Objective: To design and assess a method to leverage individuals' temporal data for predicting their healthcare cost. To achieve this goal, we first used patients' temporal data in their fine-grain form as opposed to coarse-grain form. Second, we devised novel spike detection features to extract temporal patterns that improve the performance of cost prediction. Third, we evaluated the effectiveness of different types of temporal features based on cost information, visit information and medical information for the prediction task. Materials and methods: We used three years of medical and pharmacy claims data from 2013 to 2016 from a healthcare insurer, where the first two years were used to build the model to predict the costs in the third year. To prepare the data for modeling and prediction, the time series data of cost, visit and medical information were extracted in the form of fine-grain features (i.e., segmenting each time series into a sequence of consecutive windows and representing each window by various statistics such as sum). Then, temporal patterns of the time series were extracted and added to fine-grain features using a novel set of spike detection features (i.e., the fluctuation of data points). Gradient Boosting was applied on the final set of extracted features. Moreover, the contribution of each type of data (i.e., cost, visit and medical) was assessed. We benchmarked the proposed predictors against extant methods including those that used coarse-grain features which represent each time series with various statistics such as sum and the most recent portion of the values in the entire series. All prediction performances were measured in terms of Mean Absolute Percentage Error (MAPE). Results: Gradient Boosting applied on fine-grain predictors outperformed coarse-grain predictors with a MAPE of 3.02 versus 8.14 (p<0.01). Enhancing the fine-grain features with the temporal pattern extraction features (i.e., spike detection features) further improved the MAPE to 2.04 (p<0.01). Removing cost, visit and medical status data resulted in MAPEs of 10.24, 2.22 and 2.07 respectively (p<0.01 for the first two comparisons and p=0.63 for the third comparison). Conclusions: Leveraging fine-grain temporal patterns for healthcare cost prediction significantly improves prediction performance. Enhancing fine-grain features with extraction of temporal cost and visit patterns significantly improved the performance. However, medical features did not have a significant effect on prediction performance. Gradient Boosting outperformed all other prediction models.

Background Obesity is associated with elevated coronary artery calcium (CAC), a marker of coronary atherosclerosis that is strongly predictive of cardiovascular events. we evaluated the effects of marked weight loss achieved through roux-en-Y gastric bypass surgery (GBS) on CAC scores. Methods We performed echocardiography and computed tomography of the heart in 149 subjects 6 years after enrollment in a prospective registry evaluating the cardiovascular effects of GBS. coronary calcium scores, left ventricular ejection fraction and left ventricular mass were measured. Results At baseline most coronary risk factors were similar between the GBS and nonsurgical groups including current smoking, systolic blood pressure, LDL-C, HDL-C, and TG. However, GBS patients were younger (4.7 years), less likely to be diabetic and less likely to be postmenopausal. At 6 years after enrollment, CAC score was significantly lower in patients who underwent GBS than those without surgery (p<0.01). GBS subjects had a lower likelihood of having measureable coronary calcium (odds ratio of CAC > zero = 0.39; 95% CI of (0.17, 0.90)). Significant predictors of zero CAC were GBS, female gender, younger age, baseline BMI, and baseline LDL-C. Substituting change in BMI for group status as a predictor variable showed that BMI change also predicted CAC (p=0.045). Changes in LDL-C did not predict the CAC differences between groups (p=0.67). Conclusions Sustained weight loss achieved through bariatric surgery is associated with less coronary calcification. This effect, which appears to be independent of changes in LDL-C, may contribute to lower cardiac mortality in patients with successful GBS.

No abstract

Background: Obesity is associated with an increased risk of developing heart failure. Based on cross sectional studies, it has been hypothesized that the duration of obesity is the key factor leading to impaired cardiac function. However, longitudinal data to confirm this hypothesis are not available. Methods: We prospectively studied 62 severely obese patients at baseline, 2 and 5 years after randomization to nonsurgical therapy (NonSurg, n = 25) or Rouxen-Y gastric bypass surgery (GBS, n = 37). Echocardiography was used to measure left ventricular (LV) size and ejection fraction (EF). Results: At enrollment, the mean BMI was 46±9 and the mean age was 47±11 years (range 25– 66). GBS subjects lost 96± 26 vs. 6±18 lbs at 2 years and 78±42 vs. 17±42 lbs at 5 years compared to NonSurg (p<0.0001 for both). At baseline LVEF was not different between GBS and nonsurg (67±9 vs. 64±8%) and it did not change at 2 years (64±9 vs. 63±9%) or 5 years (63±9 vs. 63±10%). LV diastolic dimension did not change over time in control (4.3±1.0 vs. 4.2±0.6 vs. 4.5±0.3) or GBS patients (4.4±0.6 vs. 4.3±0.7 vs. 4.4±0.4). Stratifying the entire group by quartiles of age or duration of obesity (quartile 1 avg duration = 16 years, quartile 4 average duration = 56 years), we found no evidence of time-dependent changes in LV size or function. Conclusion: In this, prospective study of severely obese patients we found no evidence of progressive changes in LV size or EF over a period of 5 years. Moreover, we find no relationship between age or duration of obesity and LV size or LVEF. These data argue strongly that other factors such as the development of coronary disease are the most likely causes of heart failure in obese patients.