Population Cost Prediction on Public Healthcare Datasets

Sushmita, Shanu; Newman, Stacey; Marquardt, James; Ram, Prabhu; Prasad, Viren; Cock, Martine De; Teredesai, Ankur

doi:10.1145/2750511.2750521

Cited by 48 publications

(45 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The chart shows that the score has the same distribution as described in [6], with a spike at 0 and a long right-hand tail as expected for health care cost. It has been reported [14,16,17] that the use of clinical features yields the same performance as using only cost predictors. Despite clinical information seems not to affect prediction performance, we prefer to keep it, because having it in the model could increase the number of dimensions which may improve vector differentiation.…”

Section: Data and Problem Descriptionmentioning

confidence: 99%

“…Supervised learning methods have been vastly used to predict health care costs; the data used for these methods vary. While a few works use only demographic and clinical information (e.g., diagnosis groups, number of admissions and number of laboratory tests) [13], the majority have incorporated cost inputs (e.g., previous total costs, previous medication costs) as well [14][15][16][17], obtaining better performance. GB [18] excels as the method with the best performance for this problem [17], which is an ensemble-learning algorithm, where the final model is an ensemble of weak regression tree models, which are built in a forward stage-wise fashion.…”

Section: Health Care Cost Predictionmentioning

confidence: 99%

“…ANNs come close to the performance of GB, ANNs are an extensive collection of processing units (i.e., neurons), where each unit is connected with many others; ANNs typically consist of multiple layers, and some goal is to solve problems in the same way that the human brain would do it [20]. Another type of model with good results is the M5 Tree [16]; this algorithm is also a Regression Tree, where a Linear Regression Model is used for building the model and calculating the sum of errors as opposed to the mean [21].…”

Section: Health Care Cost Predictionmentioning

confidence: 99%

“…Naive-Bayes has been used to create a prediction system for heart disease [25]. Regression and decision trees have been applied to a variety of problems [16,26,27]. Some authors have used the Dempster-Shafer theory, also called the theory of belief functions, which is a generalization of the Bayesian theory, like Maseleno et al [28], who created an expert system for the detection of skin diseases and Peñafiel et al [29] who associated the risk of getting a stroke with health checkup data.…”

Section: Interpretabilitymentioning

confidence: 99%

“…As shown by [6,16], it is challenging to predict patients' health care costs by only using clinical information. The best indicator for future health care costs are previous costs: the additional history of health care expenses is known to improve the prediction.…”

Section: Data and Problem Descriptionmentioning

confidence: 99%

See 4 more Smart Citations

Predicting Health Care Costs Using Evidence Regression

Panay

Baloian

Pino

et al. 2019

13th International Conference on Ubiquitous Computing and Ambient ‪Intelligence UCAmI 2019‬

View full text Add to dashboard Cite

People's health care cost prediction is nowadays a valuable tool to improve accountability in health care. In this work, we study if an interpretable method can reach the performance of black-box methods for the problem of predicting health care costs. We present an interpretable regression method based on the Dempster-Shafer theory, using the Evidence Regression model and a discount function based on the contribution of each dimension. Optimal parameters are learned using gradient descent. The k-nearest neighbors' algorithm was also used to speed up computations. With the transparency of the evidence regression model, it is possible to create a set of rules based on a patient's vicinity. When making a prediction, the model gives a set of rules for such a result. We used Japanese health records from Tsuyama Chuo Hospital to test our method, which includes medical checkups, exam results, and billing information from 2016 to 2017. We compared our model to an Artificial Neural Network and Gradient Boosting method. Our results showed that our transparent model outperforms the Arti cial Neural Network and Gradient Boosting with an R 2 of 0.44.

show abstract

Section: Data and Problem Descriptionmentioning

confidence: 99%

Section: Health Care Cost Predictionmentioning

confidence: 99%

Section: Health Care Cost Predictionmentioning

confidence: 99%

Section: Interpretabilitymentioning

confidence: 99%

Section: Data and Problem Descriptionmentioning

confidence: 99%

See 3 more Smart Citations

Predicting Health Care Costs Using Evidence Regression

Panay

Baloian

Pino

et al. 2019

13th International Conference on Ubiquitous Computing and Ambient ‪Intelligence UCAmI 2019‬

View full text Add to dashboard Cite

show abstract

A Cloud-Based Platform for ECG Monitoring and Early Warning Using Big Data and Artificial Intelligence Technologies

Zhou

Zhang

et al. 2020

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Machine learning versus regression modelling in predicting individual healthcare costs from a representative sample of the nationwide claims database in France

Vimont

Leleu²,

Durand-Zaleski

2021

Eur J Health Econ

View full text Add to dashboard Cite

Background Innovative provider payment methods that avoid adverse selection and reward performance require accurate prediction of healthcare costs based on individual risk adjustment. Our objective was to compare the performances of a simple neural network (NN) and random forest (RF) to a generalized linear model (GLM) for the prediction of medical cost at the individual level. Methods A 1/97 representative sample of the French National Health Data Information System was used. Predictors selected were: demographic information; pre-existing conditions, Charlson comorbidity index; healthcare service use and costs. Predictive performances of each model were compared through individual-level (adjusted R-squared (adj-R 2 ), mean absolute error (MAE) and hit ratio (HiR)), and distribution-level metrics on different sets of covariates in the general population and by pre-existing morbid condition, using a quasi-Monte Carlo design. Results We included 510,182 subjects alive on 31st December, 2015. Mean annual costs were 1894€ (standard deviation 9326€) (median 393€, IQ range 95€; 1480€), including zero-claim subjects. All models performed similarly after adjustment on demographics. RF model had better performances on other sets of covariates (pre-existing conditions, resource counts and past year costs). On full model, RF reached an adj-R 2 of 47.5%, a MAE of 1338€ and a HiR of 67%, while GLM and NN had an adj-R 2 of 34.7% and 31.6%, a MAE of 1635€ and 1660€, and a HiR of 58% and 55 M, respectively. RF model outperformed GLM and NN for most conditions and for high-cost subjects. Conclusions RF should be preferred when the objective is to best predict medical costs. When the objective is to understand the contribution of predictors, GLM was well suited with demographics, conditions and base year cost.

show abstract

Population Cost Prediction on Public Healthcare Datasets

Cited by 48 publications

References 11 publications

Predicting Health Care Costs Using Evidence Regression

Predicting Health Care Costs Using Evidence Regression

A Cloud-Based Platform for ECG Monitoring and Early Warning Using Big Data and Artificial Intelligence Technologies

Machine learning versus regression modelling in predicting individual healthcare costs from a representative sample of the nationwide claims database in France

Contact Info

Product

Resources

About