No abstract

Background Type 2 diabetes (T2D) is the seventh leading cause of death (2017) in the United States, and by 2030 it is estimated that it will affect 439 million globally. Effective glycemic control can be challenging for patients. A tool to guide patients’ in their self-management behaviors and share this data with their physician may improve insulin adherence leading to lower HbA1c. We examined an integrated diabetes management (IDM) system that utilizes a Bluetooth-enabled insulin event capture device, a Bluetooth-enabled glucometer, and an Android smartphone app. IDM data can be viewed by clinicians in the electronic medical record (EMR). Objective The primary aim of this study is to describe how app use is related to insulin adherence, blood glucose measurements, meal snapshots, and step count. Secondarily, we assessed the impact on HbA1c levels over a 3- and 6-month period. Methods Thirty-five participants were enrolled from Boston-area hospitals in this single-arm pilot study. Use of the IDM system was defined as the number of days per week participants logged into the app and moved past the home screen. Three app use groups were created: low app use (0.33-2.46 days per week), medium app use (2.54-5.08 days per week), and high app use (>5.4 days per week). Adherence to insulin, blood glucose measurements, and meal snapshots were defined as a ratio of actual weekly events recorded by participants’ app use divided by their physician’s recommendation. Step count was defined as the total weekly steps for each participant. Daily app-generated data on app use and indicators of diabetes management were collected. HbA1c levels were assessed via blood test at enrollment, 3-months, and 6-months. Using a hierarchical linear mixed model, we examined changes in outcome measures while accounting for random intercepts and slopes to control for variation in individual outcomes over the study. Results Overall app use (average unique days using the app per week) declined from 6.19 days to 3.00 days (at 1 and 24 weeks, respectively). Participants with high app use had significant improvement in bolus and basal insulin adherence per week (0.009 P=.041 [95% CI 0.0004 to 0.018] and 0.016 P<.001 [95% CI 0.0079 to 0.023], respectively), but participants had no significant improvements in blood glucose and meal snapshot adherence or absolute step count. HbA1c significantly decreased per week (coefficient –0.025 [95% CI –0.044 to –0.007], P=.007) with an overall change of 0.6. Participants with high app use significantly improved their HbA1c per week (–0.037 P=.016 [–0.066 to –0.0067]) compared to participants with medium and low app use, yielding a total improvement of 0.88 over 24 weeks. Conclusions Results show that bolus and basal insulin may have increased with higher app use. HbA1c significantly improved over the course of the study, along with significantly greater improvement in HbA1c among participants with higher app use compared to participants in the middle or low app use groups. This study is not designed or intended to evaluate efficacy but provides results to guide the future design and development of this prototype IDM system.

Background Wellness devices for health tracking have gained popularity in recent years. Additionally, portable and readily accessible wellness devices have several advantages when compared to traditional medical devices found in clinical environments The VitalWellness device is a portable wellness device that can potentially aide vital sign measuring for those interested in tracking their health. Objective In this diagnostic accuracy study, we evaluated the performance of the VitalWellness device, a wireless, compact, non-invasive device that measures four vital signs (blood pressure (BP), heart rate (HR), respiratory rate (RR), and body temperature using the index finger and forehead. Methods Volunteers age ≥18 years were enrolled to provide blood pressure (BP), heart rate (HR), respiratory rate (RR), and body temperature. We recruited participants with vital signs that fell within and outside of the normal physiological range. A sub-group of eligible participants were asked to undergo an exercise test, aerobic step test and/or a paced breathing test to analyze the VitalWellness device’s performance on vital signs outside of the normal physiological ranges for HR and RR. Vital signs measurements were collected with the VitalWellness device and FDA-approved reference devices. Mean, standard deviation, mean difference, standard deviation of difference, standard error of mean difference, and correlation coefficients were calculated for measurements collected; these measurements were plotted on a scatter plot and a Bland-Altman plot. Sensitivity analyses were performed to evaluate the performance of the VitalWellness device by gender, skin color, finger size, and in the presence of artifacts. Results 265 volunteers enrolled in the study and 2 withdrew before study completion. Majority of the volunteers were female (62%), predominately white (63%), graduated from college or post college (67%), and employed (59%). There was a moderately strong linear relationship between VitalWellness BP and reference BP (r=0.7, P<.05) and VitalWellness RR and reference RR measurements (r=0.7, P<.05). The VitalWellness HR readings were significantly in line with the reference HR readings (r=0.9, P<.05). There was a weaker linear relationship between VitalWellness temperature and reference temperature (r=0.3, P<.05). There were no differences in performance of the VitalWellness device by gender, skin color or in the presence of artifacts. Finger size was associated with differential performance for RR. Conclusions Overall, the VitalWellness device performed well in taking BP, HR, and RR when compared to FDA-approved reference devices and has potential serve as a wellness device. To test adaptability and acceptability, future research may evaluate user’s interactions and experiences with the VitalWellness device at home. In addition, the next phase of the study will evaluate transmitting vital sign information from the VitalWellness device to an online secured database where information can be shared with HCPs within seconds of measurement.

Background Wellness devices for health tracking have gained popularity in recent years. Additionally, portable and readily accessible wellness devices have several advantages when compared to traditional medical devices found in clinical environments. Building tools for patients to manage their health independently may benefit their health in the long run by improving health care providers’ (HCPs) awareness of their patients’ health information outside of the clinic. Increased access to portable wellness devices that track vital signs may increase how patients and HCPs track and monitor chronic conditions which can improve health outcomes. The VitalWellness is a portable wellness device that can potentially aid vital sign measuring for those interested in tracking their health. Objective In this diagnostic accuracy study, we evaluated the clinical performance of the VitalWellness, a wireless, compact, non-invasive device that measures four vital signs using the index finger and forehead against reference vital signs devices used in the hospital setting. Methods Volunteers age ≥18 years were enrolled to provide blood pressure (BP), heart rate (HR), respiratory rate (RR), and body temperature. We recruited volunteers with vital signs that fell within and outside of the normal physiological range, depending on the measurements they consented to undergo. A subgroup of eligible volunteers were asked to undergo an exercise test, aerobic step test and/or a paced breathing test to analyze the VitalWellness device's performance on vital signs outside of the normal physiological ranges for HR and RR. Vital signs measurements were collected with the VitalWellness device and FDA-approved reference devices. Mean, standard deviation, mean difference, standard deviation of difference, standard error of mean difference, and correlation coefficients were calculated for measurements collected; these measurements were plotted on a scatter plot and a Bland-Altman plot. Sensitivity analyses were performed to evaluate the performance of the VitalWellness device by gender, skin color, finger size, and in the presence of artifacts. Results We enrolled 265 volunteers in the study and 2 withdrew before study completion. The majority of volunteers were female (62%), predominately white (63%), graduated from college or post college (67%), and employed (59%). There was a moderately strong linear relationship between VitalWellness BP and reference BP (r=0.7, P<.05) and bewteen VitalWellness RR and reference RR measurements (r=0.7, P<.05). The VitalWellness HR readings were significantly in line with the reference HR readings (r=0.9, P<.05). There was a weaker linear relationship between VitalWellness temperature and reference temperature (r=0.3, P<.05). There were no differences in performance of the VitalWellness device by gender, skin color or in the presence of artifacts. Finger size was associated with differential performance for RR. Conclusions Overall, the VitalWellness device performed well in taking BP, HR and RR when compared to FDA-approved reference devices and has potential serve as a wellness device. To test adaptability and acceptability, future research may evaluate user’s interactions and experiences with the VitalWellness device at home. In addition, the next phase of the study will evaluate transmitting vital sign information from the VitalWellness device to an online secured database where information can be shared with HCPs within seconds of measurement.

Background: Heart failure (HF) patients have a high readmission rate with approximately 20% of patients being readmitted within 30-days after discharge. Hospital interventions to reduce HF readmissions are resource-and effort-intensive. Widespread availability of electronic medical record data has spurred interest in using machine learning-based techniques for risk stratification of heart failure patients. The predictive performance of machine learning-based predictive models is often evaluated solely using the Area Under the Receiver Operating Characteristic (AUROC) curve. However, the AUROC is independent of prevalence therefore predictive models with the same AUROC can have differential clinical utility. Furthermore, the AUROC does not provide any insight about the presence of overfitting or decay in predictive performance of a model over time, both of which can affect its real-world performance. Objective: Our primary objective is to assess real-world performance of a 30-day readmission risk prediction model for HF patients, which had an AUROC of 0.71 in the training dataset. Methods: Predictions for risk of 30-day readmissions in HF patients in the Partners Healthcare System were prospectively obtained from the model. We assessed the positive (PPV) and negative predictive value (NPV), in addition to sensitivity, specificity, accuracy, model calibration and Brier score. Results: Four hundred twenty index admissions that were not part of the training dataset were included in this prospective evaluation. Readmission rate was 24% (101 30-day readmissions). The AUROC of the predictive model was 0.57. At a discrimination threshold of 0.2 for flagging high-risk index admissions, the sensitivity and specificity of the model were 53.46% and 63.32%, respectively. The PPV and NPV were 31.57% and 81.12%, respectively. The Brier score was 0.19. Conclusions: Our analysis offers important insights about the real-world performance of this predictive model. The NPV suggests that the model's prediction about patients at low risk for readmission are reliable. This insight can be useful in optimizing resource allocation for patients with heart failure.