Identifying incident dementia by applying machine learning to a very large administrative claims dataset

Alzheimer's disease and related dementias (ADRD) are highly prevalent conditions, and prior efforts to develop predictive models have relied on demographic and clinical risk factors using traditional logistical regression methods. We hypothesized that machine-learning algorithms using administr...

Full description

Saved in:
Bibliographic Details
Published in:PloS one 2019-07, Vol.14 (7), p.e0203246-e0203246
Main Authors: Nori, Vijay S, Hane, Christopher A, Martin, David C, Kravetz, Alexander D, Sanghavi, Darshak M
Format: Article
Language:English
Subjects:
Adult
Aged
Aged, 80 and over
Aging
Algorithms
Alzheimer's disease
Artificial intelligence
Baby boomers
Biology and Life Sciences
Bipolar disorder
Cognitive ability
Cognitive Dysfunction - diagnosis
Cognitive Dysfunction - epidemiology
Computer and Information Sciences
Data mining
Datasets
Datasets as Topic
Dementia
Dementia - diagnosis
Dementia - epidemiology
Dementia disorders
Demographic variables
Demographics
Diabetes
Diagnosis
Drugstores
Female
Geriatrics
Hallucinogenic drugs
Heart failure
Humans
Incidence
Laboratories
Learning algorithms
Machine Learning
Male
Massive data points
Medical diagnosis
Medicine and Health Sciences
Memantine
Middle Aged
Neurodegenerative diseases
Neurosciences
NMR
Nuclear magnetic resonance
Parkinson's disease
Pharmacy
Physical Sciences
Population
Prediction models
Psychotropic drugs
Research and Analysis Methods
Risk analysis
Risk factors
Teaching methods
Type 2 diabetes
Variables
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Alzheimer's disease and related dementias (ADRD) are highly prevalent conditions, and prior efforts to develop predictive models have relied on demographic and clinical risk factors using traditional logistical regression methods. We hypothesized that machine-learning algorithms using administrative claims data may represent a novel approach to predicting ADRD. Using a national de-identified dataset of more than 125 million patients including over 10,000 clinical, pharmaceutical, and demographic variables, we developed a cohort to train a machine learning model to predict ADRD 4-5 years in advance. The Lasso algorithm selected a 50-variable model with an area under the curve (AUC) of 0.693. Top diagnosis codes in the model were memory loss (780.93), Parkinson's disease (332.0), mild cognitive impairment (331.83) and bipolar disorder (296.80), and top pharmacy codes were psychoactive drugs. Machine learning algorithms can rapidly develop predictive models for ADRD with massive datasets, without requiring hypothesis-driven feature engineering.
ISSN:1932-6203
1932-6203
DOI:10.1371/journal.pone.0203246