Predicting Diagnosis of Alzheimer’s Disease and Related Dementias Using Administrative Claims

Abstract

BACKGROUND: Predictive models for earlier diagnosis of Alzheimer's disease and related dementias (ADRD) that rely on variables requiring assessment during an office visit, such as cognitive function, body mass index, or lifestyle factors, may not be broadly applicable, since that level of data may be inaccessible or inefficient. OBJECTIVE: To build a predictive model for earlier diagnosis of ADRD using only administrative claims data to enhance applicability at the health caresystem level. Building on the strength of this approach and knowledge that health care utilization (HCU) is increased before dementia diagnosis, it was hypothesized that previous HCU history would improve predictive ability of the model. METHODS: We conducted a case-control study using data from the OptumLabs Data Warehouse. ADRD was defined using ICD-9-CM codes and prescription fills for antidementia medications. We included individuals with mild cognitive impairment. Cases aged ≥ 18 years with a diagnosis between 2011-2014 were matched to controls without ADRD. HCU variables were incorporated into regression models along with comorbidities and symptoms.RESULTS: The derivation cohort comprised 24,521 cases and 95,464 controls. Final adjusted models were stratified by age. We obtained moderate accuracy (c-statistic = 0.76) for the model among younger (aged < 65 years) adults and poor discriminatory ability (c-statistic = 0.63) for the model among older adults (aged ≥ 65 years). Neurological and psychological disorders had the largest effect estimates. CONCLUSIONS: We created age-stratified predictive models for earlier diagnosis of dementia using information available in administrative claims. These models could be used in decision support systems to promote targeted cognitive screening and earlier dementia recognition for individuals aged < 65 years. These models should be validated in other cohorts.

Machine learning models to predict onset of dementia: A label learning approach

“…Recently, administrative claims data have been used to develop dementia risk models with performance similar to other models in the published literature [9,10]. Models using claims data are more widely available for large populations, offering the potential for their practical use in screening and identifying patients.…”

“…This prior work [9,10,[17][18][19][20][21][22][23][24] does not address the issue of errors in labeling the patients as cases and controls. When a patient's diagnoses are incorrectly assigned, the machine learning algorithm will learn the wrong patterns.…”

“…A number of prior dementia risk models have been published in the peer-reviewed literature [2][3][4][5][6][7][8][9][10]. These have been shown to have moderate discriminatory capabilities (area under the curve ranging from 0.60 to 0.78).…”

Deep neural network models for identifying incident dementia using claims and EHR datasets

“…The univariate odds ratios of some of the diagnosis, procedure and drug features out of thousands of features in the training cohorts with a prevalence of at least 1% in either the cases or controls are presented in Appendix C in S1 File. These features were identified as key predictors of ADRD and MCI in literature [9,11,28] and are presented here to provide insights to the cohorts. Table 2 shows a comparison of the Area Under the Curve (AUC) computed on validation and test data for the best (using grid search) boosted tree, Feed forward network, Recurrent Neural Network and RNN with pre-trained embeddings models for each of the cohorts.…”

“…These models can be broadly divided in two groups. In the first group authors select risk factors based on clinical input and quantify their effect using statistical models [3][4][5][6][7][8][9]. In the second group, which is much smaller than the first, Machine Learning (ML) data-driven models have been developed [10,11] to identify and quantify the risk factors from large healthcare data resources.…”