Identification of patients at risk of new onset heart failure: Utilizing a large statewide health information exchange to train and validate a risk prediction model

Abstract

Background New-onset heart failure (HF) is associated with poor prognosis and high healthcare utilization. Early identification of patients at increased risk incident-HF may allow for focused allocation of preventative care resources. Health information exchange (HIE) data span the entire spectrum of clinical care, but there are no HIE-based clinical decision support tools for diagnosis of incident-HF. We applied machine-learning methods to model the one-year risk of incident-HF from the Maine statewide-HIE. Methods and results We included subjects aged ≥ 40 years without prior HF ICD9/10 codes during a three-year period from 2015 to 2018, and incident-HF defined as assignment of two outpatient or one inpatient code in a year. A tree-boosting algorithm was used to model the probability of incident-HF in year two from data collected in year one, and then validated in year three. 5,668 of 521,347 patients (1.09%) developed incident-HF in the validation cohort. In the validation cohort, the model c-statistic was 0.824 and at a clinically predetermined risk threshold, 10% of patients identified by the model developed incident-HF and 29% of all incident-HF cases in the state of Maine were identified. Conclusions Utilizing machine learning modeling techniques on passively collected clinical HIE data, we developed and validated an incident-HF prediction tool that performs on par with other models that require proactively collected clinical data. Our algorithm could be integrated into other HIEs to leverage the EMR resources to provide individuals, systems, and payors with a risk stratification tool to allow for targeted resource allocation to reduce incident-HF disease burden on individuals and health care systems.

Artificial Intelligence Models in Health Information Exchange: A Systematic Review of Clinical Implications

“…Overall, eleven articles were reviewed, encompassing various designs. Specifically, there were seven cohort studies [ 29 , 30 , 33 , 34 , 36 , 38 , 39 ], of which three were retrospective [ 29 , 36 , 39 ]. In addition, there were three retrospective observational studies [ 32 , 35 , 37 ] and one study that focused on developing and evaluating a model [ 31 ].…”

“…Type of parameters obtained from the patient’s records were demographics [ 29 , 34 , 35 , 36 , 39 ], risk factors [ 30 , 33 , 34 , 35 ], radiology encounters [ 29 , 34 ], histology and cytology reports [ 31 , 32 ], laboratory test results [ 34 , 37 , 38 ], visit history (ED, outpatient and, inpatient) [ 29 , 30 , 35 , 36 , 39 ], medications [ 33 , 34 , 35 , 38 , 39 ], comorbidities [ 29 , 30 , 35 , 39 ], and social determinants of health [ 35 , 38 , 39 ].…”

Artificial intelligence and heart failure: A state‐of‐the‐art review

“…Early initiation of HF treatment is important as benefits appear soon after initiation (Table 2). [10][11][12][13][14][15][16][17][18][19] In a cohort study of 497 470 patients from Maine health information exchange network, XGBoost predicted the incidence of HF well in the validation cohort (C-statistics: 0.824) using 339 predictor variables. 17 However, due to large number of predictor variables used in the study, the generalizability of high-dimensional ML models can still be constrained, thus caution should be exercised when extrapolating results to fresh datasets.…”

“…[10][11][12][13][14][15][16][17][18][19] In a cohort study of 497 470 patients from Maine health information exchange network, XGBoost predicted the incidence of HF well in the validation cohort (C-statistics: 0.824) using 339 predictor variables. 17 However, due to large number of predictor variables used in the study, the generalizability of high-dimensional ML models can still be constrained, thus caution should be exercised when extrapolating results to fresh datasets. With its ability to handle both regression and classification tasks simultaneously, random forest algorithms can handle large datasets efficiently, with high levels of predictive accuracy.…”

Risk of heart failure depending on the state of renal filtration function in patients with uncomplicated hypertension

“…Поэтому для предупреждения развития ХСН у больных гипертонической болезнью (ГБ) и другими ФР ряд авторов рекомендуют раннее выявление пациентов с повышенным риском развития ХСН, что может не только улучшить прогноз, но и позволит целенаправленно распределять ресурсы на профилактическую помощь [5].…”