 and 0.87, respectively, and an accuracy of 0.87 when MO was predicted by at least three RO-MO models. Logistic regression analysis confirmed that the derived RO-MO system could effectively predict MO with ORs of 5.7 and 21.0 for patients classified as probably (3 predictors positive), or definitely at risk of MO (4 predictors positive), respectively. In conclusion, a combination of ML and RO - taking into consideration clinical/biochemical features, drug exposure and lifestyle - might represent a valuable approach to MO prediction in migraine and holds the potential for improving model precision through weighting the relative importance of attributes.

SUBMITTER: Ferroni P

PROVIDER: S-EPMC7327028 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning approach to predict medication overuse in migraine patients.

Ferroni Patrizia P Zanzotto Fabio M FM Scarpato Noemi N Spila Antonella A Fofi Luisa L Egeo Gabriella G Rullo Alessandro A Palmirotta Raffaele R Barbanti Piero P Guadagni Fiorella F

Computational and structural biotechnology journal 20200612

Machine learning (ML) is largely used to develop automatic predictors in migraine classification but automatic predictors for medication overuse (MO) in migraine are still in their infancy. Thus, to understand the benefits of ML in MO prediction, we explored an automated predictor to estimate MO risk in migraine. To achieve this objective, a study was designed to analyze the performance of a customized ML-based decision support system that combines support vector machines and Random Optimization ...[more]

PMID: 32637046

Similar Datasets

Project description:Chronic migraine patients are at risk of developing a medication overuse. Brain functional studies in these patients have demonstrated an orbitofrontal hypometabolism, persistent after overuse cessation. Orbitofrontal dysfunction is also present in addiction and thus could predispose migraineurs to medication overuse. The aim of this study was to investigate if orbitofrontal dysfunction can be demonstrated in patients with chronic migraine and medication overuse by performing a systematic neuropsychological evaluation focused on tests that assess frontal lobe function. Second, to establish whether it is related to the outcome of these patients. We prospectively studied 42 chronic migraine patients with medication overuse, 42 episodic migraineurs and 41 controls on a battery of neuropsychological tasks evaluating the orbitofrontal and dorsolateral functioning. Depression, anxiety, and personality traits were also assessed. Chronic migraineurs with medication overuse showed a significant impairment in orbitofrontal task performance and higher depression scores as compared to episodic migraineurs and controls. Dorsolateral dysfunction was present in both groups of migraneurs, who also had higher rates of anxiety as compared to controls. After 1 year of follow-up, migraine patient's outcome was classified according to their medication overuse status. A negative outcome that included persistent or new-onset medication overuse was present in 34% of migraineurs and was associated with baseline poor orbitofrontal task performance, and with mild dorsolateral dysfunction, higher rates of depression, anxiety and neuroticism-anxiety traits. Formal education and years with migraine did not influence outcome. Orbitofrontal dysfunction is present in patients with chronic migraine and medication overuse, and associates with a poor outcome at 1 year of follow-up. Neuropsychological evaluation in migraine may help to detect patients prone to overuse so that appropriate therapeutic attitudes can be taken.

Project description:ObjectiveThis post hoc analysis in patients medically diagnosed with chronic migraine (CM) and medication-overuse headache (MOH) evaluated reductions in the use of acute headache medication (AHM) and sustained changes in the diagnostic status of CM and MOH following eptinezumab treatment in the PROMISE-2 study.BackgroundEptinezumab, a monoclonal antibody that inhibits calcitonin gene-related peptide, is approved in the United States for the preventive treatment of migraine. A previous analysis showed that eptinezumab reduced monthly migraine days and was well tolerated in the subgroup of PROMISE-2 patients diagnosed with both CM and MOH.MethodsThe phase 3, double-blind, placebo-controlled PROMISE-2 study (NCT02974153) randomized adults with CM to eptinezumab 100 mg, 300 mg, or placebo (administered intravenously every 12 weeks for up to two doses). MOH was prospectively diagnosed at screening by trained physicians based on 3 months of medication history and International Classification of Headache Disorders-3β criteria. This post hoc analysis evaluated changes in total and class-specific days of AHM usage, the percentage of patients using AHM at or above MOH diagnostic thresholds, and the percentage of patients experiencing monthly headache and migraine day frequency below diagnostic thresholds for MOH and/or CM.ResultsIn PROMISE-2, 431/1072 (40.2%) patients with CM were diagnosed with MOH (eptinezumab 100 mg, n = 139; 300 mg, n = 147; placebo, n = 145) and were included in this analysis. Total monthly AHM use decreased from 20.6 days/month at baseline to 10.6 days/month over 24 weeks of treatment (49% decrease) with eptinezumab 100 mg, from 20.7 to 10.5 days/month (49% decrease) with eptinezumab 300 mg, and from 19.8 to 14.0 days/month (29% decrease) with placebo. Numerically greater decreases from baseline with eptinezumab were also observed for individual drug classes. In each study month, the percentages of patients who were below MOH thresholds were numerically higher for both eptinezumab doses compared with placebo, as were the percentages of patients experiencing headache and migraine frequency below CM thresholds. Of patients with available data across the entire treatment period, 29.0% (58/200) of patients treated with eptinezumab stopped meeting and remained below diagnostic thresholds for both CM and MOH during Weeks 1-24, as well as 6.3% (6/96) of patients who received placebo.ConclusionsAcross 24 weeks of treatment, eptinezumab reduced AHM use in patients diagnosed with CM and MOH. More than one-fourth (29%) of patients treated with eptinezumab did not meet the diagnostic thresholds for either CM or MOH for the entire treatment period.

Project description:BackgroundMigraine is a disabling neurologic disease that can fluctuate over time in severity, frequency, and acute medication use. Harris Poll Migraine Report Card was a US population-based survey to ascertain quantifiable distinctions amongst individuals with current versus previous high-frequency headache/migraine and acute medication overuse (HFM+AMO). The objective of this report is to compare self-reported experiences in the migraine journey of adults with HFM+AMO to those who previously experienced HFM+AMO but currently have a sustained reduction in headache/migraine frequency and acute medication use.MethodsAn online survey was available to a general population panel of adults (≥18 years) with migraine per the ID Migraine™ screener. Respondents were classified into "current HFM+AMO" (within the last few months had ≥8 headache days/month and ≥10 days/month of acute medication use; n=440) or "previous HFM+AMO" (previously had HFM+AMO, but within the last few months had ≤7 headache days/month and ≤9 days/month of acute medication use; n=110). Survey questions pertained to demographics, diagnosis, living with migraine, healthcare provider (HCP) communication, and treatment.ResultsParticipants in the current HFM+AMO group had 15.2 monthly headache days and 17.4 days of monthly acute medication use in last few months compared to 4.2 and 4.1 days for the previous HFM+AMO group, respectively. Overall, current preventive pharmacologic treatment use was low (15-16%; P>0.1 for current vs previous) in both groups. Previous HFM+AMO respondents reported better current acute treatment optimization. More respondents with current (80%) than previous HFM+AMO (66%) expressed concern with their current health (P<0.05). More than one-third of both groups wished their HCP better understood their mental/emotional health (current 37%, previous 35%; P>0.1 for current vs previous) and 47% (current) to 54% (previous) of respondents worried about asking their HCP too many questions (P>0.1 for current vs previous).ConclusionApart from optimization of acute medication, medical interventions did not significantly differentiate between the current and previous HFM+AMO groups. Use of preventive pharmacological medication was low in both groups. Adults with current HFM+AMO more often had health concerns, yet both groups expressed concerns of disease burden. Optimization of acute and preventive medication and addressing mental/emotional health concerns of patients are areas where migraine care may impact outcomes regardless of their disease burden.

Project description:ObjectiveTo determine the effect of erenumab, a human anti-calcitonin gene-related peptide receptor monoclonal antibody, in patients with chronic migraine and medication overuse.MethodsIn this double-blind, placebo-controlled study, 667 adults with chronic migraine were randomized (3:2:2) to placebo or erenumab (70 or 140 mg), stratified by region and medication overuse status. Data from patients with baseline medication overuse at baseline were used to assess changes in monthly migraine days, acute migraine-specific medication treatment days, and proportion of patients achieving ≥50% reduction from baseline in monthly migraine days.ResultsOf 667 patients randomized, 41% (n = 274) met medication overuse criteria. In the medication overuse subgroup, erenumab 70 or 140 mg groups had greater reductions than the placebo group at month 3 in monthly migraine days (mean [95% confidence interval] -6.6 [-8.0 to -5.3] and -6.6 [-8.0 to -5.3] vs -3.5 [-4.6 to -2.4]) and acute migraine-specific medication treatment days (-5.4 [-6.5 to -4.4] and -4.9 [-6.0 to -3.8] vs -2.1 [-3.0 to -1.2]). In the placebo and 70 and 140 mg groups, ≥50% reductions in monthly migraine days were achieved by 18%, 36% (odds ratio [95% confidence interval] 2.67 [1.36-5.22]) and 35% (odds ratio 2.51 [1.28-4.94]). These clinical responses paralleled improvements in patient-reported outcomes with a consistent benefit of erenumab across multiple measures of impact, disability, and health-related quality of life. The observed treatment effects were similar in the non-medication overuse subgroup.ConclusionsErenumab reduced migraine frequency and acute migraine-specific medication treatment days in patients with chronic migraine and medication overuse, improving disability and quality of life.Clinicaltrialsgov identifierNCT02066415.Classification of evidenceThis study provides Class II evidence that erenumab reduces monthly migraine days at 3 months in patients with chronic migraine and medication overuse.

Dataset Information

Machine learning approach to predict medication overuse in migraine patients.

Publications

Machine learning approach to predict medication overuse in migraine patients.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets