A Bootstrap Machine Learning Approach to Identify Rare Disease Patients\n from Electronic Health Records

2016/09/06 by Ravi Garg, Garg, Ravi, Shu Dong +6
Computer Science · Health Professions · #Machine Learning in Healthcare #Artificial Intelligence in Healthcare

paper · pdf · doi:10.48550/arxiv.1609.01586

Abstract

Rare diseases are very difficult to identify among large number of other\npossible diagnoses. Better availability of patient data and improvement in\nmachine learning algorithms empower us to tackle this problem computationally.\nIn this paper, we target one such rare disease - cardiac amyloidosis. We aim to\nautomate the process of identifying potential cardiac amyloidosis patients with\nthe help of machine learning algorithms and also learn most predictive factors.\nWith the help of experienced cardiologists, we prepared a gold standard with 73\npositive (cardiac amyloidosis) and 197 negative instances. We achieved high\naverage cross-validation F1 score of 0.98 using an ensemble machine learning\nclassifier. Some of the predictive variables were: Age and Diagnosis of cardiac\narrest, chest pain, congestive heart failure, hypertension, prim open angle\nglaucoma, and shoulder arthritis. Further studies are needed to validate the\naccuracy of the system across an entire health system and its generalizability\nfor other diseases.\n

Citations

• An Introduction to Kernel and Nearest-Neighbor Nonparametric Regression
• Wrappers for feature subset selection

Related

• Machine Learning and Ensemble Approach Onto Predicting Heart Disease
• Predicting prognosis of light-chain cardiac amyloidosis by magnetic resonance imaging and deep learning
• Machine Learning-Based Heart Disease Diagnosis: A Systematic Literature Review
• A Deep Learning-Driven Pipeline for Differentiating Hypertrophic Cardiomyopathy from Cardiac Amyloidosis Using 2D Multi-View Echocardiography
• Multi-class heart disease Detection, Classification, and Prediction using Machine Learning Models