I developed LLM-enhanced narrative interventions that generated personalized stories to support mental health by adapting to individuals’ specific challenges while embedding evidence-based psychological strategies. These systems moved beyond static content by constructing narratives in real time, allowing support to remain both contextually grounded and structured to promote reflection and cognitive shift. Participants engaged with stories that mirrored their situations and modeled small, actionable ways of navigating difficulties, which improved reflection, reduced negative thought endorsement, and made coping strategies feel more applicable to everyday life. This work was further validated in a large-scale megastudy of single-session interventions published in Nature Human Behaviour, where our AI-enhanced narrative intervention demonstrated strong immediate and sustained improvements in depressive symptoms and perceived agency.

[Paper1] [Paper2] [Web Intervention] [Media]

[Paper3] [Video]

During my internship at Microsoft Research, I investigated how conversational agents can adapt their personality and conversational traits to various scenarios, including customer service, mental health support, scheduling, and information retrieval. To explore user preferences, I developed an interactive tool featuring adjustable sliders, enabling participants to express their desired levels of five key communication traits: formality, personification, empathy, sociability, and humor. This tool was deployed across five key markets—the United States, Australia, Japan, Germany, and India—and engaged approximately 3,000 participants. Our findings highlighted distinct preferences for chatbot communication traits across applications; for instance, participants preferred higher empathy in wellbeing contexts and lower personification in coding scenarios. Similarities were noted between applications such as customer service and scheduling. These insights informed the refinement of personality features in Microsoft Copilot, a tool utilized by over 1 billion users worldwide for diverse applications.

[Paper]

Traditional interventions for academic procrastination often fail to capture the nuanced, individual-specific factors that underlie them. Large language models (LLMs) hold immense potential for addressing this gap by permitting open-ended inputs, including the ability to customize interventions to individuals’ unique needs. However, user expectations and potential limitations of LLMs in this context remain underexplored. To address this, we conducted interviews and focus group discussions with 15 university students and 6 experts, during which a technology probe for generating personalized advice for managing procrastination was presented. Our results highlight the necessity for LLMs to provide structured, deadline-oriented steps and enhanced user support mechanisms. Additionally, our results surface the need for an adaptive approach to questioning based on factors like busyness. These findings offer crucial design implications for the development of LLM-based tools for managing procrastination while cautioning the use of LLMs for therapeutic guidance.

[Paper][Web app] 

[Workshop Paper]

Young adults have high rates of mental health conditions, but most do not want or cannot access formal treatment. We therefore recruited young adults with depression or anxiety symptoms to co-design a digital tool for self-managing their mental health concerns. Through study activities — consisting of an online discussion group and a series of design workshops — participants highlighted the importance of easy-to-use digital tools that allow them to exercise independence in their self-management. They described ways that an automated messaging tool might benefit them by: facilitating experimentation with diverse concepts and experiences; allowing variable depth of engagement based on preferences, availability, and mood; and collecting feedback to personalize the tool. 

Based on our formative work, we developed a range of text message contents that include peer-support messages, reflection questions, and stories on overcoming cognitive distortions. Our system enables us to personalize and contextualize text messages. meeting the unique needs and expectations of people experiencing different problems. Our work has been featured in the media, and multiple papers have been published in top venues like CHI, CSCW, and TOCHI. The resulting two-month intervention, Small Steps SMS program, has been used by more than 5,000 individuals in North America. Shortened versions and modular components of the intervention have been deployed to another 5,000 users.

Access link to Small Steps SMS program

First-Authored Papers: [Paper1][Video1] [Paper2][Video2] [Paper3]

Co-Authored Papers: [Paper1][Paper2][Paper3]

Regular practice of gratitude has the potential to enhance psychological wellbeing and foster stronger social connections among young adults. However, there is a lack of research investigating user needs and expectations regarding gratitude-promoting applications. To address this gap, we employed a user-centered design approach to develop a mobile application that facilitates gratitude practice. Our formative study involved 20 participants who utilized an existing application, providing insights into their preferences for organizing expressions of gratitude and the significance of prompts for reflection and mood labeling after working hours. Building on these findings, we conducted a deployment study with 26 participants using our custom-designed application, which confirmed the positive impact of structured options to guide gratitude practice and highlighted the advantages of passive engagement with the application during busy periods. Our study contributes to the field by identifying key design considerations for promoting gratitude among young adults. One paper from this project has been accepted in CSCW 2024.

[Paper]

Inspired by principles of cognitive behavioral therapy, we designed a 15-minute Reflective Question Activity (RQA) that encourages self-reflection without a human or automated conversational partner. In the first study, 48 participants completed RQA and provided feedback via surveys or interviews. They appreciated the RQA's structure and found venting through writing beneficial, although some reported negative side effects like confusion and frustration. The second study, involving 215 participants from Amazon Mechanical Turk, compared the RQA with a single-question activity. It showed that the RQA led to significantly higher utility ratings and reduced stress, despite requiring more effort and having no notable difference in perceived duration. The third study, deploying the RQA over two weeks with 11 participants, highlighted challenges such as monotony, mobile phone limitations, and the need for timely prompts in new troubling situations.

[Paper]

Most of the HCI work on mental health is based on the Western metaphysical definition of mind that is less applicable outside the West. In this project, we critically examined the functioning of Kaan Pete Roi, a suicide prevention helpline in Bangladesh. We demonstrate how their service often fails in ensuring the callers' safety, providing them long-term support, and protecting the volunteers from harassment and fear. We argue that such failures are often rooted in some foundational ideas of the UK-born `befriending' model that underpins the service. Building on the decolonial philosophy of Enrique Dussel, we further argue that the `befriending' model and its underpinning Western metaphysical ideation of mind carry a colonial impulse, and discuss how community-based approaches may better address the mental health problems in the Global South. In the deployment of a custom reflection application, we uncovered additional challenges specific to the context, such as difficulties in maintaining relationships with in-laws, the pressure of managing extended family dynamics, and stress related to religious adherence.

[Paper1]

[Paper2]

[Paper3]

Machine learning algorithms often require quantitative ratings from users to effectively predict helpful content. When these ratings are unavailable, systems make implicit assumptions or imputations to fill in the missing information; however, users are generally kept unaware of these processes. In our work, we explore ways of informing the users about system imputations, and experiment with imputed ratings and various explanations required by users to correct imputations. We investigate these approaches through the deployment of a text messaging probe to 26 participants to help them manage psychological wellbeing. We provide quantitative results to report users' reactions to correct vs incorrect imputations and potential risks of biasing their ratings. Using semi-structured interviews with participants, we characterize the potential trade-offs regarding user autonomy, and draw insights about alternative ways of involving users in the imputation process. Our findings provide useful directions for future research on communicating system imputation and interpreting user non-responses. 

[Paper]

Women in the global south often seek justice to their online harassment through unveiling the harassers and the screenshots of their sent harassment texts and visual contents before the relevant authorities. Nevertheless, such evidence is often challenged for their authenticity. Our survey (n=91) and interview (n=43) with Bangladeshi online gender harassment victims revealed the depth of the problem, and we set design goals to collect evidence from Facebook Messenger with ensured authenticity. Building on the ‘shame-based model’ of gender justice, we designed ‘Unmochon’, a tool that captures authentic evidence and shares with victims’ intended group. Our user-study (n=48) revealed that diminishing authenticity problem may still leave the victim and online gender justice entangled with mob-sentiment, hegemonic legal consciousness, and several privacy aspects. Our findings open up a new discussion on how HCI-design should address online gender justice in such a complex social setting. 

[Paper][Project] 

**This work was done before starting my PhD.

Background: With the rapid growth rate of newly sequenced genomes, species tree inference from genes sampled throughout the whole genome has become a basic task in comparative and evolutionary biology. However, substantial challenges remain in leveraging these large scale molecular data. One of the foremost challenges is to develop efficient methods that can handle missing data. Popular distance-based methods, such as NJ (neighbor joining) and UPGMA (unweighted pair group method with arithmetic mean) require complete distance matrices without any missing data. 

Results: We introduce two highly accurate machine learning based distance imputation techniques. These methods are based on matrix factorization and autoencoder based deep learning architectures. We evaluated these two methods on a collection of simulated and biological datasets. Experimental results suggest that our proposed methods match or improve upon the best alternate distance imputation techniques. Moreover, these methods are scalable to large datasets with hundreds of taxa, and can handle a substantial amount of missing data. 

Conclusions: This study shows, for the first time, the power and feasibility of applying deep learning techniques for imputing distance matrices. Thus, this study advances the state-of-the-art in phylogenetic tree construction in the presence of missing data. The proposed methods are available in open source form at https://github.com/Ananya-Bhattacharjee/ImputeDistances. 

[Paper] [Project]