ABSTRACT
BACKGROUND: Remote patient monitoring (RPM) programs augment type 1 diabetes (T1D) care based on retrospective continuous glucose monitoring (CGM) data. Few methods are available to estimate the likelihood of a patient experiencing clinically significant hypoglycemia within one week. METHODS: We developed a machine learning model to estimate the probability that a patient will experience a clinically significant hypoglycemic event, defined as CGM readings below 54 mg/dL for at least 15 consecutive minutes, within one week. The model takes as input the patient's CGM time series over a given week, and outputs the predicted probability of a clinically significant hypoglycemic event the following week. We used 10-fold cross-validation and external validation (testing on cohorts different from the training cohort) to evaluate performance. We used CGM data from three different cohorts of patients with T1D: REPLACE-BG (226 patients), Juvenile Diabetes Research Foundation (JDRF; 355 patients) and Tidepool (120 patients). RESULTS: In 10-fold cross-validation, the average area under the receiver operating characteristic curve (ROC-AUC) was 0.77 (standard deviation [SD]: 0.0233) on the REPLACE-BG cohort, 0.74 (SD: 0.0188) on the JDRF cohort, and 0.76 (SD: 0.02) on the Tidepool cohort. In external validation, the average ROC-AUC across the three cohorts was 0.74 (SD: 0.0262). CONCLUSIONS: We developed a machine learning algorithm to estimate the probability of a clinically significant hypoglycemic event within one week. Predictive algorithms may provide diabetes care providers using RPM with additional context when prioritizing T1D patients for review.
ABSTRACT
BACKGROUND: To examine the influence of solar cycle and geomagnetic effects on SLE disease activity. METHODS: The data used for the analysis consisted of 327 observations of 27-day Physician Global Assessment (PGA) averages from January 1996 to February 2020. The considered geomagnetic indices were the AP index (a daily average level for geomagnetic activity), sunspot number index R (measure of the area of solar surface covered by spots), the F10.7 index (measure of the noise level generated by the sun at a wavelength of 10.7 cm at the earth's orbit), the AU index (upper auroral electrojet index), and high energy (> 60 Mev) proton flux events. Geomagnetic data were obtained from the Goddard Space Flight Center Space Physics Data Facility. A time series decomposition of the PGA averages was performed as the first step. The linear relationships between the PGA and the geomagnetic indices were examined using parametric statistical methods such as Pearson correlation and linear regression, while the nonlinear relationships were examined using nonparametric statistical methods such as Spearman's rho and Kernel regression. RESULTS: After time series deconstruction of PGA averages, the seasonality explained a significant fraction of the variance of the time series (R2 = 38.7%) with one cycle completed every 16 years. The analysis of the short-term (27-day) relationships indicated that increases in geomagnetic activity Ap index (p < 0.1) and high energy proton fluxes (> 60 Mev) (p < 0.05) were associated with decreases in SLE disease activity, while increases in the sunspot number index R anticipated decreases in the SLE disease activity expressed as PGA (p < 0.05). The short-term correlations became statistically insignificant after adjusting for multiple comparisons using Bonferroni correction. The analysis of the long-term (297 day) relationships indicated stronger negative association between changes in the PGA and changes in the sunspot number index R (p < 0.01), AP index (p < 0.01), and the F10.7 index (p < 0.01). The long-term correlations remained statistically significant after adjusting for multiple comparisons using Bonferroni correction. CONCLUSION: The seasonality of the PGA averages (one cycle every 16 years) explains a significant fraction of the variance of the time series. Geomagnetic disturbances, including the level of geomagnetic activity, sunspot numbers, and high proton flux events may influence SLE disease activity. Studies of other geographic locales are needed to validate these findings.
