Currently Viewing:
Currently Reading
Machine Learning Model Predicts Higher Resource Use Among Patients With Chronic Migraine
May 25, 2019 – AJMC Staff
Patients Using Medication for Migraine Still Have Unmet Treatment Needs, Survey Says
May 14, 2019 – Allison Inserro
Study Suggests Benefit to Integrating Care for Migraine, Depression
May 11, 2019 – AJMC Staff
Study Examines Link Between Psoriasis, Migraine
May 04, 2019 – Allison Inserro
Early Onset of Migraine May Predispose Patients to Increased Risk of Stroke
May 03, 2019 – Samantha DiGrande
Demographics, Characteristics of Youth Receiving Migraine Treatment at Infusion Center
April 26, 2019 – Allison Inserro
Study Calls for More Effective Prophylactic Treatments for Migraine
April 25, 2019 – Samantha DiGrande
Shoshana Lipson Explains How "CGRP & Migraine Community" Supports Patients
April 18, 2019
Study Finds Association Between Migraine, Fibromyalgia Onset
April 16, 2019 – Allison Inserro

Machine Learning Model Predicts Higher Resource Use Among Patients With Chronic Migraine

AJMC Staff
Higher migraine-related costs were the top indicators of a patient having chronic migraine, according to a poster presented at ISPOR 2019.
A machine learning model was able to discern which patients with migraine had higher cost, resource use, and medication use, and it also found that those patients were more likely to have a diagnosis of chronic migraine (CM) versus episodic migraine (EM), according to recent research. 

Higher migraine-related costs were the top indicators of a patient having CM, according to the poster presented at ISPOR 2019.

EM is characterized as up to 14 migraine days per month, whereas CM is defined as 15 or more headache days per month, at least 8 of which have to be typical migraine headache days. 

Diagnosis codes in the MarketScan database were used to produce 3 cohorts containing CM, EM, and patients without headache (a negative control group).

Features were derived and a list of supervised machine learning algorithms were explored, including random forest, neural networks, gradient boost, and stacked model. Particle swarm optimization algorithm was applied to explore the optimal parameters of the machine learning algorithms.

Next, 395 features were constructed. Each group—CM, EM, and negative control—had 18,832 patients. The gradient boost algorithm produced the best result, which reported the accuracy and kappa at 0.75 and 0.63, respectively.

After applying thresholds in the optimized gradient boost model, the accuracy and kappa were further increased to 0.81 and 0.72 with prediction decision made to 80% of the data set. Collinearity and high correlation were found with features within the group regarding cost, resource, and medication use. Results showed that 27% of patients with CM and 9% of those with EM received wrong predictions from the fine-tuned random forest, gradient boost, and neural network.

Migraine-related costs were ranked as top important features.

The researchers said additional research on a data source with more migraine-specific clinical outcomes is warranted.


Tian H, Wang X, Lopez Lopez C, Olson M, Kahler K, Fang J. Differentiation of episodic migraine and chronic migraine using a machine learning technique. Presented at: ISPOR 2019 Annual Meeting; New Orleans, LA; May 18-22, 2019. Poster PND73.

Copyright AJMC 2006-2020 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up