Few-Shot Learning for Identification of COVID-19 Symptoms Using Generative Pre-trained Transformer Language Models

Since the onset of the COVID-19 pandemic, social media users have shared their personal experiences related to the viral infection. Their posts contain rich information of symptoms that may provide useful hints to advancing the knowledge body of medical research and supplement the discoveries from clinical settings. Identification of symptom expressions in social media text is challenging, partially due to lack of annotated data. In this study, we investigate utilizing few-shot learning with generative pre-trained transformer language models to identify COVID-19 symptoms in Twitter posts. The results of our approach show that large language models are promising in more accurately identifying symptom expressions in Twitter posts with small amount of annotation effort, and our method can be applied to other medical and health applications where abundant of unlabeled data is available. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Tracking News Stories in Short Messages in the Era of Infodemic

Tracking news stories in documents is a way to deal with the large amount of information that surrounds us everyday, to reduce the noise and to detect emergent topics in news. Since the Covid-19 outbreak, the world has known a new problem: infodemic. News article titles are massively shared on social networks and the analysis of trends and growing topics is complex. Grouping documents in news stories lowers the number of topics to analyse and the information to ingest and/or evaluate. Our study proposes to analyse news tracking with little information provided by titles on social networks. In this paper, we take advantage of datasets of public news article titles to experiment news tracking algorithms on short messages. We evaluate the clustering performance with little amount of data per document. We deal with the document representation (sparse with TF-IDF and dense using Transformers [26]), its impact on the results and why it is key to this type of work. We used a supervised algorithm proposed by Miranda et al. [22] and K-Means to provide evaluations for different use cases. We found that TF-IDF vectors are not always the best ones to group documents, and that algorithms are sensitive to the type of representation. Knowing this, we recommend taking both aspects into account while tracking news stories in short messages. With this paper, we share all the source code and resources we handled. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

A Web-Based, Provider-Driven Mobile App to Enhance Patient Care Coordination Between Dialysis Facilities and Hospitals: Development and Pilot Implementation Study.

BACKGROUND: We piloted a web-based, provider-driven mobile app (DialysisConnect) to fill the communication and care coordination gap between hospitals and dialysis facilities. OBJECTIVE: This study aimed to describe the development and pilot implementation of DialysisConnect. METHODS: DialysisConnect was developed iteratively with focus group and user testing feedback and was made available to 120 potential users at 1 hospital (hospitalists, advanced practice providers [APPs], and care coordinators) and 4 affiliated dialysis facilities (nephrologists, APPs, nurses and nurse managers, social workers, and administrative personnel) before the start of the pilot (November 1, 2020, to May 31, 2021). Midpilot and end-of-pilot web-based surveys of potential users were also conducted. Descriptive statistics were used to describe system use patterns, ratings of multiple satisfaction items (1=not at all; 3=to a great extent), and provider-selected motivators of and barriers to using DialysisConnect. RESULTS: The pilot version of DialysisConnect included clinical information that was automatically uploaded from dialysis facilities, forms for entering critical admission and discharge information, and a direct communication channel. Although physicians comprised most of the potential users of DialysisConnect, APPs and dialysis nurses were the most active users. Activities were unevenly distributed; for example, 1 hospital-based APP recorded most of the admissions (280/309, 90.6%) among patients treated at the pilot dialysis facilities. End-of-pilot ratings of DialysisConnect were generally higher for users versus nonusers (eg, "I can see the potential value of DialysisConnect for my work with dialysis patients": mean 2.8, SD 0.4, vs mean 2.3, SD 0.6; P=.02). Providers most commonly selected reduced time and energy spent gathering information as a motivator (11/26, 42%) and a lack of time to use the system as a barrier (8/26, 31%) at the end of the pilot. CONCLUSIONS: This pilot study found that APPs and nurses were most likely to engage with the system. Survey participants generally viewed the system favorably while identifying substantial barriers to its use. These results inform how best to motivate providers to use this system and similar systems and inform future pragmatic research in care coordination among this and other populations.

Effectiveness of a Web-Based Provider Communications Platform in Reducing Hospital Readmissions Among Patients Receiving Dialysis: A Pilot Pre-Post Study.

Rationale & Objective: Suboptimal care coordination between dialysis facilities and hospitals is an important driver of 30-day hospital readmissions among patients receiving dialysis. We examined whether the introduction of web-based communications platform ("DialysisConnect") was associated with reduced hospital readmissions. Study Design: Pilot pre-post study. Setting & Participants: A total of 4,994 index admissions at a single hospital (representing 2,419 patients receiving dialysis) during the study period (January 1, 2019-May 31, 2021). Intervention: DialysisConnect was available to providers at the hospital and 4 affiliated dialysis facilities (=intervention facilities) during the pilot period (November 1, 2020-May 31, 2021). Outcomes: The primary outcome was 30-day readmission; secondary outcomes included 30-day emergency department visits and observation stays. Interrupted time series and linear models with generalized estimating equations were used to assess pilot versus prepilot differences in outcomes; difference-in-difference analyses were performed to compare these differences between intervention versus control facilities. Sensitivity analyses included a third, prepilot/COVID-19 period (March 1, 2020-October 31, 2020). Results: There was no statistically significant difference in the monthly trends in the 30-day readmissions pilot versus prepilot periods (-0.60 vs -0.13, P = 0.85) for intervention facility admissions; the difference-in-difference estimate was also not statistically significant (0.54 percentage points, P = 0.83). Similar analyses including the prepilot/COVID-19 period showed that, despite a substantial drop in admissions at the start of the pandemic, there were no statistically significant differences across the 3 periods. The age-, sex-, race-, and comorbid condition-adjusted, absolute pilot versus prepilot difference in readmissions rate was 1.8% (-3.7% to 7.3%); similar results were found for other outcomes. Limitations: Potential loss to follow-up and pandemic effects. Conclusions: In this pilot, the introduction of DialysisConnect was not associated with reduced hospital readmissions. Tailored care coordination solutions should be further explored in future, multisite studies to improve the communications gap between dialysis facilities and hospitals.

The COVID-19 nephrology compendium: AKI, CKD, ESKD and transplantation.

The pandemic of coronavirus disease 2019 (CoVID-19) has been an unprecedented period. The disease afflicts multiple organ systems, with acute kidney injury (AKI) a major complication in seriously ill patients. The incidence of AKI in patients with CoVID-19 is variable across numerous international studies, but the high incidence of AKI and its associated worse outcomes in the critical care setting are a consistent finding. A multitude of patterns and mechanisms of AKI have been elucidated, and novel strategies to address shortage of renal replacement therapy equipment have been implemented. The disease also has had consequences on longitudinal management of patients with chronic kidney disease and end stage kidney disease. Kidney transplant recipients may be especially susceptible to CoVID-19 as a result of immunosuppression, with preliminary studies demonstrating high mortality rates. Increased surveillance of disease with low threshold for testing and adjustment of immunosuppression regimen during acute periods of illness have been recommended.