To develop a noninvasive hypoglycemia detection approach using smartwatch data.
We prospectively collected data from two wrist-worn wearables (Garmin vivoactive 4S, Empatica E4) and continuous ...glucose monitoring values in adults with diabetes on insulin treatment. Using these data, we developed a machine learning (ML) approach to detect hypoglycemia (<3.9 mmol/L) noninvasively in unseen individuals and solely based on wearable data.
Twenty-two individuals were included in the final analysis (age 54.5 ± 15.2 years, HbA1c 6.9 ± 0.6%, 16 males). Hypoglycemia was detected with an area under the receiver operating characteristic curve of 0.76 ± 0.07 solely based on wearable data. Feature analysis revealed that the ML model associated increased heart rate, decreased heart rate variability, and increased tonic electrodermal activity with hypoglycemia.
Our approach may allow for noninvasive hypoglycemia detection using wearables in people with diabetes and thus complement existing methods for hypoglycemia detection and warning.
Abstract
Direct alcohol biomarkers are of growing interest for the assessment of alcohol consumption, with particular interest in phosphatidylethanol (PEth) in recent years. PEth is only formed when ...alcohol is present in the body. However, there is no statement about how much the PEth concentration increases after single moderate alcohol consumption. This study was conducted to determine the increase in PEth concentrations after a single drinking event. Additionally, a new volumetric sampling device (volumetric dried blood spot cards (DBSV)) was evaluated, which was designed to simplify further sampling processes and to allow for easy self-sampling. Dried blood samples from 31 volunteers were collected before and after single alcohol consumption with a mean maximum breath alcohol concentration of 0.4 mg/L (range: 0.30–0.55 mg/L). PEth concentrations were determined after automated extraction by liquid chromatography--tandem mass spectrometry. PEth 16:0/18:1 and PEth 16:0/18:2 concentrations increased to an average of 45 ng/mL each in patients starting below 20 ng/mL (range: 25.0–57.0 ng/mL for PEth 16:0/18:1; range 26.8–62.3 ng/mL for PEth 16:0/18:2). PEth concentrations in patients starting above 20 ng/mL increased by a mean of 30 ng/mL (range: 6.2–71.3 ng/mL for PEth 16:0/18:1; range 8.8–65.3 ng/mL for PEth 16:0/18:2). In addition, the comparison of the new sampling device DBSV with a standard filter paper card (with volumetrically applied 20 µL of blood samples) yielded a close agreement for the determined PEth concentrations in 24 forensic samples and three external controls. Therefore, the sampling device DBSV proved to be suitable for the determination of PEth concentrations in blood.
PurposeThe emergence of Internet of Things (IoT) platforms in product companies opens up new data-driven business opportunities. This paper looks at the emergence of these IoT platforms from a ...business-model perspective.Design/methodology/approachThe study applies a mixed method with two research studies: Study I–a cluster analysis based on a quantitative survey, and Study II–case studies based on qualitative interviews.FindingsThe findings reveal that there is no gradual shift in a company's business model, but in fact three distinct and sequential patterns of business model innovations: (1) platform skimming, (2) platform revenue generation and (3) platform orchestration.Research limitations/implicationsThe results are subject to the typical limitations of both quantitative and qualitative studies.Practical implicationsThe results provide guidance to managers on how to modify the components of the business model (value proposition, value creation and/or delivery and profit equation) in order to enable platforms to advance.Social implicationsAs IoT platforms continue to advance, product companies achieve better performance in terms of productivity and profitability, and more easily secure competitive advantages and jobs.Originality/valueThe paper makes three original contributions: (1) it is the first quantitative study on IoT platforms in product companies, (2) identifies three patterns of business model innovations and (3) offers a first process perspective for understanding the sequence of these patterns as IoT platforms advance.
Background: Due to the high rate of asymptomatic patients, the impact of postprandial hypoglycemia after bariatric surgery (PBH) on daily activities such as driving remains controversial. We sought ...to assess driving performance in patients with PBH. Methods: In a random single-blind crossover study, 10 active drivers with PBH (2 males, age 38±15years, BMI 27.2±4.6kg/m2, gastric bypass 5±2years ago) ingested 75g glucose (GLU) to induce hypoglycemia or aspartame (ASP). A simulator was driven during 10min before and at 20 (D1), 80 (D2), 125 (D3) and 140min (D4) after GLU/ASP, reflective of expected blood glucose (BG) increase (D1), decrease (D2), and hypoglycemia (D3 and D4). Seven driving features (traffic violations and high accelerations) were integrated in a Bayesian hierarchical regression model to assess altered driving performance in GLU vs. ASP. Results: Mean peak and nadir BG after GLU were 182±24 and 47±14mg/dl, respectively. BG was stable after ASP (85±4mg/dl). Driving performance differed between GLU and ASP at times of hypoglycemia (Figure 1), particularly D4, whereas no differences were found for D1 and D2. The posterior probability for impaired driving after GLU was 16% in D1, 82% for D2, 81% for D3 and 98% for D4. Conclusion: Our findings suggest impaired driving performance during PBH.
This research investigates the impact of feature additions on the use of an information system's (IS) existing core features. Based on prior work in marketing and IS, we hypothesize conflicting ...effects on the usage of the system as a whole and the IS core due to the goal congruence of the two feature sets. In three consecutive empirical studies, we consider the example of a utilitarian consumer IS in the form of a mobile insurance app with additional weather-related functionality. The statistical results indicate that the goal-congruent feature addition exerts a positive influence on system use, whereas the impact on core IS use is negative. More specifically, we show that the latter effect can be explained by changes in the users' percep tions of the usefulness and ease of use of the core features. From a theoretical perspective, our work goes beyond the predominant system view of technology acceptance and use by employing a more fine-grained, feature-oriented level of investigation, which opens several avenues for further research regarding the relationships between information systems and the features they comprise. From a managerial perspective, the results help to characterize the detrimental effects that feature additions may have on IS usage. These consequences become particularly relevant when revenue, cost savings, or other benefits on the part of IS operators are linked only to a subset of the entire IS functionality, as in the case of certain web portals or mobile apps.
•High jerk events measured from CAN Bus data of a nationwide field study of 72 drivers.•Negative binomial regressions indicate relationship between crash rate and jerk rate.•Connected fleet ...demonstrated to be representative of the overall vehicle population.•Crash frequency predicted through regressions on jerk rate and fleet trip frequency.•Spatial regression of crash frequency across the majority of the Swiss road network.
Despite the fact that semi-autonomous vehicles will become more and more prevalent in the coming decades, recent studies have highlighted that traffic accidents will persist as a core issue for road users, insurers, and policy makers alike. Researchers and industry players see potential in the technology embedded in semi-autonomous vehicles to combat this challenge by reliably predicting locations with a high likelihood of traffic accidents. This technology can be leveraged to detect accidents and ‘near miss incidents’, such as heavy braking and evasive manoeuvres, otherwise known as Critical Driving Events (CDEs). The locations of CDEs could identify areas of high accident exposure, offering automotive insurers a unique opportunity to reduce traffic accidents through the adoption of active loss prevention business models, such as providing safe-routing services and in-vehicle warnings. To date, there is limited empirical evidence on whether the Crash Frequency and Crash Rate of locations can be accurately identified through CDEs. To address this research gap, an 18-week naturalistic driving field study of 72 vehicles was conducted in Switzerland, covering over 690,000 km. Data collected from the CAN Bus of these vehicles indicate that there is a proportional relationship between the CDEs of the fleet, and the Crash Frequency and Crash Rate of a location. Furthermore, a nationwide spatial regression analysis was applied to determine Crash Frequency across the majority of the Swiss road network. We identify the relationship between Crash Frequency, and the CDEs and Trip Frequency of the fleet, along with additional explanatory variables for urban and highway locations. These insights provide first evidence that insurance companies and other industry players with access to a nationwide semi-autonomous fleet can determine existing and emerging locations of high accident probability, enabling more proactive business models and safety focused services.
•Comparison of numerical vs. symbolic eco-driving feedback in a large randomized control field trial over 10 weeks.•Only the symbolic eco-driving feedback led to significant fuel reduction of ...2–3%.•Effect is stable when controlling for the influence of road attributes and other covariates.
Despite the fact that more and more car dashboards are being equipped with powerful, high-resolution displays, allowing for radically new ways to design driving feedback, the question of what impact different design types and features have on real-world eco-driving remains largely unclear. To address this research gap, we conducted a randomized control field trial in Switzerland with 62 road assistance drivers over a period of 10 weeks, covering over 245,000 km. We evaluate the effect of eco-driving feedback on fuel consumption for two types of feedback: numerical feedback (which uses numbers and gauges to present numerical values) and symbolic feedback (which translates numerical values into symbolic representations). Both, numeric and symbolic eco-driving feedback were tested against a control group. Data analyses are performed on the level of 265,939 dynamic road segments with constant road characteristics to account for the significant effect of road attributes on fuel consumption. Results of a fixed-effects regression models reveal that only the symbolic feedback design led to significant reductions of 2–3% in fuel consumption. The effect is robust across different model specifications that control for the influence of road attributes and other covariates. We conclude that the design of eco-driving feedback can have a significant impact on its effectiveness for promoting a less fuel-consuming driving style. We conjecture that there is a large untapped potential for manufacturers to use modern digitalized dashboards that can improve the impact of driver feedback systems.
Aim
To develop and evaluate the concept of a non‐invasive machine learning (ML) approach for detecting hypoglycaemia based exclusively on combined driving (CAN) and eye tracking (ET) data.
Materials ...and Methods
We first developed and tested our ML approach in pronounced hypoglycaemia, and then we applied it to mild hypoglycaemia to evaluate its early warning potential. For this, we conducted two consecutive, interventional studies in individuals with type 1 diabetes. In study 1 (n = 18), we collected CAN and ET data in a driving simulator during euglycaemia and pronounced hypoglycaemia (blood glucose BG 2.0‐2.5 mmol L−1). In study 2 (n = 9), we collected CAN and ET data in the same simulator but in euglycaemia and mild hypoglycaemia (BG 3.0‐3.5 mmol L−1).
Results
Here, we show that our ML approach detects pronounced and mild hypoglycaemia with high accuracy (area under the receiver operating characteristics curve 0.88 ± 0.10 and 0.83 ± 0.11, respectively).
Conclusions
Our findings suggest that an ML approach based on CAN and ET data, exclusively, enables detection of hypoglycaemia while driving. This provides a promising concept for alternative and non‐invasive detection of hypoglycaemia.