Text Feedback Research Articles

50 Million Followers Can’t Be Wrong, or Can They? The Influence of Social Media on CEO Communication

The rise of social media introduces new ways in which executives communicate and receive feedback on their words and actions. Social media feedback is generated quickly in large volumes with a format that discourages in-depth critical feedback (e.g., availability of only “like” option), from partly unknown (and unknowable) and heterogeneous sources. In this paper, I theorize and test how long-term exposure to social media feedback influences the communication patterns of CEOs. Through applying novel machine learning methods on 820,000 communication threads of CEOs from S&P 1500, I found that long-term exposure to aggregate social media feedback increases the frequency and affective tone of communication. The relationship is moderated by recent textual feedback. These findings have important implications for the literature on CEO communications and feedback.

User Engagement Among Diverse Adults in a 12-Month Text Message-Delivered Diabetes Support Intervention: Results from a Randomized Controlled Trial.

BackgroundText message–delivered interventions are a feasible and scalable approach for improving chronic disease self-care and reducing health disparities; however, information on long-term user engagement with these interventions is limited.ObjectiveThe aim of this study is to examine user engagement in a 12-month text message–delivered intervention supporting diabetes self-care, called REACH (Rapid Education/Encouragement And Communications for Health), among racially and socioeconomically diverse patients with type 2 diabetes (T2D). We explored time trends in engagement, associations between patient characteristics and engagement, and whether the addition of a human component or allowing patients to change their text frequency affected engagement. Qualitative data informed patients’ subjective experience of their engagement.MethodsWe recruited patients with T2D for a randomized trial evaluating mobile phone support relative to enhanced treatment as usual. This analysis was limited to participants assigned to the intervention. Participants completed a survey and hemoglobin A1c (HbA1c) test and received REACH text messages, including self-care promotion texts, interactive texts asking about medication adherence, and adherence feedback texts. For the first 6 months, texts were sent daily, and half of the participants also received monthly phone coaching. After 6 months, coaching stopped, and participants had the option to receive fewer texts for the subsequent 6 months. We defined engagement via responses to the interactive texts and responses to a follow-up interview. We used regression models to analyze associations with response rate and thematic and structural analysis to understand participants’ reasons for responding to the texts and their preferred text frequency.ResultsThe participants were, on average, aged 55.8 (SD 9.8) years, 55.2% (137/248) female, and 52.0% (129/248) non-White; 40.7% (101/248) had ≤ a high school education, and 40.7% (101/248) had an annual household income <US $25,000. The median response rate to interactive texts was 91% (IQR 75%-97%) over 12 months. Engagement gradually declined throughout the intervention but remained high. Engagement did not differ by age, gender, education, income, diabetes duration, insulin status, health literacy, or numeracy. Black race and worse baseline medication adherence and HbA1c were each associated with lower engagement, although the effects were small. Nearly half of the participants chose to continue receiving daily texts for the last 6 months of the intervention. Participants who continued daily text messages said they wanted to continue experiencing benefits to their health, whereas those who chose fewer texts said that the daily texts had helped them create routines and they no longer needed them as often. Engagement was not impacted by receiving coaching or by participants’ chosen text frequency.ConclusionsWell-designed interactive text messages can engage diverse patients in a self-care intervention for at least 1 year. Variation in and reasons for frequency preference suggest that offering a frequency choice may be important to users’ engagement.

Assessing the black box of feedback neglect in a digital educational game for elementary school

ABSTRACT Background: Previous research shows that critical constructive feedback, that scaffolds students to improve on tasks, often remains untapped. The paper’s aim is to illuminate at what stages students provided with such feedback drop out of feedback processing. Methods: In our model, students can drop out at any of five stages of feedback processing: (1) noticing, (2) decoding, (3) making sense, (4) acting upon, and (5) using feedback to make progress. Eye-tracking was used to measure noticing and decoding of feedback. Behavioral data-logging tracked students’ use of feedback and potential progress. Three feedback signaling conditions were experimentally compared: a pedagogical agent, an animated arrow, and no signaling (control condition). Findings: Students dropped out at each stage and few made it past the final stage. The agent condition led to significantly less feedback neglect at the two first stages, suggesting that students who are not initially inclined to notice and read feedback text can be influenced into doing so. Contribution: The study provides a model and method to build more fine-grained knowledge of students’ (non)processing of feedback. More knowledge on at what stages students drop out and why can inform methods to counteract drop out and scaffold more productive and fruitful responses.

A new image encryption algorithm based on optimized Lorenz chaotic system

SummaryIn this article, an improved Lorenz system was constructed, in the first stage, a nonlinear term in general Lorenz system was replaced by the sum of term of exponential function and term of the square of single variable, then the dynamic characteristics of the improved Lorenz system were analyzed through the analysis on the phase diagrams and its Lyapunov exponents, and it is proved that the new system has good chaotic characteristics. In the second stage, to improve the security and anti‐attack ability of image encryption algorithm, a new image encryption algorithm was proposed based on this improved Lorenz chaotic system, which was used to generate two groups of key sequences which were used in scrambling encryption and alternative encryption. In the scrambling encryption stage, the gray values of image pixels were sorted in ascending order, and combining with the sequence of scrambled encryption key, the pixel position scrambling was realized. In the phase of the pixel value replacement encryption stage, the gray values of the pixels were changed by the method of cipher text feedback. In the third stage, the quantitative analysis of the effect of encryption was carried out by using histogram, information entropy, correlation coefficient of adjacent pixels. The experimental results show that the improved Lorenz formula is a nonlinear chaotic system, the encrypted algorithm based on this chaotic system can effectively resist chosen plaintext attack, in the encrypted image, the grey value shave random‐like distribution behavior and the adjacent pixels have zero correlation. The algorithm has a 299 bit key space and it is very sensitive to the key.

Topic Modeling and Sentiment Analysis of Martial Arts Learning Textual Feedback on YouTube

Topic Modeling and Sentiment Analysis of Martial Arts Learning Textual Feedback on YouTube

Quantitative Evaluation of Sentiment Analysis on E-Commerce Data

With the emergence of technology and Big Data in the modern world, it is important to govern such vast amounts of data. Amazon and other e-commerce websites play a vital role in the delivery of goods services to the user. Nonetheless, such services are frequently accompanied by consumer reviews and ratings of the things they sell. Such reviews and ratings provided in the form of textual feedback serve to improve service delivery and product quality in the event that a consumer is dissatisfied. Thus, the process of evaluations and ratings is regarded as a significant component of the customer satisfaction process. Thus, it is vital to analyse them so that useful insights can be gleaned from them and the correct judgements can be made regarding the enhancement of the product. With emphasis on this concept, the authors of the proposed paper intend to develop a model capable of performing sentiment analysis on customer-provided product reviews and analysing products with positive and negative comments. On the basis of these reviews, three machine learning algorithms, primarily Decision Trees, XGBoost, and AdaBoost, are utilised to undertake sentiment analysis on products. It was noticed that XGBoost produced a greater detection accuracy of 84.45%.

Algorithmic Improvements for Deep Reinforcement Learning Applied to Interactive Fiction

Text-based games are a natural challenge domain for deep reinforcement learning algorithms. Their state and action spaces are combinatorially large, their reward function is sparse, and they are partially observable: the agent is informed of the consequences of its actions through textual feedback. In this paper we emphasize this latter point and consider the design of a deep reinforcement learning agent that can play from feedback alone. Our design recognizes and takes advantage of the structural characteristics of text-based games. We first propose a contextualisation mechanism, based on accumulated reward, which simplifies the learning problem and mitigates partial observability. We then study different methods that rely on the notion that most actions are ineffectual in any given situation, following Zahavy et al.'s idea of an admissible action. We evaluate these techniques in a series of text-based games of increasing difficulty based on the TextWorld framework, as well as the iconic game Zork. Empirically, we find that these techniques improve the performance of a baseline deep reinforcement learning agent applied to text-based games.

P97 Implementation of a regional trainee-led leadership and management programme for rheumatology trainees

Abstract Background We identified a lack of formal learning opportunities provided by Health Education England North West (HEE (NW)) to achieve the many leadership and management requirements stipulated by the Joint Royal College of Physicians’ Training Board Speciality Training (ST) rheumatology curriculum. This gap was highlighted through discussion at the HEE (NW) Specialty Training Committee by trainee representatives and regional Training Programme Directors (TPD), in 2017. Furthermore, the merger of two pre-existing Deaneries (Mersey and NW) into the NW Local Education and Training Board (LETB) created a pressing need to develop a network across the region. Methods Two regional ST representatives engaged with trainees across the new LETB through formal surveys and informal conversations regarding training to address this. Specialty trainees (STRs) elected a bi-annual half-day session in conjunction with an existing clinical regional event, the North West Rheumatology Club (NWRC). Sessions have covered a range of topics (Table 1) and provide an opportunity for networking across the deaneries. Learning objectives are mapped against ST curriculum. HEE (NW) TPDs support the programme, facilitating study leave. Meeting costs are supported by the NWRC. Trainees shape future sessions submitting topic ideas and have autonomy over content and speakers. Formal anonymous sessional feedback uses a Likert scale evaluating whether learning objectives have been met and quality of session delivery. Now entering its third year the scheme is well-established, with succession planning in place. Results With feedback response rates of 85% (mean across 4 sessions), median scores are 5 (IQR 4.9-5) for meeting learning objectives and 5 (IQR 5-5) for session delivery. Free text feedback is overwhelmingly positive. Six months after sessions trainees reported ‘feeling more encouraged to raise concerns’, ‘being more proactive in whistleblowing‘, ‘having a better understanding of conflict situations’, and ‘better awareness of reasons for difference in individual behaviours’. Conclusion Filling unmet training needs in leadership and management through trainee-led sessions provides an excellent learning opportunity for STRs both attending and leading the events, and facilitates cross-regional collaboration. This programme demonstrates the role of stakeholders to inform programme design successfully, showing excellent long-term impact in an important and previously underserved curriculum requirement. Disclosures R. Benson None. C. Sharp None. E. Macphie Grants/research support; EM is the Secretary of the North West Rheumatology Club; meetings are supported by an unrestricted educational grant from UCB.

Effectiveness of an Ecological Momentary Intervention for Reducing Risky Alcohol Consumption Among Young Adults: Protocol for a Three-Arm Randomized Controlled Trial.

BackgroundRecent research has investigated the utility of mobile phone–delivered interventions for reducing risky single-occasion drinking, also known as binge drinking. In the past five years, focus has been placed on ecological momentary interventions (EMIs), which aim to deliver intervention content in correspondence to real-time assessments of behavior, also known as ecological momentary assessments (EMAs).ObjectiveThis study aims to assess the effect of a fully automated, tailored, mobile phone–delivered EMI termed Mobile Intervention for Drinking in Young people (MIDY) on young people's risky single-occasion drinking behavior.MethodsWe will use a three-armed randomized controlled trial design to determine the impact of MIDY on peak consumption of alcohol among young people. A list of mobile telephone numbers for random digit dialing will be generated, and researchers will telephone potential participants to screen for eligibility. Participants will be randomized into one of three intervention groups. For 6 weeks, EMI, EMA, and attention control groups will complete hourly EMA surveys on their mobile phones on Friday and Saturday nights. EMI participants will receive personalized feedback in the form of text messages corresponding to their EMA survey responses, which focus on alcohol consumption, spending, and mood. EMA participants will not receive feedback. A third group will also complete EMA and receive feedback text messages at the same time intervals, but these will be focused on sedentary behavior and technology use. All groups will also complete a short survey on Saturday and Sunday mornings, with the primary outcome measure taken on Sunday mornings. A more detailed survey will be sent on the final Sunday of the 6-week period, and then again 1 year after recruitment.ResultsThe primary outcome measure will be an observed change (ie, reduction) in the mean peak number of drinks consumed in a single night over the 6-week intervention period between the EMI and attention control groups as measured in the weekly EMA. We expect to see a greater reduction in mean peak drinking in the EMI group compared to that in the attention control group. As a secondary aim, we will assess whether mean peak drinking is reduced in the EMA group compared to the attention control group. We will use a random-effects mixed-modeling approach using maximum-likelihood estimation to provide estimates of differences in peak drinking across time periods between those receiving the intervention (EMI) and attention control participants. An intention-to-treat approach will be taken for the analysis. Individuals and study groups will be modeled as random and fixed factors, respectively.ConclusionsThis study extends our previous work investigating the efficacy of a mobile EMI (MIDY) for reducing risky drinking among young adults in Australia, and will add to the expanding literature on the use of mobile interventions for reducing risky alcohol consumption.Trial RegistrationAustralian New Zealand Clinical Trials Registration (ANZCTR): ACTRN12617001509358p; http://www.anzctr.org.au/ACTRN12617001509358p.aspxInternational Registered Report Identifier (IRRID)DERR1-10.2196/14190

Learner participation regulation supported by long-term peer moderation and participation feedback during asynchronous discussions

Active participation in asynchronous online discussions is crucial for learning, but low student participation remains a persistent problem. This design research project presents two interventions drawn from self/co-regulation and feedback literature: (1) extended peer moderation periods and (2) multimodal participation feedback. In peer moderation, students are assigned as leaders within the discussion and are responsible for posting critical questions. Their active participation influences others through co-regulation. However, assigned moderators may not perform their leadership roles as expected. How moderators facilitate discussion may depend on their own self-regulation since they exhibit self-regulating behaviors of planning, monitoring and evaluating while choosing facilitation strategies. Feedback, like visual representations of students’ participation, enables students to monitor their own participation. Therefore, to support moderators’ development of self-regulation, we developed written guidance for moderating a discussion and feedback to help moderators monitor and facilitate discussions more effectively. To track assigned peer moderators’ participation, we used a computational model using social network analysis. This information was then provided to students as a composite, three-part feedback: a graphic, tabulated participation levels, and textual feedback. We tested our design within a graduate-level asynchronous online course for 12 weeks to study how peer moderators conducted their assigned leadership role during the extended moderation period and how multimodal feedback supported students’ regulation of participation. Results show that not all moderators accepted their roles, but those who did showed high participation levels and leadership behaviors. Dynamic relationships among moderators created different self/co-regulation in the group. Students found the multimodal participation feedback useful for monitoring their participation. From these findings, we propose several design solutions for supporting peer moderators’ development of regulation through scaffolds and multimodal feedback.

Perceptions of resuscitation care among in-hospital cardiac arrest responders: a qualitative analysis

BackgroundIn-hospital cardiac arrests (IHCA) occur commonly and are associated with poor survival and variable outcomes. This study aimed to directly survey IHCA responders to understand their perceptions of resuscitation care.MethodsAs part of a quality improvement initiative, we surveyed participating providers of IHCAs at our institution from Jan 2014 to May 2016. The survey included unstructured free text feedback, which was the focus of this study. We systematically coded the free text and organized identifiable latent themes using thematic analysis.We used the natural timeline of an IHCA – pre-arrest, arrest, and post-arrest – for organization of the identifiable latent themes, and created a separate category for holistic remarks that arched across the timeline.ResultsWe identified 172 IHCAs with a mean of 1.7 responses per arrest (range: 1–8 responses). The mean age of this patient population was 59 years at the time of arrest, and 107 (62%) were men.We identified several themes - [1] issues around code activation and code status characterized the pre-arrest period [2] ,team interactions and issues around supplies/equipment dominated the intra-arrest period, and [3] code cessation and transitions of care typified the post-arrest period. Holistic remarks focused on attentiveness paid by the arrest team to patient comfort and family. Some comments reflected positive experiences but most focused on areas of improvement consistent with the initiative’s purpose. In certain cases, we identified a tension between the need to balance established resuscitation protocols with flexibility required by real-life circumstances.ConclusionsDirectly surveying those who participated in IHCAs led to novel insights about their experiences. Our findings suggest that parsing through such qualitative feedback can help hospitals identify areas of improvement, modulate expectations, temper emotions, and refine protocols.

Implementation of Entrustable Professional Activities into a General Surgery Residency

Concerns over resident ability to practice effectively after graduation have led to the competency-based medical education movement. Entrustable professional activities (EPAs) may facilitate competency-based medical education in surgery, but implementation is challenging. This manuscript describes 1 strategy used to implement EPAs into an academic general surgery residency. A mobile application was developed incorporating 5 EPAs developed by the American Board of Surgery; residents and faculty from the Departments of Surgery, Emergency Medicine, and Hospital Medicine at a single tertiary care center were trained in its use. Entrustment levels and free text feedback were collected. Self-assessment was paired with supervisor assessment, and faculty assessments were used to inform clinical competency committee entrustment decisions. Feedback was regularly solicited from app users and results distributed on a monthly basis. One thousand seven hundred and twenty microassessments were collected over the first 16 months of implementation; 898 (47.8%) were performed by faculty with 569 (66.0%) matched pairs. Engagement was skewed with small numbers of high performers in both resident and faculty groups. Continued development of resident and faculty was required to sustain engagement with the program. Nonsurgical specialties contributed significantly to resident assessments (496, 28.8%). EPAs are being successfully integrated into the assessment framework at our institution. EPA implementation in surgery residency is a long-term process that requires investment, but may address limitations in the current assessment framework.

Medical training in palliative care at the Hannover Medical School: development of the interdisciplinary and cross-sectoral education in the cross-sectional subject Q13 (WEISE-Q13)

The implementation of the cross-sectional subject of palliative care (Q13) at medical faculties in Germany is heterogeneous. In faculties without achair for palliative medicine, other solutions are needed. This study describes the development and implementation process of the new educational concept in Q13 at the Hannover Medical School and its evaluation by medical students and lecturers. The new educational concept was developed in 2017/18 following the six-step approach of curriculum development by Kern, and was implemented and tested in the academic year 2018/19. Evaluation was assessed in two steps, both in qualitative and quantitative terms (mixed methods): an anonymous written evaluation by students with the option of free text answers and feedback discussions with lecturers. Students and lecturers welcomed the new educational concept in equal measures. Students thought that education in Q13 stimulates reflection on the individual attitude towards the issues of death and dying and prepares them for encounters with terminally ill and dying people. The case-based learning units in particular led to aconsolidation of theoretical knowledge (i.e. pain management) with adequate patient orientation. Teachers positively highlighted the organization and the educational material. Content structuring was considered to be particularly helpful. With the new teaching concept, medical students feel well prepared for the treatment of terminally ill and dying patients. The combination of different learning formats with an interdisciplinary and cross-sectoral focus promotes education with apractical orientation.

The Use of Web-Based Support Groups Versus Usual Quit-Smoking Care for Men and Women Aged 21-59 Years: Protocol for a Randomized Controlled Trial.

BackgroundExisting smoking cessation treatments are challenged by low engagement and high relapse rates, suggesting the need for more innovative, accessible, and interactive treatment strategies. Twitter is a Web-based platform that allows people to communicate with each other throughout the day using their phone.ObjectiveThis study aims to leverage the social media platform of Twitter for fostering peer-to-peer support to decrease relapse with quitting smoking. Furthermore, the study will compare the effects of coed versus women-only groups on women’s success with quitting smoking.MethodsThe study design is a Web-based, three-arm randomized controlled trial with two treatment arms (a coed or women-only Twitter support group) and a control arm. Participants are recruited online and are randomized to one of the conditions. All participants will receive 8 weeks of combination nicotine replacement therapy (patches plus their choice of gum or lozenges), serial emails with links to Smokefree.gov quit guides, and instructions to record their quit date online (and to quit smoking on that date) on a date falling within a week of initiation of the study. Participants randomized to a treatment arm are placed in a fully automated Twitter support group (coed or women-only), paired with a buddy (matched on age, gender, location, and education), and encouraged to communicate with the group and buddy via daily tweeted discussion topics and daily automated feedback texts (a positive tweet if they tweet and an encouraging tweet if they miss tweeting). Recruited online from across the continental United States, the sample consists of 215 male and 745 female current cigarette smokers wanting to quit, aged between 21 and 59 years. Self-assessed follow-up surveys are completed online at 1, 3, and 6 months after the date they selected to quit smoking, with salivary cotinine validation at 3 and 6 months. The primary outcome is sustained biochemically confirmed abstinence at the 6-month follow-up.ResultsFrom November 2016 to September 2018, 960 participants in 36 groups were recruited for the randomized controlled trial, in addition to 20 participants in an initial pilot group. Data analysis will commence soon for the randomized controlled trial based on data from 896 of the 960 participants (93.3%), with 56 participants lost to follow-up and 8 dropouts.ConclusionsThis study combines the mobile platform of Twitter with a support group for quitting smoking. Findings will inform the efficacy of virtual peer-to-peer support groups for quitting smoking and potentially elucidate gender differences in quit rates found in prior research.Trial RegistrationClinicalTrials.gov NCT02823028; https://clinicaltrials.gov/ct2/show/NCT02823028International Registered Report Identifier (IRRID)DERR1-10.2196/16417

Ranking chat logs for crime investigation

Aiming at the needs of criminal investigators to quickly find chat logs related to crime case, a ranking method for chat texts has been proposed. First, query keywords were used to search chat texts, and the feedback texts were divided into relevant texts and irrelevant texts with manual annotation, then non-negative matrix factorization (NMF) was utilized to obtain the implicit semantic relationship of chat texts, finally chat logs could be ranked based on the scores calculated by similarity. Experiments show that the method proposed in this paper can quickly retrieve and get the messages of interest in a lot of chat logs, which can facilitate crime investigation.

Application of reflective journals to assess self-directed learning in a blended learning setting: a case study in Hong Kong

This paper presents the case study of a blended learning subject offered to a group of engineering students, in which a reflective journal was adopted as part of the assessment. It also presents an innovative attempt to explore the relationship between the reflections of students and their performance in the subject. The feedbacks from students were collected and analysed using text-mining techniques, and a machine-learning algorithm was used to identify the association between the reflective feedback text and the corresponding final grades of the students. The supervised machine-learning algorithm produces an inferred function from the training data so as to make predictions about the output values of the testing data. The results prove that reflective journals can be a valuable means of assessing student learning in a blended learning environment, and also offers a good reference for educators to have a better understanding regarding the performance of students.

The different effects of daily-life instant response social media and an educational feedback system on flipped learning: from the evidence of behavioral analysis

ABSTRACT In this study, we used the application software installed on students’ smart phones as instantly interactive tools during class time in a flipped learning context. One popular daily-life instant interaction application, Line, provided the students with text or multimedia feedback in time sequence; one non-daily instant interaction application, PAL+, provided the students with inscribed feedback in time sequence; and one educational instant interaction application, SpeakUP!, provided the students with anonymous feedback in time sequence. We compared the students’ behaviors of using the different interactive systems for interactions in the flipped classroom class time. The study applied learning analytics techniques and used behavioral analysis to identify the role of the daily-life instant response application in flipped learning, and found how the instant feedback systems enhanced discussion in the classroom.

Student Perceptions of Feedback in Higher Education

Text Feedback is critical to students’ academic development in higher education. Despite this, evidence suggests that students do not consistently engage with feedback or recognise the value of it. This study explored student perceptions of feedback in one university in England. Data were collected using focus groups. The results indicated that the participants valued feedback that is detailed and personal. They also demonstrated a preference for verbal feedback rather than written feedback. Participants recognised the benefits and limitations of peer feedback and there was evidence to suggest that participants valued the judgements of their lecturers above those of their peers. The data indicate that lecturers should utilise a range of feedback modes, including face-to-face, verbal, written, audio and video feedback. The study suggests that the use of written feedback in higher education may not be effective because students may not engage effectively with it, particularly if they achieve a high grade. Taking into consideration the important role that feedback plays in promoting learning, it is vital that modes of feedback are used which students are likely to engage with. Given the fact that students have different preferences, it is therefore suggested that lecturers utilise a variety of modes of feedback. The limitation of this study was the small sample size and therefore the results are not generalisable. https://doi.org/10.26803/ijlter.18.11.3

Patient Text Feedback Based Optimized Deep Learned Model to Identify the Impact of Therapy

Text classification or Text mining is a very demanding field because the content created by the user in natural language is not easily understandable. It becomes very important to systematically identify and extract subjective information from user content so that it can be easily understandable. The whole process is done by assigning a particular class to text. In the field of opinion mining, most of the work has been done in common areas like restaurants, electronic goods, movie feedback, etc. and a lot of work needs to be done in the area of healthcare and medical. So, the proposed work has been carried over healthcare. The aim of this study is to classify the text feedback of patient using optimized deep learning model to identify the impact of therapy. In proposed method comparison of CNN with machine learning algorithms has been done, in which, CNN gave better results in terms of accuracy (99.98%), precision (0.981), recall (0.981), mean squared error (0.282). Further, we have implemented the CNN with N-gram technique and found that this method improved the results of CNN based on precision (0.999), recall (0.999), mean squared error (0.001), area under curve (0.998) but accuracy remained the same.

The Use of Web-Based Support Groups versus Usual Quit-Smoking Care for Men and Women 21–59 Years Old: A Protocol for a Randomized Controlled Trial

Background: Existing smoking cessation treatments are challenged by low engagement and high relapse rates, suggesting the need for more innovative, accessible and interactive treatment strategies. Twitter is a web-based platform that allows people to communicate with each other throughout the day right from their phone. Objectives: This study aims to leverage the social media platform of Twitter for fostering peer-to-peer support to decrease relapse with quitting smoking. Further, the study will compare the effects of co-ed versus women-only groups on women’s success with quitting smoking. Methods: The study design is a web-based 3-arm randomized controlled trial with two treatment arms (a co-ed or women-only Twitter support group) and a control arm. Participants are recruited online and are randomized to one of the conditions. All participants receive 8 weeks of combination nicotine replacement therapy (patches plus their choice of gum or lozenges); serial emails with links to Smokefree.gov quit guides; and instructions to record their quit date online (and to quit smoking on that date), on a date falling within a week of study start. Participants randomized to a treatment arm are placed in a fully-automated Twitter support group (co-ed or women-only), paired with a buddy (matched on age, gender, location, and education), and encouraged to communicate with the group and buddy via daily tweeted discussion topics and daily automated feedback texts (a positive tweet if they tweet and an encouraging tweet if they miss tweeting). Recruited online from across the continental United States, the sample consists of 215 men and 745 women current cigarette smokers wanting to quit, between the ages of 21-59. Self-assessed follow-up surveys are completed online at 1, 3, and 6 months past the date they selected to quit smoking with salivary cotinine validation at 3- and 6-months. The primary outcome is sustained biochemically-confirmed abstinence at 6-months follow-up. Results: From November 2016 through September 2018, 960 participants in 36 groups have been recruited for the randomized controlled trial, in addition to 20 participants in an initial pilot group. Data analysis will commence soon for the randomized controlled trial, based on data from 913 of the 960 participants (95.6%) with 47 participants lost to follow-up. Conclusions: This study combines the mobile platform of Twitter with a support group for quitting smoking. Findings will inform the efficacy of virtual peer-to-peer support groups for quitting smoking and potentially elucidate gender differences in quit rates found in prior research. Trial Registration ClinicalTrial.gov Registration # NCT02823028 (Last Updated January 2019).