Standardized Comparison of Voice-Based Information and Documentation Systems to Established Systems in Intensive Care: Crossover Study.

BACKGROUND: The medical teams in intensive care units (ICUs) spend increasing amounts of time at computer systems for data processing, input, and interpretation purposes. As each patient creates about 1000 data points per hour, the available information is abundant, making the interpretation difficult and time-consuming. This data flood leads to a decrease in time for evidence-based, patient-centered care. Information systems, such as patient data management systems (PDMSs), are increasingly used at ICUs. However, they often create new challenges arising from the increasing documentation burden. OBJECTIVE: New concepts, such as artificial intelligence (AI)-based assistant systems, are hence introduced to the workflow to cope with these challenges. However, there is a lack of standardized, published metrics in order to compare the various data input and management systems in the ICU setting. The objective of this study is to compare established documentation and retrieval processes with newer methods, such as PDMSs and voice information and documentation systems (VIDSs). METHODS: In this crossover study, we compare traditional, paper-based documentation systems with PDMSs and newer AI-based VIDSs in terms of performance (required time), accuracy, mental workload, and user experience in an intensive care setting. Performance is assessed on a set of 6 standardized, typical ICU tasks, ranging from documentation to medical interpretation. RESULTS: A total of 60 ICU-experienced medical professionals participated in the study. The VIDS showed a statistically significant advantage compared to the other 2 systems. The tasks were completed significantly faster with the VIDS than with the PDMS (1-tailed t59=12.48; Cohen d=1.61; P<.001) or paper documentation (t59=20.41; Cohen d=2.63; P<.001). Significantly fewer errors were made with VIDS than with the PDMS (t59=3.45; Cohen d=0.45; P=.03) and paper-based documentation (t59=11.2; Cohen d=1.45; P<.001). The analysis of the mental workload of VIDS and PDMS showed no statistically significant difference (P=.06). However, the analysis of subjective user perception showed a statistically significant perceived benefit of the VIDS compared to the PDMS (P<.001) and paper documentation (P<.001). CONCLUSIONS: The results of this study show that the VIDS reduced error rate, documentation time, and mental workload regarding the set of 6 standardized typical ICU tasks. In conclusion, this indicates that AI-based systems such as the VIDS tested in this study have the potential to reduce this workload and improve evidence-based and safe patient care.

Two-stage visual speech recognition for intensive care patients.

In this work, we propose a framework to enhance the communication abilities of speech-impaired patients in an intensive care setting via reading lips. Medical procedure, such as a tracheotomy, causes the patient to lose the ability to utter speech with little to no impact on the habitual lip movement. Consequently, we developed a framework to predict the silently spoken text by performing visual speech recognition, i.e., lip-reading. In a two-stage architecture, frames of the patient's face are used to infer audio features as an intermediate prediction target, which are then used to predict the uttered text. To the best of our knowledge, this is the first approach to bring visual speech recognition into an intensive care setting. For this purpose, we recorded an audio-visual dataset in the University Hospital of Aachen's intensive care unit (ICU) with a language corpus hand-picked by experienced clinicians to be representative of their day-to-day routine. With a word error rate of 6.3%, the trained system reaches a sufficient overall performance to significantly increase the quality of communication between patient and clinician or relatives.

A Visualization and Benchmarking Simulator for Clinical Data.

The availability of Big Data has increased significantly in many areas in recent years. Insights from these data sets lead to optimized processes in many industries, which is why understanding as well as gaining knowledge through analyses of these data sets is becoming increasingly relevant. In the medical field, especially in intensive care units, fast and appropriate treatment is crucial due to the usually critical condition of patients. The patient data recorded here is often very heterogeneous and the resulting database models are very complex, so that accessing and thus using this data requires technical background knowledge. We have focused on the development of a web application that is primarily aimed at clinical staff and researchers. It is an easily accessible visualization and benchmarking tool that provides a graphical interface for the MIMIC-III database. The anonymized datasets contained in MIMIC-III include general information about patients as well as characteristics such as vital signs and laboratory measurements. These datasets are of great interest because they can be used to improve digital decision support systems and clinical processes. Therefore, in addition to visualization, the application can be used by researchers to validate anomaly detection algorithms and by clinical staff to assess disease progression. For this purpose, patient data can be individualized through modifications such as increasing and decreasing vital signs and laboratory parameters so that disease progression can be simulated and subsequently analyzed according to the user's specific needs.

Predicting Abnormalities in Laboratory Values of Patients in the Intensive Care Unit Using Different Deep Learning Models: Comparative Study.

BACKGROUND: In recent years, the volume of medical knowledge and health data has increased rapidly. For example, the increased availability of electronic health records (EHRs) provides accurate, up-to-date, and complete information about patients at the point of care and enables medical staff to have quick access to patient records for more coordinated and efficient care. With this increase in knowledge, the complexity of accurate, evidence-based medicine tends to grow all the time. Health care workers must deal with an increasing amount of data and documentation. Meanwhile, relevant patient data are frequently overshadowed by a layer of less relevant data, causing medical staff to often miss important values or abnormal trends and their importance to the progression of the patient's case. OBJECTIVE: The goal of this work is to analyze the current laboratory results for patients in the intensive care unit (ICU) and classify which of these lab values could be abnormal the next time the test is done. Detecting near-future abnormalities can be useful to support clinicians in their decision-making process in the ICU by drawing their attention to the important values and focus on future lab testing, saving them both time and money. Additionally, it will give doctors more time to spend with patients, rather than skimming through a long list of lab values. METHODS: We used Structured Query Language to extract 25 lab values for mechanically ventilated patients in the ICU from the MIMIC-III and eICU data sets. Additionally, we applied time-windowed sampling and holding, and a support vector machine to fill in the missing values in the sparse time series, as well as the Tukey range to detect and delete anomalies. Then, we used the data to train 4 deep learning models for time series classification, as well as a gradient boosting-based algorithm and compared their performance on both data sets. RESULTS: The models tested in this work (deep neural networks and gradient boosting), combined with the preprocessing pipeline, achieved an accuracy of at least 80% on the multilabel classification task. Moreover, the model based on the multiple convolutional neural network outperformed the other algorithms on both data sets, with the accuracy exceeding 89%. CONCLUSIONS: In this work, we show that using machine learning and deep neural networks to predict near-future abnormalities in lab values can achieve satisfactory results. Our system was trained, validated, and tested on 2 well-known data sets to ensure that our system bridged the reality gap as much as possible. Finally, the model can be used in combination with our preprocessing pipeline on real-life EHRs to improve patients' diagnosis and treatment.

[Artificial Intelligence: Challenges and Applications in Intensive Care Medicine]. / Künstliche Intelligenz: Herausforderungen und Nutzen in der Intensivmedizin.

The high workload in intensive care medicine arises from the exponential growth of medical knowledge, the flood of data generated by the permanent and intensive monitoring of intensive care patients, and the documentation burden. Artificial intelligence (AI) is predicted to have a great impact on ICU work in the near future as it will be applicable in many areas of critical care medicine. These applications include documentation through speech recognition, predictions for decision support, algorithms for parameter optimisation and the development of personalised intensive care medicine. AI-based decision support systems can augment human therapy decisions. Primarily through machine learning, a sub-discipline of AI, self-adaptive algorithms can learn to recognise patterns and make predictions. For actual use in clinical settings, the explainability of such systems is a prerequisite. Intensive care staff spends a large amount of their working hours on documentation, which has increased up to 50% of work time with the introduction of PDMS. Speech recognition has the potential to reduce this documentation burden. It is not yet precise enough to be usable in the clinic. The application of AI in medicine, with the help of large data sets, promises to identify diagnoses more quickly, develop individualised, precise treatments, support therapeutic decisions, use resources with maximum effectiveness and thus optimise the patient experience in the near future.

An Innovative Telemedical Network to Improve Infectious Disease Management in Critically Ill Patients and Outpatients (TELnet@NRW): Stepped-Wedge Cluster Randomized Controlled Trial.

BACKGROUND: Evidence-based infectious disease and intensive care management is more relevant than ever. Medical expertise in the two disciplines is often geographically limited to university institutions. In addition, the interconnection between inpatient and outpatient care is often insufficient (eg, no shared electronic health record and no digital transfer of patient findings). OBJECTIVE: This study aims to establish and evaluate a telemedical inpatient-outpatient network based on expert teleconsultations to increase treatment quality in intensive care medicine and infectious diseases. METHODS: We performed a multicenter, stepped-wedge cluster randomized trial (February 2017 to January 2020) to establish a telemedicine inpatient-outpatient network among university hospitals, hospitals, and outpatient physicians in North Rhine-Westphalia, Germany. Patients aged ≥18 years in the intensive care unit or consulting with a physician in the outpatient setting were eligible. We provided expert knowledge from intensivists and infectious disease specialists through advanced training courses and expert teleconsultations with 24/7/365 availability on demand respectively once per week to enhance treatment quality. The primary outcome was adherence to the 10 Choosing Wisely recommendations for infectious disease management. Guideline adherence was analyzed using binary logistic regression models. RESULTS: Overall, 159,424 patients (10,585 inpatients and 148,839 outpatients) from 17 hospitals and 103 outpatient physicians were included. There was a significant increase in guideline adherence in the management of Staphylococcus aureus infections (odds ratio [OR] 4.00, 95% CI 1.83-9.20; P<.001) and in sepsis management in critically ill patients (OR 6.82, 95% CI 1.27-56.61; P=.04). There was a statistically nonsignificant decrease in sepsis-related mortality from 29% (19/66) in the control group to 23.8% (50/210) in the intervention group. Furthermore, the extension of treatment with prophylactic antibiotics after surgery was significantly less likely (OR 9.37, 95% CI 1.52-111.47; P=.04). Patients treated by outpatient physicians, who were regularly participating in expert teleconsultations, were also more likely to be treated according to guideline recommendations regarding antibiotic therapy for uncomplicated upper respiratory tract infections (OR 1.34, 95% CI 1.16-1.56; P<.001) and asymptomatic bacteriuria (OR 9.31, 95% CI 3.79-25.94; P<.001). For the other recommendations, we found no significant effects, or we had too few observations to generate models. The key limitations of our study include selection effects due to the applied on-site triage of patients as well as the limited possibilities to control for secular effects. CONCLUSIONS: Telemedicine facilitates a direct round-the-clock interaction over broad distances between intensivists or infectious disease experts and physicians who care for patients in hospitals without ready access to these experts. Expert teleconsultations increase guideline adherence and treatment quality in infectious disease and intensive care management, creating added value for critically ill patients. TRIAL REGISTRATION: ClinicalTrials.gov NCT03137589; https://clinicaltrials.gov/ct2/show/NCT03137589.

Development and validation of a reinforcement learning algorithm to dynamically optimize mechanical ventilation in critical care.

The aim of this work was to develop and evaluate the reinforcement learning algorithm VentAI, which is able to suggest a dynamically optimized mechanical ventilation regime for critically-ill patients. We built, validated and tested its performance on 11,943 events of volume-controlled mechanical ventilation derived from 61,532 distinct ICU admissions and tested it on an independent, secondary dataset (200,859 ICU stays; 25,086 mechanical ventilation events). A patient "data fingerprint" of 44 features was extracted as multidimensional time series in 4-hour time steps. We used a Markov decision process, including a reward system and a Q-learning approach, to find the optimized settings for positive end-expiratory pressure (PEEP), fraction of inspired oxygen (FiO2) and ideal body weight-adjusted tidal volume (Vt). The observed outcome was in-hospital or 90-day mortality. VentAI reached a significantly increased estimated performance return of 83.3 (primary dataset) and 84.1 (secondary dataset) compared to physicians' standard clinical care (51.1). The number of recommended action changes per mechanically ventilated patient constantly exceeded those of the clinicians. VentAI chose 202.9% more frequently ventilation regimes with lower Vt (5-7.5 mL/kg), but 50.8% less for regimes with higher Vt (7.5-10 mL/kg). VentAI recommended 29.3% more frequently PEEP levels of 5-7 cm H2O and 53.6% more frequently PEEP levels of 7-9 cmH2O. VentAI avoided high (>55%) FiO2 values (59.8% decrease), while preferring the range of 50-55% (140.3% increase). In conclusion, VentAI provides reproducible high performance by dynamically choosing an optimized, individualized ventilation strategy and thus might be of benefit for critically ill patients.

The Role of Ribonuclease 1 and Ribonuclease Inhibitor 1 in Acute Kidney Injury after Open and Endovascular Thoracoabdominal Aortic Aneurysm Repair.

Acute kidney injury (AKI) is one of the most common post-operative complications and is closely associated with increased mortality after open and endovascular thoracoabdominal aortic aneurysm (TAAA) repair. Ribonuclease (RNase) 1 belongs to the group of antimicrobial peptides elevated in septic patients and indicates the prediction of two or more organ failures. The role of RNase 1 and its antagonist RNase inhibitor 1 (RNH1) after TAAA repair is unknown. In this study, we analyzed RNase 1 and RNH1 serum levels in patients undergoing open (n = 14) or endovascular (n = 19) TAAA repair to determine their association with post-operative AKI and in-hospital mortality. Increased RNH1 serum levels after open TAAA repair as compared with endovascular TAAA repair immediately after surgery and 12, 48, and 72 h after surgery (all p < 0.05) were observed. Additionally, elevated RNase 1 and RNH1 serum levels 12, 24, and 48 h after surgery were shown to be significantly associated with AKI (all p < 0.05). RNH1 serum levels before and RNase 1 serum levels 12 h after TAAA repair were significantly correlated with in-hospital mortality (both p < 0.05). On the basis of these findings, RNase 1 and RNH1 may be therapeutically relevant and may represent biomarkers for post-operative AKI and in-hospital mortality.

Impact of the COVID-19 Pandemic on Urologists in Germany.

BACKGROUND: In order to contain the coronavirus disease 2019 (COVID-19) pandemic, Germany has implemented drastic restrictions on public or social life, while health institutions are invoked to postpone elective procedures. Although urologists are less involved in the direct treatment of COVID-19 patients, the current situation strongly affects the urological work routine. OBJECTIVE: To analyze the impact of the COVID-19 pandemic on various aspects of work and personal life among urologists in Germany. DESIGN, SETTING, AND PARTICIPANTS: A total of 589 urologists in Germany participated in an online survey between March 27 and April 11, 2020. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Participants were stratified into subgroups according to professional characteristics. RESULTS AND LIMITATIONS: Most urologists rated Germany as "well prepared" and the increasing restrictions of social life as "very positive." Routine operation was more restricted in hospitals than in the outpatient sector (p = 0.046). Moreover, urologists from the outpatient sector felt significantly less prepared for the COVID-19 pandemic (p = 0.001), reported a higher shortage of protective medical equipment (p < 0.001), and described a tendency toward a higher level of threat (p = 0.054). Although restrictions regarding telemedicine approaches were reported by 60% of participants, the outpatient sector used telehealth more frequently than hospitals (25.5% vs 17.0%, p < 0.001). Limitations include the national design and the restricted survey period. CONCLUSIONS: This survey systematically evaluates the impact of the COVID-19 pandemic on personal and professional aspects of German urologists. We identified several issues, such as a higher shortage of medical protective equipment in the outpatient sector that could trigger specific measures to further improve the quality of urological care in Germany. PATIENT SUMMARY: We evaluated a potential impact of the coronavirus disease 2019 (COVID-19) pandemic on professional and personal aspects of the urologists in Germany. Our results suggest that the outpatient sector should receive specific attention as, for example, shortage of protective equipment was more common.

Perception of the 2020 SARS-CoV-2 pandemic among medical professionals in Germany: results from a nationwide online survey.

Background: The COVID-19 pandemic represents an unprecedented global challenge and implicates a wide range of burden on medical professionals. Here, we evaluated the perception of the COVID-19 pandemic among medical professionals in Germany. Methods: A total of n = 2827 medical professionals participated in an online survey between 27 March and 11 April. Results: While most participants stated that Germany was well prepared and rated the measures taken by their employer as positive, subgroup analyses revealed decisive differences. The preventive measures were rated significantly worse by nurses compared to doctors (p < 0.001) and by participants from ambulatory healthcare centres compared to participants from maximum-care hospitals (p < 0.001). Importantly, shortage of protective medical equipment was reported more commonly in the ambulatory sector (p < 0.001) and in East German federal states (p = 0.004). Moreover, the majority of health care professionals (72.4%) reported significant restrictions of daily work routine. Finally, over 60% of medical professionals had concerns regarding their own health, which were more pronounced among female participants (p = 0.024). Conclusion: This survey may indicate starting points on how medical professionals could be supported in carrying out their important activities during the ongoing and future healthcare challenges.

Telemedicine in Germany During the COVID-19 Pandemic: Multi-Professional National Survey.

BACKGROUND: In an effort to contain the effects of the coronavirus disease (COVID-19) pandemic, health care systems worldwide implemented telemedical solutions to overcome staffing, technical, and infrastructural limitations. In Germany, a multitude of telemedical systems are already being used, while new approaches are rapidly being developed in response to the crisis. However, the extent of the current implementation within different health care settings, the user's acceptance and perception, as well as the hindering technical and regulatory obstacles remain unclear. OBJECTIVE: The aim of this paper is to assess the current status quo of the availability and routine use of telemedical solutions, user acceptance, and the subjectively perceived burdens on telemedical approaches. Furthermore, we seek to assess the perception of public information quality among professional groups and their preferred communication channels. METHODS: A national online survey was conducted on 14 consecutive days in March and April 2020, and distributed to doctors, nurses, and other medical professionals in the German language. RESULTS: A total of 2827 medical professionals participated in the study. Doctors accounted for 65.6% (n=1855) of the professionals, 29.5% (n=833) were nursing staff, and 4.9% (n=139) were identified as others such as therapeutic staff. A majority of participants rated the significance of telemedicine within the crisis as high (1065/2730, 39%) or neutral (n=720, 26.4%); however, there were significant differences between doctors and nurses (P=.01) as well as between the stationary sector compared to the ambulatory sector (P<.001). Telemedicine was already in routine use for 19.6% (532/2711) of German health care providers and in partial use for 40.2% (n=1090). Participants working in private practices (239/594, 40.2%) or private clinics (23/59, 39.0%) experienced less regulatory or technical obstacles compared to university hospitals (586/1190, 49.2%). A majority of doctors rated the public information quality on COVID-19 as good (942/1855, 50.8%) or very good (213/1855, 11.5%); nurses rated the quality of public information significantly lower (P<.001). Participant's age negatively correlated with the perception of telemedicine's significance (ρ=-0.23; P<.001). CONCLUSIONS: Telemedicine has a broad acceptance among German medical professionals. However, to establish telemedical structures within routine care, technical and regulatory burdens must be overcome.

A Deep Learning Approach for Managing Medical Consumable Materials in Intensive Care Units via Convolutional Neural Networks: Technical Proof-of-Concept Study.

BACKGROUND: High numbers of consumable medical materials (eg, sterile needles and swabs) are used during the daily routine of intensive care units (ICUs) worldwide. Although medical consumables largely contribute to total ICU hospital expenditure, many hospitals do not track the individual use of materials. Current tracking solutions meeting the specific requirements of the medical environment, like barcodes or radio frequency identification, require specialized material preparation and high infrastructure investment. This impedes the accurate prediction of consumption, leads to high storage maintenance costs caused by large inventories, and hinders scientific work due to inaccurate documentation. Thus, new cost-effective and contactless methods for object detection are urgently needed. OBJECTIVE: The goal of this work was to develop and evaluate a contactless visual recognition system for tracking medical consumable materials in ICUs using a deep learning approach on a distributed client-server architecture. METHODS: We developed Consumabot, a novel client-server optical recognition system for medical consumables, based on the convolutional neural network model MobileNet implemented in Tensorflow. The software was designed to run on single-board computer platforms as a detection unit. The system was trained to recognize 20 different materials in the ICU, while 100 sample images of each consumable material were provided. We assessed the top-1 recognition rates in the context of different real-world ICU settings: materials presented to the system without visual obstruction, 50% covered materials, and scenarios of multiple items. We further performed an analysis of variance with repeated measures to quantify the effect of adverse real-world circumstances. RESULTS: Consumabot reached a >99% reliability of recognition after about 60 steps of training and 150 steps of validation. A desirable low cross entropy of <0.03 was reached for the training set after about 100 iteration steps and after 170 steps for the validation set. The system showed a high top-1 mean recognition accuracy in a real-world scenario of 0.85 (SD 0.11) for objects presented to the system without visual obstruction. Recognition accuracy was lower, but still acceptable, in scenarios where the objects were 50% covered (P<.001; mean recognition accuracy 0.71; SD 0.13) or multiple objects of the target group were present (P=.01; mean recognition accuracy 0.78; SD 0.11), compared to a nonobstructed view. The approach met the criteria of absence of explicit labeling (eg, barcodes, radio frequency labeling) while maintaining a high standard for quality and hygiene with minimal consumption of resources (eg, cost, time, training, and computational power). CONCLUSIONS: Using a convolutional neural network architecture, Consumabot consistently achieved good results in the classification of consumables and thus is a feasible way to recognize and register medical consumables directly to a hospital's electronic health record. The system shows limitations when the materials are partially covered, therefore identifying characteristics of the consumables are not presented to the system. Further development of the assessment in different medical circumstances is needed.

Prediction of Prolonged ICU Stay in Cardiac Surgery Patients as a Useful Method to Identify Nutrition Risk in Cardiac Surgery Patients: A Post Hoc Analysis of a Prospective Observational Study.

BACKGROUND: Cardiovascular surgery patients with a prolonged intensive care unit (ICU) stay may benefit most from early nutrition support. Using established scoring systems for nutrition assessment and operative risk stratification, we aimed to develop a model to predict a prolonged ICU stay ≥5 days in order to identify patients who will benefit from early nutrition interventions. METHODS: This is a retrospective analysis of a prospective observational study of patients undergoing elective valvular, coronary artery bypass grafting, or combined cardiac surgery. The nutrition risk was assessed by well-established screening tools. Patients' preoperative EuroSCORE (European System for Cardiac Operative Risk Evaluation), primary disease, and intraoperative cardiopulmonary bypass (CPB) time were included as independent variables in a multivariate logistic regression analysis to predict a prolonged ICU stay (>4 days). RESULTS: The number of cardiac surgery patients included was 1193. Multivariate analysis revealed that for prediction of ICU stay >4 days, both Nutritional Risk Screening 2002 (area under the curve (AUC): 0.716, P = .020) and Mini Nutritional Assessment (MNA) score (AUC: 0.715, P = .037) were significant, whereas for prediction of ICU stay >5 days, only the MNA score showed significant results (AUC: 0.762, P = .011). CONCLUSION: Present data provide first evidence about the combined use of EuroSCORE, primary disease, CPB time, and nutrition risk screening tools for prediction of prolonged ICU stay in cardiac surgery patients. If prospectively evaluated in adequately designed studies, this model may help to identify patients with prolonged ICU stay to initiate early postoperative nutrition therapy and thus, facilitate an enhanced recovery.

Self-directed learning can outperform direct instruction in the course of a modern German medical curriculum - results of a mixed methods trial.

BACKGROUND: Modernised medical curricula in Germany (so called "reformed study programs") rely increasingly on alternative self-instructed learning forms such as e-learning and curriculum-guided self-study. However, there is a lack of evidence that these methods can outperform conventional teaching methods such as lectures and seminars. This study was conducted in order to compare extant traditional teaching methods with new instruction forms in terms of learning effect and student satisfaction. METHODS: In a randomised trial, 244 students of medicine in their third academic year were assigned to one of four study branches representing self-instructed learning forms (e-learning and curriculum-based self-study) and instructed learning forms (lectures and seminars). All groups participated in their respective learning module with standardised materials and instructions. Learning effect was measured with pre-test and post-test multiple-choice questionnaires. Student satisfaction and learning style were examined via self-assessment. RESULTS: Of 244 initial participants, 223 completed the respective module and were included in the study. In the pre-test, the groups showed relatively homogenous scores. All students showed notable improvements compared with the pre-test results. Participants in the non-self-instructed learning groups reached scores of 14.71 (seminar) and 14.37 (lecture), while the groups of self-instructed learners reached higher scores with 17.23 (e-learning) and 15.81 (self-study). All groups improved significantly (p < .001) in the post-test regarding their self-assessment, led by the e-learning group, whose self-assessment improved by 2.36. CONCLUSIONS: The study shows that students in modern study curricula learn better through modern self-instructed methods than through conventional methods. These methods should be used more, as they also show good levels of student acceptance and higher scores in personal self-assessment of knowledge.