Publications

Rigorous blood glucose management is vital for individuals with diabetes to prevent states of too low blood glucose (hypoglycemia). While there are continuous glucose monitors available, they are expensive and not available for many patients. Related work suggests a correlation between the blood glucose level and physiological measures, such as heart rate variability. We therefore propose a machine learning model to detect hypoglycemia on basis of data from smartwatch sensors gathered in a proof-of-concept study. In further work, we want to integrate our model in wearables and warn individuals with diabetes of possible hypoglycemia. However, presenting just the detection output alone might be confusing to a patient especially if it is a false positive result. We thus use SHAP (SHapley Additive exPlanations) values for feature attribution and a method for subsequently explaining the model decision in a comprehensible way on smartwatches.

Access paper here: http://dx.doi.org/10.1145/3334480.3382808

The globally increasing prevalence and incident rates of mental diseases is one of the most serious public health challenges according to the World Health Organization. Today, treatment is based on professional therapies which require a high amount of financial resources and personnel effort, however IT-supported interventions in ubiquitous devices promise help and a new leverage beyond traditional therapies. We identify the car as a space for new treatments since drivers often have time and the environment in the automobile is highly controlled. In-vehicle information systems can reach people in their daily routine and could introduce innovative prevention measures. In this research in progress paper, we address the open question how the car can improve a driver’s affective state while driving. First, we thoroughly describe the design of a study we conducted to motivate other researchers for this topic. Second, we analyse 631 completed interventions collected in a 2-month field study with 10 drivers. First analyses indicate that we can positively influence the short-term affective state of drivers with at least one of ourintervention types. We provide first practical examples of how to reach the masses of everyday drivers.

Access paper here: https://aisel.aisnet.org/ecis2020_rip/8/

The formation of IT companies and even of entire new technological ecosystems depends heavily on external financing. Consequently, the IS community has intensely studied various financing sources such as venture capital, initial public offerings or debt. Blockchain technology has led to the emergence of a system of decentralized finance (DeFi) which includes decentralized versions of equity and debt financing. In particular, equity-like fundraisings referred to as initial coin offerings (ICO) have received serious traction. In this paper, we investigate the role of user-generated content (UGC) for ICO success. Specifically, we leverage signaling theory to analyze how the activity on blogs and discussion forums is related to the amount of capital raised and the valuation in ICOs. We analyze data of 216 ICOs and provide first results indicating the importance of discussion forum activity for ICO success. Furthermore, we find that blogs seem less relevant than in traditional finance.

Equipment manufacturers are currently utilizing new digital technologies such as the Internet of Things (IoT), Artificial Intelligence, or Big Data, for new digital offerings. However, these offerings seldom enhance revenue, because companies struggle with business model (BM) dynamics. By analyzing 27 companies through an explorative case-study approach, the authors consider how companies can successfully achieve revenue enhancement through digital offerings. The result is a threefold framework for revenue enhancement through digital offerings. First, this framework distinguishes between three phases of BM dynamics: 1) augmenting products through a “hardware plus” logic, 2) developing a portfolio of multiple logics for creating customer value, 3) integrating this portfolio through platform logic. Second, the framework emphasizes that three barriers, which we refer to as confidence, mixing, and collaboration barrier, limit the progress from Phases 1 to 3. Third, the framework reveals that each phase contains certain modifications of BM components. In the first phase, companies adapt their BM components slightly, so as to advance toward a “hardware plus” logic. In the second phase, companies embrace more radical BM innovations in order to convert services into an outcome-based BM and develop a new software subscription BM. In the third phase, companies modify BM components in order to integrate the BMs internally and to open them up for external collaboration partners.

Access the paper here: https://doi.org/10.1016/j.indmarman.2020.10.006

Peer-to-peer (P2P) energy markets are gaining interest in the energy sector as a means to increase the share of decentralised energy resources (DER), thus fostering a clean, resilient and decentralised supply of energy. Various reports have touted P2P energy markets as ideal use case for blockchain-technology, as it offers advantages such as fault-tolerant operation, trust elegation, immutability, transparency, resilience, and automation. However, relatively little is known about the influence of hardware and communication infrastructure limitations on blockchain systems in real-life applications. In this article, we demonstrate the implementation of a real-world blockchain managed microgrid in Walenstadt, Switzerland. The 37 participating households are equipped with 75 special smart-meters that include single board computers (SBC) that run their own, application-specific private blockchain. Using the field-test setup, we provide an empirical evaluation of the feasibility of a Byzantine fault tolerant blockchain system. Furthermore, we artificially throttle bandwidth between nodes to simulate how the bandwidth of communication infrastructure impacts its performance. We find that communication networks with a bandwidth smaller than 1000 kbit/s - which includes WPAN, LoRa, narrowband IoT, and narrowband PLC - lead to insufficient throughput of the operation of a blockchain-managed microgrid. While larger numbers of validators may provide higher decentralisation and fault-tolerant operation, they considerably reduce throughput. The results from the field-test in the Walenstadt microgrid show that the blockchain running on the smart-meter SBCs can provide a maximum throughput of 10 transactions per second. The blockchain throughput halts almost entirely if the system is run by more than 40 validators. Based on the field test, we provide simplified guidelines for utilities or grid operators interested in implementing local P2P markets based on BFT systems.

Access the paper here: https://doi.org/10.1016/j.comcom.2020.04.038

Despite the demonstrated opportunities for revenue enhancement through digitalization, companies often experience a digitalization paradox. This paradox suggests that although companies may invest in digitalization, they often fail to achieve the expected revenue enhancement. In reporting research on 52 companies, we make the following four contributions: First, we focus on industrial companies in the business-to-business context, which largely have been neglected in previous research on digitalization. Second, we introduce the digitalization paradox as an important phenomenon in the discussion of revenue enhancement through digitalization. Third, we describe three growth paths: (1) commercializing digital solutions, (2) utilizing product connectivity, and (3) establishing an IoT-platform-based application business. For each growth path, the article takes a dynamic perspective on business models, highlighting triggers and modifications in business-model components (including value proposition, value-creation activities, and profit equation). Fourth, while the described modifications require initial investments to let these growth paths develop, we highlight how growth traps can prevent investments in business-model modifications from leading to revenue enhancement and how they can ultimately lead to the digitalization paradox.

Access the paper here: https://doi.org/10.1016/j.bushor.2020.01.005

The Internet of Things (loT) promises to deliver enormous business value. More specifically, loT solutions disrupt existing business models by opening up novel service opportunities. In order to help companies understand the opportunities and challenges of this development, we shed light on different loT revenue models. Based on an inductive case study approach, we identify nine direct and indirect revenue patterns. The different types of revenue patterns all use loT-enabled services to create value for customers; the extent and the monetization of services, however, vary.

In recent years, mobile health (mHealth) technologies have received increasing attention from industry and researchers. Such technologies have been the focus of both criticism and high expectations. In this paper, we analyze the integration of mHealth tools in everyday life. Insights into the actual use of such tools have empirical importance and could contribute to our theoretical understanding of mHealth technologies. Our research is based on 23 interviews with the participants of a smartphone-based mobile health intervention aimed at increasing physical activity. We followed the principles of grounded theory during data collection and our analysis is framed by the domestication approach. Our results reveal that the intervention design can result in the participants feeling ill-represented by the reductive nature of the data they generate. The results also reveal the inadequacy between biomedical standards and the social contexts of use. In addition, we describe how middle-class users perceive step-counting through the prism of a moralizing ethos of self-responsibility. Our research has practical implications for the developers and participants of mHealth interventions and theoretical implications regarding mHealth as a societal practice. We also suggest that mHealth-related public policies may fail to reach certain population groups, namely those who do not share the values that surround those technologies and their uses.

Access the paper here: https://doi.org/10.1080/09581596.2020.1725445

Background The Assistant to Lift your Level of activitY (Ally) app is a smartphone application that combines financial incentives with chatbot-guided interventions to encourage users to reach personalized daily step goals. Purpose To evaluate the effects of incentives, weekly planning, and daily self-monitoring prompts that were used as intervention components as part of the Ally app. Methods We conducted an 8 week optimization trial with n = 274 insurees of a health insurance company in Switzerland. At baseline, participants were random-ized to different incentive conditions (cash incentives vs. charity incentives vs. no incentives). Over the course of the study, participants were randomized weekly to different planning conditions (action planning vs. coping planning vs. no planning) and daily to receiving or not receiving a self-monitoring prompt. Primary outcome was the achievement of personalized daily step goals. Results Study participants were more active and healthier than the general Swiss population. Daily cash incentives increased step-goal achievement by 8.1%, 95% confidence interval (CI): [2.1, 14.1] and, only in the no-incentive control group, action planning increased step-goal achievement by 5.8%, 95% CI: [1.2, 10.4]. Charity incentives, self-monitoring prompts, and coping planning did not affect physical activity. Engagement with planning interventions and self-monitoring prompts was low and 30% of participants stopped using the app over the course of the study. Conclusions Daily cash incentives increased physical activity in the short term. Planning interventions and self-monitoring prompts require revision before they can be included in future versions of the app. Selection effects and engagement can be important challenges for physical-activity apps. Clinical Trial Information This study was registered on ClinicalTrials.gov, NCT03384550.

Access paper here: https://doi.org/10.1093/abm/kaaa002

Background: Ongoing pain is one of the most common diseases and has major physical, psychological, social, and economic impacts. A mobile health intervention utilizing a fully automated text-based health care chatbot (TBHC) may offer an innovative way not only to deliver coping strategies and psychoeducation for pain management but also to build a working alliance between a participant and the TBHC.

Objective: The objectives of this study are twofold: (1) to describe the design and implementation to promote the chatbot painSELfMAnagement (SELMA), a 2-month smartphone-based cognitive behavior therapy (CBT) TBHC intervention for pain self-management in patients with ongoing or cyclic pain, and (2) to present findings from a pilot randomized controlled trial, in which effectiveness, influence of intention to change behavior, pain duration, working alliance, acceptance, and adherence were evaluated.

Methods: Participants were recruited online and in collaboration with pain experts, and were randomized to interact with SELMA for 8 weeks either every day or every other day concerning CBT-based pain management (n=59), or weekly concerning content not related to pain management (n=43). Pain-related impairment (primary outcome), general well-being, pain intensity, and the bond scale of working alliance were measured at baseline and postintervention. Intention to change behavior and pain duration were measured at baseline only, and acceptance postintervention was assessed via self-reporting instruments. Adherence was assessed via usage data.

Results: From May 2018 to August 2018, 311 adults downloaded the SELMA app, 102 of whom consented to participate and met the inclusion criteria. The average age of the women (88/102, 86.4%) and men (14/102, 13.6%) participating was 43.7 (SD 12.7) years. Baseline group comparison did not differ with respect to any demographic or clinical variable. The intervention group reported no significant change in pain-related impairment (P=.68) compared to the control group postintervention. The intention to change behavior was positively related to pain-related impairment (P=.01) and pain intensity (P=.01). Working alliance with the TBHC SELMA was comparable to that obtained in guided internet therapies with human coaches. Participants enjoyed using the app, perceiving it as useful and easy to use. Participants of the intervention group replied with an average answer ratio of 0.71 (SD 0.20) to 200 (SD 58.45) conversations initiated by SELMA. Participants’ comments revealed an appreciation of the empathic and responsible interaction with the TBHC SELMA. A main criticism was that there was no option to enter free text for the patients’ own comments.

Conclusions: SELMA is feasible, as revealed mainly by positive feedback and valuable suggestions for future revisions. For example, the participants’ intention to change behavior or a more homogenous sample (eg, with a specific type of chronic pain) should be considered in further tailoring of SELMA.

Trial Registration: German Clinical Trials Register DRKS00017147; https://tinyurl.com/vx6n6sx, Swiss National Clinical Trial Portal: SNCTP000002712; https://www.kofam.ch/de/studienportal/suche/70582/studie/46326.

Access paper here: https://doi.org/10.2196/15806

Background: More than 800 000 people commit suicide every year. This calls for better predictors and prevention. Yet, the temporal dynamics of suicidal ideation make this challenging. A solution might be ecological momentary assessments (EMA). The real-time, real-world data collection might capture the temporal dynamics and help identify predictors. We tested the feasibility of EMA in a high-risk population and examined the predictive value of sleep quality, a promising predictor and modifiable risk factor.

Methods: We included patients across diagnoses with suicidal ideations from the University Hospital of Psychiatry Zurich, Switzerland. Patients used an in-house developed app with 5-times daily EMA for 28 days during the most critical period for suicides: hospital discharge. EMA data was analyzed with a random intercept, random slope model.

Results: We included 15 patients (females: n=12, age: M=31.20±12.71 years) who were clinically depressed (Beck Depression Inventory: M=30.63±14.19) and reported suicidal ideations (107 of 455 EMA in the first week, 23.5%). Of the 455 EMA in the first week, more than half were completed (n=230, 50.5%). We found no evidence that sleep quality predicted suicidal ideation (β=-0.25; 95%CI,-0.59, 0.04; P = 0.12).

Conclusions: We showed that EMA allow to assess suicidal ideations, even in a high-risk population, and to capture its temporal dynamics which is important for an adequate estimation of suicidal ideations to identify reliable predictors and ultimately develop effective treatments. Sleep quality was not a predictor for suicidal ideation, which might be due to the limited sample size and requires further investigation in a larger sample.

Access the paper here: https://doi.org/10.1016/j.biopsych.2020.02.1149

Research indicates that it might be possible to change personality traits through intervention, but this clinical research has primarily focused on changing neuroticism. To date, there are no established, proven techniques for changing other domains of personality, such as conscientiousness and openness. This research examined the effects of a two‐week smartphone‐based intervention to either change one facet of conscientiousness (i.e. self‐discipline) or one facet of openness to experience (i.e. openness to action). Two intervention studies (total N = 255) with two active intervention groups for mutual comparisons were conducted. Results of self‐reports and observer reports showed that people who wanted to become more self‐disciplined were less self‐disciplined at pretest. Similarly, people who wanted to become more open to action were less open to action at pretest. The results showed that people who chose the self‐discipline intervention showed greater increases in self‐discipline, and people who chose the openness to action intervention showed greater increases in openness to action compared with the other group. Changes were maintained until follow‐up two and six weeks after the end of the intervention. Future work is needed to examine whether these personality changes are enduring or reflect temporary accentuation as a result of participation in the intervention.

Access the paper here: http://dx.doi.org/10.1002/per.2267

Desires to increase in extraversion and conscientiousness as well as to decrease in neuroticism are the three most prevalent personality change goals. This study describes characteristics of people who wanted to change one of these personality traits (total N = 1,196) with the help of a digital personality change intervention. The extent to which characteristics predicted the selection of one change goal over the other two was explored using machine learning. Individuals desired to change traits with lower (in case of desires to increase) or higher (in case of desires to decrease) self- and observer-reports and with greater self-other discrepancies. This identification of characteristics of people who desired to change certain personality traits informs future interventions.

Background: Asthma is one of the most prevalent chronic respiratory diseases. Despite increased investment in treatment, little progress has been made in the early recognition and treatment of asthma exacerbations over the last decade. Nocturnal cough monitoring may provide an opportunity to identify patients at risk for imminent exacerbations. Recently developed approaches enable smartphone-based cough monitoring. These approaches, however, have not undergone longitudinal overnight testing nor have they been specifically evaluated in the context of asthma. Also, the problem of distinguishing partner coughs from patient coughs when two or more people are sleeping in the same room using contact-free audio recordings remains unsolved.
Objective: The objective of this study was to evaluate the automatic recognition and segmentation of nocturnal asthmatic coughs and cough epochs in smartphone-based audio recordings that were collected in the field. We also aimed to distinguish partner coughs from patient coughs in contact-free audio recordings by classifying coughs based on sex.
Methods: We used a convolutional neural network model that we had developed in previous work for automated cough recognition. We further used techniques (such as ensemble learning, minibatch balancing, and thresholding) to address the imbalance in the data set. We evaluated the classifier in a classification task and a segmentation task. The cough-recognition classifier served as the basis for the cough-segmentation classifier from continuous audio recordings. We compared automated cough and cough-epoch counts to human-annotated cough and cough-epoch counts. We employed Gaussian mixture models to build a classifier for cough and cough-epoch signals based on sex.
Results: We recorded audio data from 94 adults with asthma (overall: mean 43 years; SD 16 years; female: 54/94, 57%; male 40/94, 43%). Audio data were recorded by each participant in their everyday environment using a smartphone placed next to their bed; recordings were made over a period of 28 nights. Out of 704,697 sounds, we identified 30,304 sounds as coughs. A total of 26,166 coughs occurred without a 2-second pause between coughs, yielding 8238 cough epochs. The ensemble classifier performed well with a Matthews correlation coefficient of 92% in a pure classification task and achieved comparable cough counts to that of human annotators in the segmentation of coughing. The count difference between automated and human-annotated coughs was a mean –0.1 (95% CI –12.11, 11.91) coughs. The count difference between automated and human-annotated cough epochs was a mean 0.24 (95% CI –3.67, 4.15) cough epochs. The Gaussian mixture model cough epoch–based sex classification performed best yielding an accuracy of 83%.
Conclusions: Our study showed longitudinal nocturnal cough and cough-epoch recognition from nightly recorded smartphone-based audio from adults with asthma. The model distinguishes partner cough from patient cough in contact-free recordings by identifying cough and cough-epoch signals that correspond to the sex of the patient. This research represents a step towards enabling passive and scalable cough monitoring for adults with asthma.

Access the paper here: https://doi.org/10.2196/18082

Background: Conversational agents, also known as chatbots, are computer programs designed to simulate human text or verbal conversations. They are increasingly used in a range of fields, including health care. By enabling better accessibility, personalization, and efficiency, conversational agents have the potential to improve patient care.
Objective: This study aimed to review the current applications, gaps, and challenges in the literature on conversational agents in health care and provide recommendations for their future research, design, and application.
Methods: We performed a scoping review. A broad literature search was performed in MEDLINE (Medical Literature Analysis and Retrieval System Online; Ovid), EMBASE (Excerpta Medica database; Ovid), PubMed, Scopus, and Cochrane Central with the search terms “conversational agents,” “conversational AI,” “chatbots,” and associated synonyms. We also searched the gray literature using sources such as the OCLC (Online Computer Library Center) WorldCat database and ResearchGate in April 2019. Reference lists of relevant articles were checked for further articles. Screening and data extraction were performed in parallel by 2 reviewers. The included evidence was analyzed narratively by employing the principles of thematic analysis.
Results: The literature search yielded 47 study reports (45 articles and 2 ongoing clinical trials) that matched the inclusion criteria. The identified conversational agents were largely delivered via smartphone apps (n=23) and used free text only as the main input (n=19) and output (n=30) modality. Case studies describing chatbot development (n=18) were the most prevalent, and only 11 randomized controlled trials were identified. The 3 most commonly reported conversational agent applications in the literature were treatment and monitoring, health care service support, and patient education.
Conclusions: The literature on conversational agents in health care is largely descriptive and aimed at treatment and monitoring and health service support. It mostly reports on text-based, artificial intelligence–driven, and smartphone app–delivered conversational agents. There is an urgent need for a robust evaluation of diverse health care conversational agents’ formats, focusing on their acceptability, safety, and effectiveness.

Access the paper here: https://doi.org/10.2196/17158

Smartphones promise great potential for personality science to study people's everyday life behaviours. Even though personality psychologists have become increasingly interested in the study of personality states, associations between smartphone data and personality states have not yet been investigated. This study provides a first step towards understanding how smartphones may be used for behavioural assessment of personality states. We explored the relationships between Big Five personality states and data from smartphone sensors and usage logs. On the basis of the existing literature, we first compiled a set of behavioural and situational indicators, which are potentially related to personality states. We then applied them on an experience sampling data set containing 5748 personality state responses that are self‐assessments of 30 minutes timeframes and corresponding smartphone data. We used machine learning analyses to investigate the predictability of personality states from the set of indicators. The results showed that only for extraversion, smartphone data (specifically, ambient noise level) were informative beyond what could be predicted based on time and day of the week alone. The results point to continuing challenges in realizing the potential of smartphone data for psychological research.

Access the paper here: https://doi.org/10.1002/per.2309

Just-In-Time Adaptive Intervention (JITAI) is an emerging technique with great potential to support health behavior by providing the right type and amount of support at the right time. A crucial aspect of JITAIs is properly timing the delivery of interventions, to ensure that a user is receptive and ready to process and use the support provided. Some prior works have explored the association of context and some user-specific traits on receptivity, and have built post-study machine-learning models to detect receptivity. For effective intervention delivery, however, a JITAI system needs to make in-the-moment decisions about a user’s receptivity. To this end, we conducted a study in which we deployed machine-learning models to detect receptivity in the natural environment, i.e., in free-living conditions.
We leveraged prior work regarding receptivity to JITAIs and deployed a chatbot-based digital coach – Walkie – that provided physical-activity interventions and motivated participants to achieve their step goals. The Walkie app included two types of machine-learning model that used contextual information about a person to predict when a person is receptive: a static model that was built before the study started and remained constant for all participants and an adaptive model that continuously learned the receptivity of individual participants and updated itself as the study progressed. For comparison, we included a control model that sent intervention messages at random times. The app randomly selected a delivery model for each intervention message. We observed that the machine-learning models led up to a 40% improvement in receptivity as compared to the control model. Further, we evaluated the temporal dynamics of the different models and observed that receptivity to messages from the adaptive model increased over the course of the study.

Access the paper here​: https://arxiv.org/abs/2011.08302

The growing adoption of photovoltaic panels on roof-tops increases the pressure on grid operators for offsetting surplus or deficiency in generation. A multi-carrier energy system allows energy to be converted and stored using different energy carriers, thus relieving the stress from grid operators. However, these systems require efficient operation to unfold their full potential.

This paper proposes a novel blockchain-enabled process to coordinate, allocate, and settle intra-day energy transactions in a district multi-carrier energy system with electricity and heating sub-networks. An incentive mechanism is designed for an optimal allocation of local green energy generation. The mechanism is implemented for the Ethereum blockchain and operates fully on-chain. The design leaves energy producers the freedom to choose their preferred pricing strategy for profit maximization while restricting them to behavior favoring the common good. We test three pricing strategies, with different levels of knowledge on users’ pricing behaviors, that energy producers may adopt. The price-availability-based allocation system guarantees consumers the lowest possible cost.

Access the paper here: https://doi.org/10.1186/s42162-018-0040-4

As a basic building block of the smart grid, advanced metering infrastructure (AMI) is substantial for gathering and sending consumption and production data of consumers. The applications facilitated by blockchain technology like local peer to peer (P2P) markets challenge the centrally organized utility industry with its disruptive potential and rely also heavily on AMIs as data source. However, such technologies pose a number of engineering challenges in early stage pilot projects: Unlike centrally managed AMIs, local P2P markets in particular require AMIs to exchange data with their peer devices, which increases the communication requirements due to the decentral nature of blockchain networks. In this paper, we compare the bandwidth requirement of real-time AMI with the requirements for a blockchain managed peer to peer market. By benchmarking both a normal operation and a high throughput scenario we find a ten times higher demand in bandwidth of the blockchain-based solution compared to real-time AMI and select the appropriate communication technology for an upcoming field test.

Access the paper here: https://www.alexandria.unisg.ch/255396/

With the advent of decentralised energy resources (DERs), there has been increased pressure on classic grid infrastructure to manage non-dispatchable resources. In face of these challenges, microgrids provide a new way of managing and distributing DERs and are characterised as core building blocks of smart grids. In this context blockchain technology enables transaction-based systems for keeping track of energy flows in between producing and consuming parties. However, such systems are not intuitive and introduce challenges for the user»s understanding. In our current work, we introduce a user-centric approach to utilise a transactional data structure, providing transparency and understanding for when and from where electric energy is consumed. We present our approach for an engaging user interface and a preliminary study with feedback from solar installation owners and close with remarks on our future research plans.

Access the paper here: https://dl.acm.org/citation.cfm?doid=3170427.3188610

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.

About 17% of the worldwide CO2-emissions can be ascribed to road transportation. Using information systems (IS)-enabled feedback has shown to be very efficient in promoting a less fuel-consuming driving style. Today, in-car IS that provide feedback on driving behavior are in the midst of a fundamental change. Increasing digitalization of in-car IS enables virtually any kind of feedback. Still, we see a gap in the empirical evidence on how to leverage this potential, raising questions on future HCI-based feedback design. To address this knowledge gap, we designed an eco-driving feedback IS and, building upon construal level theory, hypothesize that abstract feedback is more effective in reducing fuel consumption than concrete feedback. Deployed in a large field experiment with 56 participants covering over 297,000km, we provide first empirical evidence that supports this hypothesis. Despite its limitations, this research may have general implications for the design of real-time feedback.

As a basic building block of the smart grid, advanced metering infrastructure (AMI) is substantial for gathering and sending consumption and production data of consumers. The applications facilitated by blockchain technology like local peer to peer (P2P) markets challenge the centrally organized utility industry with its disruptive potential and rely also heavily on AMIs as data source. However, such technologies pose a number of engineering challenges in early stage pilot projects: Unlike centrally managed AMIs, local P2P markets in particular require AMIs to exchange data with their peer devices, which increases the communication requirements due to the decentral nature of blockchain networks. In this paper, we compare the bandwidth requirement of real-time AMI with the requirements for a blockchain managed peer to peer market. By benchmarking both a normal operation and a high throughput scenario we find a ten times higher demand in bandwidth of the blockchain-based solution compared to real-time AMI and select the appropriate communication technology for an upcoming field test.

Driver identification is a growing topic which offers a streamlined user experience in the connected car, but potentially also highlights privacy issues of our interconnected lives. Recent studies have reported the ability for individuals to be reliably identified out of a group based on their driving behavior. In particular, the state-of-the-art study claims that, in a controlled setting, data collected on how a driver operated the brake pedal could separately identify each of the 15 drivers from the others. The paper at hand was not able to validate these strong scientific claims using naturalistic driving data. In line with the results of other studies using similar data, the replicated identification accuracy dropped to values between 40% and 70% by applying the outlined methods. Nevertheless, this paper further contributes to the field by presenting and evaluating an alternative feature collection technique in order to achieve identification results between 80% and 99.5% in this challenging setting, thus advancing the state-of-the-art. These findings demonstrate the real-world capabilities of data-enabled driver identification, which both facilitates new use-cases and potentially raises privacy questions. As such, important key features from the identification models are presented to assist both researchers and practitioners in this rapidly developing topic.