Solving the explainable AI conundrum by bridging clinicians' needs and developers' goals.

Explainable artificial intelligence (XAI) has emerged as a promising solution for addressing the implementation challenges of AI/ML in healthcare. However, little is known about how developers and clinicians interpret XAI and what conflicting goals and requirements they may have. This paper presents the findings of a longitudinal multi-method study involving 112 developers and clinicians co-designing an XAI solution for a clinical decision support system. Our study identifies three key differences between developer and clinician mental models of XAI, including opposing goals (model interpretability vs. clinical plausibility), different sources of truth (data vs. patient), and the role of exploring new vs. exploiting old knowledge. Based on our findings, we propose design solutions that can help address the XAI conundrum in healthcare, including the use of causal inference models, personalized explanations, and ambidexterity between exploration and exploitation mindsets. Our study highlights the importance of considering the perspectives of both developers and clinicians in the design of XAI systems and provides practical recommendations for improving the effectiveness and usability of XAI in healthcare.

ICU Cockpit: a platform for collecting multimodal waveform data, AI-based computational disease modeling and real-time decision support in the intensive care unit.

ICU Cockpit: a secure, fast, and scalable platform for collecting multimodal waveform data, online and historical data visualization, and online validation of algorithms in the intensive care unit. We present a network of software services that continuously stream waveforms from ICU beds to databases and a web-based user interface. Machine learning algorithms process the data streams and send outputs to the user interface. The architecture and capabilities of the platform are described. Since 2016, the platform has processed over 89 billion data points (N = 979 patients) from 200 signals (0.5-500 Hz) and laboratory analyses (once a day). We present an infrastructure-based framework for deploying and validating algorithms for critical care. The ICU Cockpit is a Big Data platform for critical care medicine, especially for multimodal waveform data. Uniquely, it allows algorithms to seamlessly integrate into the live data stream to produce clinical decision support and predictions in clinical practice.

RWD-Cockpit: Application for Quality Assessment of Real-world Data.

BACKGROUND: Digital technologies are transforming the health care system. A large part of information is generated as real-world data (RWD). Data from electronic health records and digital biomarkers have the potential to reveal associations between the benefits and adverse events of medicines, establish new patient-stratification principles, expose unknown disease correlations, and inform on preventive measures. The impact for health care payers and providers, the biopharmaceutical industry, and governments is massive in terms of health outcomes, quality of care, and cost. However, a framework to assess the preliminary quality of RWD is missing, thus hindering the conduct of population-based observational studies to support regulatory decision-making and real-world evidence. OBJECTIVE: To address the need to qualify RWD, we aimed to build a web application as a tool to translate characterization of some quality parameters of RWD into a metric and propose a standard framework for evaluating the quality of the RWD. METHODS: The RWD-Cockpit systematically scores data sets based on proposed quality metrics and customizable variables chosen by the user. Sleep RWD generated de novo and publicly available data sets were used to validate the usability and applicability of the web application. The RWD quality score is based on the evaluation of 7 variables: manageability specifies access and publication status; complexity defines univariate, multivariate, and longitudinal data; sample size indicates the size of the sample or samples; privacy and liability stipulates privacy rules; accessibility specifies how the data set can be accessed and to what granularity; periodicity specifies how often the data set is updated; and standardization specifies whether the data set adheres to any specific technical or metadata standard. These variables are associated with several descriptors that define specific characteristics of the data set. RESULTS: To address the need to qualify RWD, we built the RWD-Cockpit web application, which proposes a framework and applies a common standard for a preliminary evaluation of RWD quality across data sets-molecular, phenotypical, and social-and proposes a standard that can be further personalized by the community retaining an internal standard. Applied to 2 different case studies-de novo-generated sleep data and publicly available data sets-the RWD-Cockpit could identify and provide researchers with variables that might increase quality. CONCLUSIONS: The results from the application of the framework of RWD metrics implemented in the RWD-Cockpit application suggests that multiple data sets can be preliminarily evaluated in terms of quality using the proposed metrics. The output scores-quality identifiers-provide a first quality assessment for the use of RWD. Although extensive challenges remain to be addressed to set RWD quality standards, our proposal can serve as an initial blueprint for community efforts in the characterization of RWD quality for regulated settings.

The Energy Consumption of Radiology: Energy- and Cost-saving Opportunities for CT and MRI Operation.

Background Awareness of energy efficiency has been rising in the industrial and residential sectors but only recently in the health care sector. Purpose To measure the energy consumption of modern CT and MRI scanners in a university hospital radiology department and to estimate energy- and cost-saving potential during clinical operation. Materials and Methods Three CT scanners, four MRI scanners, and cooling systems were equipped with kilowatt-hour energy measurement sensors (2-Hz sampling rate). Energy measurements, the scanners' log files, and the radiology information system from the entire year 2015 were analyzed and segmented into scan modes, as follows: net scan (actual imaging), active (room time), idle, and system-on and system-off states (no standby mode was available). Per-examination and peak energy consumption were calculated. Results The aggregated energy consumption imaging 40 276 patients amounted to 614 825 kWh, dedicated cooling systems to 492 624 kWh, representing 44.5% of the combined consumption of 1 107 450 kWh (at a cost of U.S. $199 341). This is equivalent to the usage in a town of 852 people and constituted 4.0% of the total yearly energy consumption at the authors' hospital. Mean consumption per CT examination over 1 year was 1.2 kWh, with a mean energy cost (±standard deviation) of $0.22 ± 0.13. The total energy consumption of one CT scanner for 1 year was 26 226 kWh ($4721 in energy cost). The net consumption per CT examination over 1 year was 3580 kWh, which is comparable to the usage of a two-person household in Switzerland; however, idle state consumption was fourfold that of net consumption (14 289 kWh). Mean MRI consumption over 1 year was 19.9 kWh per examination, with a mean energy cost of $3.57 ± 0.96. The mean consumption for a year in the system-on state was 82 174 kWh per MRI examination and 134 037 kWh for total consumption, for an energy cost of $24 127. Conclusion CT and MRI energy consumption is substantial. Considerable energy- and cost-saving potential is present during nonproductive idle and system-off modes, and this realization could decrease total cost of ownership while increasing energy efficiency. © RSNA, 2020.

[Patient tracking with beacon technology : Pilot project in a radiological practice]. / Patiententracking mit Beacon-Technologie : Pilotprojekt in einer radiologischen Praxis.

Waiting times are still assumed to be unavoidable in medicine. However, waiting time is an essential factor of patient satisfaction. Because patient expectations are increasing, medical institutions should address the issue. Above all, this requires transparency about the current processes in the facilities. Conventional information systems often do not offer sufficient solutions to ensure this in real time combined with helpful visualization. In a pilot project in a radiological practice, the use of a patient tracking system based on beacon technology was tested. The aim was to track the actual location of the patients in the practice, to determine the patient status (e.g. patient waiting) and to display the entire processes on a smart dashboard. The successful pilot project has shown that the technology meets all requirements, that patients accept the system and that staff are familiar with the new processes after some time. For the first time, patient flows, including waiting times, were displayed clearly and in real time on a dashboard. This made it possible to control processes and waiting times that had previously never been recorded in a structured manner and were usually only recognized in the event of complaints. From a technical point of view, the system is arbitrarily scalable, whereby the connection to different information systems will be a challenge. If this succeeds, however, the possibilities are manifold. The created transparency makes it possible to reduce waiting times and to actively inform patients about waiting times and thus contribute to increasing patient satisfaction.

Making the Radiology Workflow Visible in Order to Inform Optimization Strategies.

Medical imaging is undergoing rapid change, induced by the increasing amount of image data, and advances in fields such as artificial intelligence. In order for a radiology service provider to respond to these challenges, it needs to adapt its workflow. To inform optimization strategies, the way that processes and resources interact in the real world must be understood. We report on our experiences with an approach that consists of merging a variety of data sources into a data model that allows efficient interactive queries, and then providing highly interactive visualizations to explore the data. Two examples are discussed: animation of patient flow through the radiology workflow, and the use of energy consumption patterns to characterize operational modalities.

Visualization beyond the desktop--the next big thing.

Visualization is coming of age. With visual depictions being seamlessly integrated into documents, and data visualization techniques being used to understand increasingly large and complex datasets, the term "visualization"' is becoming used in everyday conversations. But we are on a cusp; visualization researchers need to develop and adapt to today's new devices and tomorrow's technology. Today, people interact with visual depictions through a mouse. Tomorrow, they'll be touching, swiping, grasping, feeling, hearing, smelling, and even tasting data. The next big thing is multisensory visualization that goes beyond the desktop.