Prediction of drug effectiveness in rheumatoid arthritis patients based on machine learning algorithms

Published in

International Conferences on Biomedical and Bio...(2022)

External Links:

Generate Graph DownloadPDF

TL;DR

This study introduced a Drug Response Prediction (DRP) framework that provides a complete pipeline to predict disease activity scores and identify the group that does not respond well to anti-TNF treatments, thus showing promise in supporting clinical decisions based on EHR information.

Abstract

Rheumatoid arthritis (RA) is the most common inflammatory arthritis, affecting 1% of the population. It is an autoimmune condition resulting in significant joint destruction and morbidity. Machine learning (ML) has the potential to identify patterns in patient electronic health records (EHR) to forecast the best clinical treatment to improve patient outcomes. This study introduced a Drug Response Prediction (DRP) framework with two main goals: 1) design a data processing pipeline to extract information from tabular clinical data, and then preprocess it for functional use, 2) predict RA patients’ response to drugs and evaluate classification models’ performance. We propose a novel two-stage ML framework based on European Alliance of Associations for Rheumatology (EULAR) criteria cutoffs to model drug effectiveness. In the first stage, the ML models regress the changes in the Disease Activity Score in 28 joints (ΔDAS28) of patients who are bio-naïve to anti-tumor necrosis factor (TNF) treatments; in the second stage, the patient’s responses to drugs are classified using predicted ΔDAS28 scores with thresholds. We empirically show that such division into subtasks significantly improves the accuracy of predicting drug effectiveness in RA patients. Furthermore, regression of ΔDAS28 scores makes our model more interpretable to health care providers, and the classification of the change between initial and 3-month DAS scores would give an easy-to-understand binary recommendation. Our model Stacked-Ensemble DRP was developed and cross-validated using data from 425 RA patients. The evaluation used a subset of 124 patients (30%) from the same data source. In the evaluation of the test set, two-stage DRP leads to improved classification accuracy over other end-to-end classification models for binary classification. Our proposed method provides a complete pipeline to predict disease activity scores and identify the group that does not respond well to anti-TNF treatments, thus showing promise in supporting clinical decisions based on EHR information. The code is open source and is available on GitHub: https://github.com/Gaskell-1206/Ensemble_DRP.

Prediction of drug effectiveness in rheumatoid arthritis patients based on machine learning algorithms

Published in

International Conferences on Biomedical and Bio...(2022)

External Links:

Generate Graph DownloadPDF

TL;DR

Abstract

References40 items

DAS28‐CRP Cutoffs for High Disease Activity and Remission Are Lower Than DAS28‐ESR in Rheumatoid Arthritis

Machine learning approaches to drug response prediction: challenges and recent progress

Two-Stage Ransomware Detection Using Dynamic Analysis and Machine Learning Techniques

Machine Learning to Predict Anti–Tumor Necrosis Factor Drug Responses of Rheumatoid Arthritis Patients by Integrating Clinical and Genetic Markers

DAS28-CRP and DAS28-ESR cut-offs for high disease activity in rheumatoid arthritis are not interchangeable

Crowdsourced assessment of common genetic contribution to predicting anti-TNF treatment response in rheumatoid arthritis

A novel method predicting clinical response using only background clinical data in RA patients before treatment with infliximab

XGBoost: A Scalable Tree Boosting System

“Design characteristics of the CORRONA CERTAIN study: a comparative effectiveness study of biologic agents for rheumatoid arthritis patients”

The early clinical course of infliximab treatment in rheumatoid arthritis: results from the REMARK observational study.

Gradient boosting machines, a tutorial

Crowdsourcing genetic prediction of clinical utility in the Rheumatoid Arthritis Responder Challenge

Obesity and reduction of the response rate to anti–tumor necrosis factor α in rheumatoid arthritis: An approach to a personalized medicine

Impact of patient and disease characteristics on therapeutic success during adalimumab treatment of patients with rheumatoid arthritis: data from a German noninterventional observational study

The value of rheumatoid factor and anti-citrullinated protein antibodies as predictors of response to infliximab in rheumatoid arthritis: an exploratory study.

The lifetime risk of adult-onset rheumatoid arthritis and other inflammatory autoimmune rheumatic diseases.

Scikit-learn: Machine Learning in Python

Body mass index and clinical response to infliximab in rheumatoid arthritis.

Convergent Random Forest predictor: methodology for predicting drug response from genome-scale data applied to anti-TNF response.

Prediction of DAS28-ESR remission at 6 months by baseline variables in patients with rheumatoid arthritis treated with etanercept in Japanese population

Are biologics more effective than classical disease-modifying antirheumatic drugs?

The impact of rheumatoid arthritis and biologics on employers and payers.

Disease Activity Score 28 (DAS28) using C-reactive protein underestimates disease activity and overestimates EULAR response criteria compared with DAS28 using erythrocyte sedimentation rate in a large observational cohort of rheumatoid arthritis patients in Japan

Extremely randomized trees

Remission and active disease in rheumatoid arthritis: defining criteria for disease activity states.

Generalized, Linear, and Mixed Models

Improved Classification of Recent HIV-1 Infection by Employing a Two-Stage Sensitive/Less-Sensitive Test Strategy

Generalized Linear Models

Bagging Predictors

Experiments with a New Boosting Algorithm

Judging disease activity in clinical practice in rheumatoid arthritis: first step in the development of a disease activity score.

The Strength of Weak Learnability

The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis.

H2O AutoML: Scalable Automatic Machine Learning

Deep Learning

Super Learner

Disease activity scores using C-reactive protein: CRP may replace ESR in the assessment of RA disease activity

Factors predicting response to treatment in rheumatoid arthritis: the importance of disease duration.

Random decision forests

Improved classiﬁcation of recent hiv-1 9