The rapidly evolving science about the Coronavirus Disease 2019 (COVID-19) pandemic created unprecedented health information needs and dramatic changes in policies globally. We describe a platform, ...Watson Assistant (WA), which has been used to develop conversational agents to deliver COVID-19 related information. We characterized the diverse use cases and implementations during the early pandemic and measured adoption through a number of users, messages sent, and conversational turns (ie, pairs of interactions between users and agents). Thirty-seven institutions in 9 countries deployed COVID-19 conversational agents with WA between March 30 and August 10, 2020, including 24 governmental agencies, 7 employers, 5 provider organizations, and 1 health plan. Over 6.8 million messages were delivered through the platform. The mean number of conversational turns per session ranged between 1.9 and 3.5. Our experience demonstrates that conversational technologies can be rapidly deployed for pandemic response and are adopted globally by a wide range of users.
The development and implementation of clinical decision support (CDS) that trains itself and adapts its algorithms based on new data-here referred to as Adaptive CDS-present unique challenges and ...considerations. Although Adaptive CDS represents an expected progression from earlier work, the activities needed to appropriately manage and support the establishment and evolution of Adaptive CDS require new, coordinated initiatives and oversight that do not currently exist. In this AMIA position paper, the authors describe current and emerging challenges to the safe use of Adaptive CDS and lay out recommendations for the effective management and monitoring of Adaptive CDS.
•We applied automated classification methods to patient generated portal messages.•Classifier features included words from messages, NLP concepts, and NLP semantics.•Random forest and logistic ...regression approaches accurately classified messages.•Automated classification may aid in managing growing volumes of portal messages.
Secure messaging through patient portals is an increasingly popular way that consumers interact with healthcare providers. The increasing burden of secure messaging can affect clinic staffing and workflows. Manual management of portal messages is costly and time consuming. Automated classification of portal messages could potentially expedite message triage and delivery of care.
We developed automated patient portal message classifiers with rule-based and machine learning techniques using bag of words and natural language processing (NLP) approaches. To evaluate classifier performance, we used a gold standard of 3253 portal messages manually categorized using a taxonomy of communication types (i.e., main categories of informational, medical, logistical, social, and other communications, and subcategories including prescriptions, appointments, problems, tests, follow-up, contact information, and acknowledgement). We evaluated our classifiers’ accuracies in identifying individual communication types within portal messages with area under the receiver-operator curve (AUC). Portal messages often contain more than one type of communication. To predict all communication types within single messages, we used the Jaccard Index. We extracted the variables of importance for the random forest classifiers.
The best performing approaches to classification for the major communication types were: logistic regression for medical communications (AUC: 0.899); basic (rule-based) for informational communications (AUC: 0.842); and random forests for social communications and logistical communications (AUCs: 0.875 and 0.925, respectively). The best performing classification approach of classifiers for individual communication subtypes was random forests for Logistical-Contact Information (AUC: 0.963). The Jaccard Indices by approach were: basic classifier, Jaccard Index: 0.674; Naïve Bayes, Jaccard Index: 0.799; random forests, Jaccard Index: 0.859; and logistic regression, Jaccard Index: 0.861. For medical communications, the most predictive variables were NLP concepts (e.g., Temporal_Concept, which maps to ‘morning’, ‘evening’ and Idea_or_Concept which maps to ‘appointment’ and ‘refill’). For logistical communications, the most predictive variables contained similar numbers of NLP variables and words (e.g., Telephone mapping to ‘phone’, ‘insurance’). For social and informational communications, the most predictive variables were words (e.g., social: ‘thanks’, ‘much’, informational: ‘question’, ‘mean’).
This study applies automated classification methods to the content of patient portal messages and evaluates the application of NLP techniques on consumer communications in patient portal messages. We demonstrated that random forest and logistic regression approaches accurately classified the content of portal messages, although the best approach to classification varied by communication type. Words were the most predictive variables for classification of most communication types, although NLP variables were most predictive for medical communication types. As adoption of patient portals increases, automated techniques could assist in understanding and managing growing volumes of messages. Further work is needed to improve classification performance to potentially support message triage and answering.
Artificial intelligence (AI) represents a valuable tool that could be used to improve the safety of care. Major adverse events in healthcare include: healthcare-associated infections, adverse drug ...events, venous thromboembolism, surgical complications, pressure ulcers, falls, decompensation, and diagnostic errors. The objective of this scoping review was to summarize the relevant literature and evaluate the potential of AI to improve patient safety in these eight harm domains. A structured search was used to query MEDLINE for relevant articles. The scoping review identified studies that described the application of AI for prediction, prevention, or early detection of adverse events in each of the harm domains. The AI literature was narratively synthesized for each domain, and findings were considered in the context of incidence, cost, and preventability to make projections about the likelihood of AI improving safety. Three-hundred and ninety-two studies were included in the scoping review. The literature provided numerous examples of how AI has been applied within each of the eight harm domains using various techniques. The most common novel data were collected using different types of sensing technologies: vital sign monitoring, wearables, pressure sensors, and computer vision. There are significant opportunities to leverage AI and novel data sources to reduce the frequency of harm across all domains. We expect AI to have the greatest impact in areas where current strategies are not effective, and integration and complex analysis of novel, unstructured data are necessary to make accurate predictions; this applies specifically to adverse drug events, decompensation, and diagnostic errors.
Adverse drug events (ADEs) represent one of the most prevalent types of health-care-related harm, and there is substantial room for improvement in the way that they are currently predicted and ...detected. We conducted a scoping review to identify key use cases in which artificial intelligence (AI) could be leveraged to reduce the frequency of ADEs. We focused on modern machine learning techniques and natural language processing. 78 articles were included in the scoping review. Studies were heterogeneous and applied various AI techniques covering a wide range of medications and ADEs. We identified several key use cases in which AI could contribute to reducing the frequency and consequences of ADEs, through prediction to prevent ADEs and early detection to mitigate the effects. Most studies (73 94% of 78) assessed technical algorithm performance, and few studies evaluated the use of AI in clinical settings. Most articles (58 74% of 78) were published within the past 5 years, highlighting an emerging area of study. Availability of new types of data, such as genetic information, and access to unstructured clinical notes might further advance the field.
Vascular surgery is a subspecialty that attracts future surgeons with challenging technical procedures and complex decision making. Despite its appeal, continued promotion of the field is necessary ...to recruit and retain the best and brightest candidates. Recruitment of medical students and residents may be limited by the lifestyle inherent to vascular surgery and the length of residency training. The young adults of the current applicant and resident pool differ from prior generations in their desire for hands-on mentoring, aspirations to affect change daily, a penchant for technology, and strong emphasis on work-life balance. Furthermore, the percentage of women pursuing careers in vascular surgery is not representative of the eligible workforce. Women are now the majority of graduates in all of higher education, and thus, vascular surgery may need to make a concerted effort to appeal to women in order to attract the most talented young professionals to the field. Recruiting strategies for both men and women of Generation Y should target a diverse group of potential candidates with an awareness of the unique characteristics and needs of this generation of rising surgeons.
Recurrent vaso-occlusive pain episodes, the most common complication of sickle cell disease (SCD), cause frequent health care utilization. Studies exploring associations between patient activation ...and acute health care utilization for pain are lacking. We tested the hypothesis that increased activation and self-efficacy are associated with decreased health care utilization for pain in SCD.
In this cross-sectional study of adults with SCD at a tertiary medical center, we collected demographics, SCD phenotype, Patient Activation Measure levels, and self-efficacy scores using structured questionnaires. We reviewed charts to obtain disease-modifying therapy and acute health care utilization, defined as emergency room visits and hospitalizations, for vaso-occlusive pain episodes. Negative binomial regression analyses were used to test the hypothesis.
We surveyed 67 adults with SCD. The median age was 27.0 years, 53.7% were female, and 95.5% were African American. Median health care utilization for pain over one year (range) was 2.0 (0-24). Only one-third of participants (38.8%) were at the highest activation level (median range = 3 1-4). Two-thirds (65.7%) of participants had high self-efficacy (median range = 32.0 13-45). Regressions showed significant association between health care utilization and activation (incidence rate ratio IRR = 0.663, P = 0.045), self-efficacy (IRR = 0.947, P = 0.038), and male sex (IRR = 0.390, P = 0.003). Two outliers with high activation, self-efficacy, and health care utilization also had addictive behavior.
Many individuals with SCD have suboptimal activation and reduced self-efficacy. Higher activation and self-efficacy were associated with lower health care utilization for pain. Additional studies are needed to evaluate interventions to improve activation and self-efficacy and reduce acute health care utilization for pain.
Artificial intelligence (AI) represents a valuable tool that could be widely used to inform clinical and public health decision-making to effectively manage the impacts of a pandemic. The objective ...of this scoping review was to identify the key use cases for involving AI for pandemic preparedness and response from the peer-reviewed, preprint, and grey literature. The data synthesis had two parts: an in-depth review of studies that leveraged machine learning (ML) techniques and a limited review of studies that applied traditional modeling approaches. ML applications from the in-depth review were categorized into use cases related to public health and clinical practice, and narratively synthesized. One hundred eighty-three articles met the inclusion criteria for the in-depth review. Six key use cases were identified: forecasting infectious disease dynamics and effects of interventions; surveillance and outbreak detection; real-time monitoring of adherence to public health recommendations; real-time detection of influenza-like illness; triage and timely diagnosis of infections; and prognosis of illness and response to treatment. Data sources and types of ML that were useful varied by use case. The search identified 1167 articles that reported on traditional modeling approaches, which highlighted additional areas where ML could be leveraged for improving the accuracy of estimations or projections. Important ML-based solutions have been developed in response to pandemics, and particularly for COVID-19 but few were optimized for practical application early in the pandemic. These findings can support policymakers, clinicians, and other stakeholders in prioritizing research and development to support operationalization of AI for future pandemics.
Cybersecurity is an increasingly important concern for reliable healthcare delivery and is particularly salient for robotic surgery. Surgical robots are complex systems with numerous points of ...vulnerability, and there have been real-world demonstrations of successful cyberattacks on surgical robots. There are several ways to improve the risk profile of robotic surgery, including recognizing system complexity, investing in regular software updates, following cybersecurity best-practices, and increasing transparency for all stakeholders. As robotic surgery continues to technologically advance, ensuring overall system safety from a cybersecurity perspective is paramount.
Background
Contact tracing in association with quarantine and isolation is an important public health tool to control outbreaks of infectious diseases. This strategy has been widely implemented ...during the current COVID-19 pandemic. The effectiveness of this nonpharmaceutical intervention is largely dependent on social interactions within the population and its combination with other interventions. Given the high transmissibility of SARS-CoV-2, short serial intervals, and asymptomatic transmission patterns, the effectiveness of contact tracing for this novel viral agent is largely unknown.
Objective
This study aims to identify and synthesize evidence regarding the effectiveness of contact tracing on infectious viral disease outcomes based on prior scientific literature.
Methods
An evidence-based review was conducted to identify studies from the PubMed database, including preprint medRxiv server content, related to the effectiveness of contact tracing in viral outbreaks. The search dates were from database inception to July 24, 2020. Outcomes of interest included measures of incidence, transmission, hospitalization, and mortality.
Results
Out of 159 unique records retrieved, 45 (28.3%) records were reviewed at the full-text level, and 24 (15.1%) records met all inclusion criteria. The studies included utilized mathematical modeling (n=14), observational (n=8), and systematic review (n=2) approaches. Only 2 studies considered digital contact tracing. Contact tracing was mostly evaluated in combination with other nonpharmaceutical interventions and/or pharmaceutical interventions. Although some degree of effectiveness in decreasing viral disease incidence, transmission, and resulting hospitalizations and mortality was observed, these results were highly dependent on epidemic severity (R0 value), number of contacts traced (including presymptomatic and asymptomatic cases), timeliness, duration, and compliance with combined interventions (eg, isolation, quarantine, and treatment). Contact tracing effectiveness was particularly limited by logistical challenges associated with increased outbreak size and speed of infection spread.
Conclusions
Timely deployment of contact tracing strategically layered with other nonpharmaceutical interventions could be an effective public health tool for mitigating and suppressing infectious outbreaks by decreasing viral disease incidence, transmission, and resulting hospitalizations and mortality.
PDF
