Sequential Data–Based Patient Similarity Framework for Patient Outcome Prediction: Algorithm Development

Abstract

Background Sequential information in electronic medical records is valuable and helpful for patient outcome prediction but is rarely used for patient similarity measurement because of its unevenness, irregularity, and heterogeneity. Objective We aimed to develop a patient similarity framework for patient outcome prediction that makes use of sequential and cross-sectional information in electronic medical record systems. Methods Sequence similarity was calculated from timestamped event sequences using edit distance, and trend similarity was calculated from time series using dynamic time warping and Haar decomposition. We also extracted cross-sectional information, namely, demographic, laboratory test, and radiological report data, for additional similarity calculations. We validated the effectiveness of the framework by constructing k–nearest neighbors classifiers to predict mortality and readmission for acute myocardial infarction patients, using data from (1) a public data set and (2) a private data set, at 3 time points—at admission, on Day 7, and at discharge—to provide early warning patient outcomes. We also constructed state-of-the-art Euclidean-distance k–nearest neighbor, logistic regression, random forest, long short-term memory network, and recurrent neural network models, which were used for comparison. Results With all available information during a hospitalization episode, predictive models using the similarity model outperformed baseline models based on both public and private data sets. For mortality predictions, all models except for the logistic regression model showed improved performances over time. There were no such increasing trends in predictive performances for readmission predictions. The random forest and logistic regression models performed best for mortality and readmission predictions, respectively, when using information from the first week after admission. Conclusions For patient outcome predictions, the patient similarity framework facilitated sequential similarity calculations for uneven electronic medical record data and helped improve predictive performance.

Predicting outcomes at the individual patient level: what is the best method?

“…We found that the endto-end and similarity-based approaches showed comparably low predictive performance for any number of clusters k ranging from 2 to 8. This finding is in contrast with previous studies, [8][9][10] where the similarity-based approach led to performance benefits. One possible explanation is that the available clinical and demographic variables were not strong predictors of changes in PHQ-9 scores in our data set.…”

“…The idea behind this approach is that we try to better capture the heterogeneity of patient characteristics by bypassing the need to create a single, complex model, valid for all possible patient subgroups and types. The two-stage method has previously yielded improved predictive performance compared with the usual end-to-end approach in some applications, such as for predicting mortality and readmission in patients with acute myocardial infarction and for diagnosis and outcome prediction in patients with kidney diseases 9 10. It remains however unclear whether applying this method in other settings, such as in mental health outcomes prediction, may also lead to improved performance.…”

Improving the Performance of Outcome Prediction for Inpatients With Acute Myocardial Infarction Based on Embedding Representation Learned From Electronic Medical Records: Development and Validation Study

“…We maintained patients’ first hospitalization data to evaluate the proposed method. Demographic data (age and gender) and the following AMI-related features were maintained in both data sets: AMI-relevant items of laboratory tests that at least 95% of patients carried out, AMI-relevant radiological features extracted from radiology reports [ 34 ], 7 commonly prescribed medications, and all recorded disease diagnoses and procedures. For laboratory tests performed more than once, only the results obtained in the first test (usually at admission) were retained, which could reflect a patient’s health status and the severity of illness.…”

Year 2022 in Medical Natural Language Processing: Availability of Language Models as a Step in the Democratization of NLP in the Biomedical Area

“…Information 40,87,88,89], mortality [90], risks [91], and patient outcomes [89,92,93]. IE and prediction tasks will certainly continue to perform on clinical and health data in the next years.…”