scholarly journals Investigating the impact of weakly supervised data on text mining models of publication transparency: a case study on randomized controlled trials

Author(s):  
Linh Hoang ◽  
Lan Jiang ◽  
Halil Kilicoglu

AbstractLack of large quantities of annotated data is a major barrier in developing effective text mining models of biomedical literature. In this study, we explored weak supervision strategies to improve the accuracy of text classification models developed for assessing methodological transparency of randomized controlled trial (RCT) publications. Specifically, we used Snorkel, a framework to programmatically build training sets, and UMLS-EDA, a data augmentation method that leverages a small number of existing examples to generate new training instances, for weak supervision and assessed their effect on a BioBERT-based text classification model proposed for the task in previous work. Performance improvements due to weak supervision were limited and were surpassed by gains from hyperparameter tuning. Our analysis suggests that refinements to the weak supervision strategies to better deal with multi-label case could be beneficial.

2020 ◽  
Author(s):  
Martina Bientzle ◽  
Marie Eggeling ◽  
Simone Korger ◽  
Joachim Kimmerle

BACKGROUND: Successful shared decision making (SDM) in clinical practice requires that future clinicians learn to appreciate the value of patient participation as early as in their medical training. Narratives, such as patient testimonials, have been successfully used to support patients’ decision-making process. Previous research suggests that narratives may also be used for increasing clinicians’ empathy and responsiveness in medical consultations. However, so far, no studies have investigated the benefits of narratives for conveying the relevance of SDM to medical students.METHODS: In this randomized controlled experiment, N = 167 medical students were put into a scenario where they prepared for medical consultation with a patient having Parkinson disease. After receiving general information, participants read either a narrative patient testimonial or a fact-based information text. We measured their perceptions of SDM, their control preferences (i.e., their priorities as to who should make the decision), and the time they intended to spend for the consultation.RESULTS: Participants in the narrative patient testimonial condition referred more strongly to the patient as the one who should make decisions than participants who read the information text. Participants who read the patient narrative also considered SDM in situations with more than one treatment option to be more important than participants in the information text condition. There were no group differences regarding their control preferences. Participants who read the patient testimonial indicated that they would schedule more time for the consultation.CONCLUSIONS: These findings show that narratives can potentially be useful for imparting the relevance of SDM and patient-centered values to medical students. We discuss possible causes of this effect and implications for training and future research.


2020 ◽  
Vol 4 (Supplement_1) ◽  
pp. 774-774
Author(s):  
David Rein ◽  
Madeleine Hackney ◽  
Michele Dougherty ◽  
Camille Vaughan ◽  
Laurie Imhof ◽  
...  

Abstract The STEADI Options trial uses a randomized, controlled-trial design to assess the effectiveness and cost-effectiveness of the STEADI Initiative . Beginning March, 2020, we will randomize 3,000 adults ≥ 65 years of age at risk for falls seen in an Emory Clinic primary care practice to: (1) full STEADI; (2) a STEADI-derived gait, balance, and strength assessment with physical therapy referrals; (3) a STEADI-derived medication review and management; or (4) usual care. This presentation will discuss decisions made by the study team to facilitate implementation of STEADI including electronically conducting screening prior to the date of encounter, the use of dedicated nursing staff to conduct assessments, implementation of strength, balance, orthostatic hypotension, and vision testing, methods to facilitate medication review, and communication of assessment information to providers. The results from this study will be used to estimate the impact of STEADI on falls, service utilization, and costs over one year.


2021 ◽  
Vol 20 (1) ◽  
Author(s):  
Joseph M. Wagman ◽  
Kenyssony Varela ◽  
Rose Zulliger ◽  
Abuchahama Saifodine ◽  
Rodaly Muthoni ◽  
...  

Abstract Background The need to develop new products and novel approaches for malaria vector control is recognized as a global health priority. One approach to meeting this need has been the development of new products for indoor residual spraying (IRS) with novel active ingredients for public health. While initial results showing the impact of several of these next-generation IRS products have been encouraging, questions remain about how to best deploy them for maximum impact. To help address these questions, a 2-year cluster-randomized controlled trial to measure the impact of IRS with a microencapsulated formulation of pirimiphos-methyl (PM) in an area with high ownership of long-lasting insecticidal nets (LLINs) was conducted in a high-transmission district of central Mozambique with pyrethroid resistant vectors. Presented here are the results of the vector surveillance component of the trial. Methods The 2 year, two-armed trial was conducted in Mopeia District, Zambezia Province, Mozambique. In ten sentinel villages, five that received IRS with PM in October–November 2016 and again in October–November 2017 and five that received no IRS, indoor light trap collections and paired indoor-outdoor human landing collections catches (HLCs) were conducted monthly from September 2016 through October 2018. A universal coverage campaign in June 2017, just prior to the second spray round, distributed 131,540 standard alpha-cypermethrin LLINs across all study villages and increased overall net usage rates in children under 5 years old to over 90%. Results The primary malaria vector during the trial was Anopheles funestus sensu lato (s.l.), and standard World Health Organization (WHO) tube tests with this population indicated variable but increasing resistance to pyrethroids (including alpha-cypermethrin, from > 85% mortality in 2017 to 7% mortality in 2018) and uniform susceptibility to PM (100% mortality in both years). Over the entire duration of the study, IRS reduced An. funestus s.l. densities by 48% (CI95 33–59%; p < 0.001) in indoor light traps and by 74% (CI95 38–90%; p = 0.010) during indoor and outdoor HLC, though in each study year reductions in vector density were consistently greatest in those months immediately following the IRS campaigns and waned over time. Overall there was no strong preference for An. funestus to feed indoors or outdoors, and these biting behaviours did not differ significantly across study arms: observed indoor-outdoor biting ratios were 1.10 (CI95 1.00–1.21) in no-IRS villages and 0.88 (CI95 0.67–1.15) in IRS villages. The impact of IRS was consistent in reducing HLC exposures both indoors (75% reduction: CI95 47–88%; p = 0. < 0.001) and outdoors (68% reduction: CI95 22–87%; p = 0.012). While substantially fewer Anopheles gambiae s.l. were collected during the study, trends show a similar impact of IRS on this key vector group as well, with a 33% (CI95 7–53%; p = 0.019) reduction in mosquitoes collected in light traps and a non-statistically significant 39% reduction (p = 0.249) in HLC landing rates. Conclusion IRS with PM used in addition to pyrethroid-only LLINs substantially reduced human exposures to malaria vectors during both years of the cluster-randomized controlled trial in Mopeia—a high-burden district where the primary vector, An. funestus s.l., was equally likely to feed indoors or outdoors and demonstrated increasing resistance to pyrethroids. Findings suggest that IRS with PM can provide effective vector control, including in some settings where pyrethroid-only ITNs are widely used. Trial registrationclinicaltrials.gov, NCT02910934. Registered 22 September 2016, https://www.clinicaltrials.gov/ct2/show/NCT02910934.


2021 ◽  
pp. 019459982199474
Author(s):  
Maggie Xing ◽  
Dorina Kallogjeri ◽  
Jay F. Piccirillo

Objective To evaluate the effectiveness of cognitive training in improving tinnitus bother and to identify predictors of patient response. Study Design Prospective open-label randomized controlled trial. Setting Online. Methods Participants were adults with subjective idiopathic nonpulsatile tinnitus causing significant tinnitus-related distress. The intervention group trained by using auditory-intensive exercises for 20 minutes per day, 5 days per week, for 8 weeks. The active control group trained on the same schedule with non–auditory intensive games. Surveys were completed at baseline, 8 weeks, and 12 weeks. Results A total of 64 participants completed the study. The median age was 63 years (range, 25-69) in the intervention group and 61 years (34-68) in the control group. Mixed model analysis revealed that within-subject change in Tinnitus Functional Index in the intervention group was not different than the control group, with marginal mean differences (95% CI): 0.24 (–11.20 to 10.7) and 2.17 (–8.50 to 12.83) at 8 weeks and 2.33 (–8.6 to 13.3) and 3.36 (–7.91 to 14.6) at 12 weeks, respectively. When the 2 study groups were compared, the control group had higher Tinnitus Functional Index scores than the intervention group by 10.5 points at baseline (95% CI, –0.92 to 29.89), 8.1 at 8 weeks (95% CI, –3.27 to 19.42), and 9.4 at 12 weeks (95% CI, –2.45 to 21.34). Conclusion Auditory-intensive cognitive training was not associated with changes in self-reported tinnitus bother. Given the potential for neuroplasticity to affect tinnitus, we believe that future studies on cognitive training for tinnitus remain relevant.


Author(s):  
Dorien Vanden Bossche ◽  
Susan Lagaert ◽  
Sara Willems ◽  
Peter Decat

Background: During the COVID-19 pandemic, many primary care professionals were overburdened and experienced difficulties reaching vulnerable patients and meeting the increased need for psychosocial support. This randomized controlled trial (RCT) tested whether a primary healthcare (PHC) based community health worker (CHW) intervention could tackle psychosocial suffering due to physical distancing measures in patients with limited social networks. Methods: CHWs provided 8 weeks of tailored psychosocial support to the intervention group. Control group patients received ‘care as usual’. The impact on feelings of emotional support, social isolation, social participation, anxiety and fear of COVID-19 were measured longitudinally using a face-to-face survey to determine their mean change from baseline. Self-rated change in psychosocial health at 8 weeks was determined. Results: We failed to find a significant effect of the intervention on the prespecified psychosocial health measures. However, the intervention did lead to significant improvement in self-rated change in psychosocial health. Conclusions: This study confirms partially the existing evidence on the effectiveness of CHW interventions as a strategy to address mental health in PHC in a COVID context. Further research is needed to elaborate the implementation of CHWs in PHC to reach vulnerable populations during and after health crises.


2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Ricardo Castro ◽  
Eduardo Kattan ◽  
Giorgio Ferri ◽  
Ronald Pairumani ◽  
Emilio Daniel Valenzuela ◽  
...  

Abstract Background Persistent hyperlactatemia has been considered as a signal of tissue hypoperfusion in septic shock patients, but multiple non-hypoperfusion-related pathogenic mechanisms could be involved. Therefore, pursuing lactate normalization may lead to the risk of fluid overload. Peripheral perfusion, assessed by the capillary refill time (CRT), could be an effective alternative resuscitation target as recently demonstrated by the ANDROMEDA-SHOCK trial. We designed the present randomized controlled trial to address the impact of a CRT-targeted (CRT-T) vs. a lactate-targeted (LAC-T) fluid resuscitation strategy on fluid balances within 24 h of septic shock diagnosis. In addition, we compared the effects of both strategies on organ dysfunction, regional and microcirculatory flow, and tissue hypoxia surrogates. Results Forty-two fluid-responsive septic shock patients were randomized into CRT-T or LAC-T groups. Fluids were administered until target achievement during the 6 h intervention period, or until safety criteria were met. CRT-T was aimed at CRT normalization (≤ 3 s), whereas in LAC-T the goal was lactate normalization (≤ 2 mmol/L) or a 20% decrease every 2 h. Multimodal perfusion monitoring included sublingual microcirculatory assessment; plasma-disappearance rate of indocyanine green; muscle oxygen saturation; central venous-arterial pCO2 gradient/ arterial-venous O2 content difference ratio; and lactate/pyruvate ratio. There was no difference between CRT-T vs. LAC-T in 6 h-fluid boluses (875 [375–2625] vs. 1500 [1000–2000], p = 0.3), or balances (982[249–2833] vs. 15,800 [740–6587, p = 0.2]). CRT-T was associated with a higher achievement of the predefined perfusion target (62 vs. 24, p = 0.03). No significant differences in perfusion-related variables or hypoxia surrogates were observed. Conclusions CRT-targeted fluid resuscitation was not superior to a lactate-targeted one on fluid administration or balances. However, it was associated with comparable effects on regional and microcirculatory flow parameters and hypoxia surrogates, and a faster achievement of the predefined resuscitation target. Our data suggest that stopping fluids in patients with CRT ≤ 3 s appears as safe in terms of tissue perfusion. Clinical Trials: ClinicalTrials.gov Identifier: NCT03762005 (Retrospectively registered on December 3rd 2018)


Sign in / Sign up

Export Citation Format

Share Document