Development and validation of 15-month mortality prediction models: a retrospective observational comparison of
Gregory D Berg1, Virginia F Gurley2
1Analytics, AxisPoint Health, Westminster, Colorado, USA.
Objective:
The objective is to develop and validate a predictive model for 15-month mortality using a random sample of community-dwelling Medicare beneficiaries.
Data Source:
The Centres for Medicare & Medicaid Services' Limited Data Set files containing the five per cent samples for 2014 and 2015.
Participants:
The data analysed contains de-identified administrative claims information at the beneficiary level, including diagnoses, procedures and demographics for 2.7 million beneficiaries.
Setting:
US national sample of Medicare beneficiaries.
Study Design:
Eleven different models were used to predict 15-month mortality risk: logistic regression (using both stepwise and least absolute shrinkage and selection operator (LASSO) selection of variables as well as models using an age gender baseline, Charlson scores, Charlson conditions, Elixhauser conditions and all variables), naïve Bayes, decision tree with adaptive boosting, neural network and support vector machines (SVMs) validated by simple cross validation. Updated Charlson score weights were generated from the predictive model using only Charlson conditions.
Primary Outcome Measure:
C-statistic.
Results:
The c-statistics was 0.696 for the naïve Bayes model and 0.762 for the decision tree model. For models that used the Charlson score or the Charlson variables the c-statistic was 0.713 and 0.726, respectively, similar to the model using Elixhauser conditions of 0.734. The c-statistic for the SVM model was 0.788 while the four models that performed the best were the logistic regression using all variables, logistic regression after selection of variables by the LASSO method, the logistic regression using a stepwise selection of variables and the neural network with c-statistics of 0.798, 0.798, 0.797 and 0.795, respectively.
Conclusions:
Improved means for identifying individuals in the last 15 months of life is needed to improve the patient experience of care and reducing the per capita cost of healthcare. This study developed and validated a predictive model for 15-month mortality with higher generalisability than previous administrative claims-based studies.
Insights
Researchers developed and validated a predictive model for 15-month mortality risk in Medicare beneficiaries. The best models, including logistic regression and neural networks, achieved high accuracy, improving identification of individuals nearing end-of-life.
Area of Science:
- Health Services Research
- Biostatistics
- Gerontology
Background:
- Identifying individuals at high risk of mortality is crucial for improving patient care and reducing healthcare costs.
- Existing predictive models based on administrative claims data have limitations in generalizability.
Purpose of the Study:
- To develop and validate a predictive model for 15-month mortality risk.
- To compare the performance of various machine learning models for mortality prediction.
Main Methods:
- Utilized a random sample of 2.7 million community-dwelling Medicare beneficiaries from 2014-2015.
- Evaluated eleven predictive models including logistic regression, naïve Bayes, decision trees, neural networks, and support vector machines (SVMs).
- Validated models using cross-validation and assessed performance via the c-statistic.
Main Results:
- The Support Vector Machine (SVM) model achieved a c-statistic of 0.788.
- Logistic regression models (all variables, LASSO, stepwise) and the neural network model demonstrated the highest performance with c-statistics ranging from 0.795 to 0.798.
- Models incorporating Charlson or Elixhauser conditions showed moderate predictive accuracy.
Conclusions:
- Developed and validated a highly generalizable predictive model for 15-month mortality using administrative claims data.
- The study highlights the potential of advanced machine learning techniques for improving end-of-life care identification.
- The findings can inform strategies to enhance patient experience and manage healthcare expenditures.
Related Concept Videos
01:12Observational Learning
01:15Techniques of therapeutic communication I: Active Listening, Sharing Observations, Validation, and Using Touch
Therapeutic communication is not the same as social interaction. Social interaction has no goal or purpose and consists of casual information sharing, whereas therapeutic communication has a plan or purpose for the conversation. Therapeutic...
02:57The Sense of Self: Reflected Self-Appraisal and Social Comparison
01:26National Nursing Organizations I
01:29Reliability and Validity
02:30Naturalistic Observations