Mohit Jain

Akshat Sanghvi, Naren Akash, Raza Imam et al.

Large language models (LLMs) are increasingly used for health-related decision support. Yet most evaluations treat diagnosis as a single-shot task with complete information provided upfront, often as a multiple-choice selection. This diverges from clinical practice, where diagnosis is interactive and open-ended, involving sequential hypothesis refinement through targeted questioning. We address this gap. We build MeDxBench, a large-scale benchmark of 4,421 clinical cases across 20 specialties. We further propose MeDxAgent, a multi-agent consultation system for interactive diagnosis, and systematically study its prompt-, flow- and agent-level design choices. MeDxAgent achieves a 10.3% accuracy gain over the baseline on MeDxBench, closing 52.3% of the gap to a full-information oracle. We find that specific design choices: collecting demographics first, passing summarized dialogue for diagnosis, and feeding candidate diagnoses for targeted questioning, improve accuracy, mirroring how physicians reason, though their effect emerges fully only in combination. Code and dataset will be released upon publication.

Yanjie Ze, Nicklas Hansen, Yinbo Chen et al.

A prominent approach to visual Reinforcement Learning (RL) is to learn an internal state representation using self-supervised methods, which has the potential benefit of improved sample-efficiency and generalization through additional learning signal and inductive biases. However, while the real world is inherently 3D, prior efforts have largely been focused on leveraging 2D computer vision techniques as auxiliary self-supervision. In this work, we present a unified framework for self-supervised learning of 3D representations for motor control. Our proposed framework consists of two phases: a pretraining phase where a deep voxel-based 3D autoencoder is pretrained on a large object-centric dataset, and a finetuning phase where the representation is jointly finetuned together with RL on in-domain data. We empirically show that our method enjoys improved sample efficiency in simulated manipulation tasks compared to 2D representation learning methods. Additionally, our learned policies transfer zero-shot to a real robot setup with only approximate geometric correspondence, and successfully solve motor control tasks that involve grasping and lifting from a single, uncalibrated RGB camera. Code and videos are available at https://yanjieze.com/3d4rl/ .

Mohammad Aflah Khan, Neemesh Yadav, Mohit Jain et al.

Hate speech detection is a challenging natural language processing task that requires capturing linguistic and contextual nuances. Pre-trained language models (PLMs) offer rich semantic representations of text that can improve this task. However there is still limited knowledge about ways to effectively combine representations across PLMs and leverage their complementary strengths. In this work, we shed light on various combination techniques for several PLMs and comprehensively analyze their effectiveness. Our findings show that combining embeddings leads to slight improvements but at a high computational cost and the choice of combination has marginal effect on the final outcome. We also make our codebase public at https://github.com/aflah02/The-Art-of-Embedding-Fusion-Optimizing-Hate-Speech-Detection .

Suril Mehta, Nipun Kwatra, Mohit Jain et al.

The use of observed wearable sensor data (e.g., photoplethysmograms [PPG]) to infer health measures (e.g., glucose level or blood pressure) is a very active area of research. Such technology can have a significant impact on health screening, chronic disease management and remote monitoring. A common approach is to collect sensor data and corresponding labels from a clinical grade device (e.g., blood pressure cuff), and train deep learning models to map one to the other. Although well intentioned, this approach often ignores a principled analysis of whether the input sensor data has enough information to predict the desired metric. We analyze the task of predicting blood pressure from PPG pulse wave analysis. Our review of the prior work reveals that many papers fall prey data leakage, and unrealistic constraints on the task and the preprocessing steps. We propose a set of tools to help determine if the input signal in question (e.g., PPG) is indeed a good predictor of the desired label (e.g., blood pressure). Using our proposed tools, we have found that blood pressure prediction using PPG has a high multi-valued mapping factor of 33.2% and low mutual information of 9.8%. In comparison, heart rate prediction using PPG, a well-established task, has a very low multi-valued mapping factor of 0.75% and high mutual information of 87.7%. We argue that these results provide a more realistic representation of the current progress towards to goal of wearable blood pressure measurement via PPG pulse wave analysis.

Ishani Mondal, Kabir Ahuja, Mohit Jain et al.

The COVID-19 pandemic has brought out both the best and worst of language technology (LT). On one hand, conversational agents for information dissemination and basic diagnosis have seen widespread use, and arguably, had an important role in combating the pandemic. On the other hand, it has also become clear that such technologies are readily available for a handful of languages, and the vast majority of the global south is completely bereft of these benefits. What is the state of LT, especially conversational agents, for healthcare across the world's languages? And, what would it take to ensure global readiness of LT before the next pandemic? In this paper, we try to answer these questions through survey of existing literature and resources, as well as through a rapid chatbot building exercise for 15 Asian and African languages with varying amount of resource-availability. The study confirms the pitiful state of LT even for languages with large speaker bases, such as Sinhala and Hausa, and identifies the gaps that could help us prioritize research and investment strategies in LT for healthcare.

Siddhartha Gairola, Pallavi Joshi, Anand Balasubramaniam et al.

Keratoconus is a severe eye disease that leads to deformation of the cornea. It impacts people aged 10-25 years and is the leading cause of blindness in that demography. Corneal topography is the gold standard for keratoconus diagnosis. It is a non-invasive process performed using expensive and bulky medical devices called corneal topographers. This makes it inaccessible to large populations, especially in the Global South. Low-cost smartphone-based corneal topographers, such as SmartKC, have been proposed to make keratoconus diagnosis accessible. Similar to medical-grade topographers, SmartKC outputs curvature heatmaps and quantitative metrics that need to be evaluated by doctors for keratoconus diagnosis. An automatic scheme for evaluation of these heatmaps and quantitative values can play a crucial role in screening keratoconus in areas where doctors are not available. In this work, we propose a dual-head convolutional neural network (CNN) for classifying keratoconus on the heatmaps generated by SmartKC. Since SmartKC is a new device and only had a small dataset (114 samples), we developed a 2-stage transfer learning strategy -- using historical data collected from a medical-grade topographer and a subset of SmartKC data -- to satisfactorily train our network. This, combined with our domain-specific data augmentations, achieved a sensitivity of 91.3% and a specificity of 94.2%.

Aditya Aggarwal, Siddhartha Gairola, Uddeshya Upadhyay et al.

Refractive error is the most common eye disorder and is the key cause behind correctable visual impairment, responsible for nearly 80% of the visual impairment in the US. Refractive error can be diagnosed using multiple methods, including subjective refraction, retinoscopy, and autorefractors. Although subjective refraction is the gold standard, it requires cooperation from the patient and hence is not suitable for infants, young children, and developmentally delayed adults. Retinoscopy is an objective refraction method that does not require any input from the patient. However, retinoscopy requires a lens kit and a trained examiner, which limits its use for mass screening. In this work, we automate retinoscopy by attaching a smartphone to a retinoscope and recording retinoscopic videos with the patient wearing a custom pair of paper frames. We develop a video processing pipeline that takes retinoscopic videos as input and estimates the net refractive error based on our proposed extension of the retinoscopy mathematical model. Our system alleviates the need for a lens kit and can be performed by an untrained examiner. In a clinical trial with 185 eyes, we achieved a sensitivity of 91.0% and specificity of 74.0% on refractive error diagnosis. Moreover, the mean absolute error of our approach was 0.75$\pm$0.67D on net refractive error estimation compared to subjective refraction measurements. Our results indicate that our approach has the potential to be used as a retinoscopy-based refractive error screening tool in real-world medical settings.

Ananditha Raghunath, William Thies, Mohit Jain

When medical chatbots provide advice that conflicts with users' lived care experiences, users are left to interpret, negotiate, and evaluate the legitimacy of that guidance. In India, the widespread overuse of antibiotics, antidiarrheals, and injections has shifted patient expectations away from the guideline-aligned advice that chatbots are trained to provide. We present a mixed-methods, vignette-based study with 200 urban Indian adults examining preferences for and against guideline-aligned, norm-divergent advice in chatbot transcripts. We find that a majority of users reject such advice, drawing on diverse rationales grounded in their lived expectations. Through the design and introduction of context-aware nudges, we support expectation alignment that shifts preferences towards transcripts containing guideline-aligned advice. In doing so, we surface key tensions in the equitable design of medical chatbots in the Global South.

Pragnya Ramjee, Bhuvan Sachdeva, Satvik Golechha et al.

The healthcare landscape is evolving, with patients seeking reliable information about their health conditions and available treatment options. Despite the abundance of information sources, the digital age overwhelms individuals with excess, often inaccurate information. Patients primarily trust medical professionals, highlighting the need for expert-endorsed health information. However, increased patient loads on experts has led to reduced communication time, impacting information sharing. To address this gap, we developed CataractBot. CataractBot answers cataract surgery related questions instantly using an LLM to query a curated knowledge base, and provides expert-verified responses asynchronously. It has multimodal and multilingual capabilities. In an in-the-wild deployment study with 49 patients and attendants, 4 doctors, and 2 patient coordinators, CataractBot demonstrated potential, providing anytime accessibility, saving time, accommodating diverse literacy levels, alleviating power differences, and adding a privacy layer between patients and doctors. Users reported that their trust in the system was established through expert verification. Broadly, our results could inform future work on expert-mediated LLM bots.

Varun Gumma, Ananditha Raghunath, Mohit Jain et al. · microsoft-research

Assessing the capabilities and limitations of large language models (LLMs) has garnered significant interest, yet the evaluation of multiple models in real-world scenarios remains rare. Multilingual evaluation often relies on translated benchmarks, which typically do not capture linguistic and cultural nuances present in the source language. This study provides an extensive assessment of 24 LLMs on real world data collected from Indian patients interacting with a medical chatbot in Indian English and 4 other Indic languages. We employ a uniform Retrieval Augmented Generation framework to generate responses, which are evaluated using both automated techniques and human evaluators on four specific metrics relevant to our application. We find that models vary significantly in their performance and that instruction tuned Indic models do not always perform well on Indic language queries. Further, we empirically show that factual correctness is generally lower for responses to Indic queries compared to English queries. Finally, our qualitative work shows that code-mixed and culturally relevant queries in our dataset pose challenges to evaluated models.

Bhuvan Sachdeva, Naren Akash, Tajamul Ashraf et al.

Cataract surgery is the most common surgical procedure globally, with a disproportionately higher burden in developing countries. While automated surgical video analysis has been explored in general surgery, its application to ophthalmic procedures remains limited. Existing works primarily focus on Phaco cataract surgery, an expensive technique not accessible in regions where cataract treatment is most needed. In contrast, Manual Small-Incision Cataract Surgery (MSICS) is the preferred low-cost, faster alternative in high-volume settings and for challenging cases. However, no dataset exists for MSICS. To address this gap, we introduce Sankara-MSICS, the first comprehensive dataset containing 53 surgical videos annotated for 18 surgical phases and 3,527 frames with 13 surgical tools at the pixel level. We benchmark this dataset on state-of-the-art models and present ToolSeg, a novel framework that enhances tool segmentation by introducing a phase-conditional decoder and a simple yet effective semi-supervised setup leveraging pseudo-labels from foundation models. Our approach significantly improves segmentation performance, achieving a $23.77\%$ to $38.10\%$ increase in mean Dice scores, with a notable boost for tools that are less prevalent and small. Furthermore, we demonstrate that ToolSeg generalizes to other surgical settings, showcasing its effectiveness on the CaDIS dataset.

Bhuvan Sachdeva, Sneha Kumari, Rudransh Agarwal et al.

Cataract surgery is one of the most commonly performed surgeries worldwide, yet intraoperative complications such as iris prolapse, posterior capsule rupture (PCR), and vitreous loss remain major causes of adverse outcomes. Automated detection of such events could enable early warning systems and objective training feedback. In this work, we propose CataractCompDetect, a complication detection framework that combines phase-aware localization, SAM 2-based tracking, complication-specific risk scoring, and vision-language reasoning for final classification. To validate CataractCompDetect, we curate CataComp, the first cataract surgery video dataset annotated for intraoperative complications, comprising 53 surgeries, including 23 with clinical complications. On CataComp, CataractCompDetect achieves an average F1 score of 70.63%, with per-complication performance of 81.8% (Iris Prolapse), 60.87% (PCR), and 69.23% (Vitreous Loss). These results highlight the value of combining structured surgical priors with vision-language reasoning for recognizing rare but high-impact intraoperative events. Our dataset and code will be publicly released upon acceptance.

Harshita Sharma, Maxwell C. Reynolds, Valentina Salvatelli et al.

AI-assisted report generation offers the opportunity to reduce radiologists' workload stemming from expanded screening guidelines, complex cases and workforce shortages, while maintaining diagnostic accuracy. In addition to describing pathological findings in chest X-ray reports, interpreting lines and tubes (L&T) is demanding and repetitive for radiologists, especially with high patient volumes. We introduce MAIRA-X, a clinically evaluated multimodal AI model for longitudinal chest X-ray (CXR) report generation, that encompasses both clinical findings and L&T reporting. Developed using a large-scale, multi-site, longitudinal dataset of 3.1 million studies (comprising 6 million images from 806k patients) from Mayo Clinic, MAIRA-X was evaluated on three holdout datasets and the public MIMIC-CXR dataset, where it significantly improved AI-generated reports over the state of the art on lexical quality, clinical correctness, and L&T-related elements. A novel L&T-specific metrics framework was developed to assess accuracy in reporting attributes such as type, longitudinal change and placement. A first-of-its-kind retrospective user evaluation study was conducted with nine radiologists of varying experience, who blindly reviewed 600 studies from distinct subjects. The user study found comparable rates of critical errors (3.0% for original vs. 4.6% for AI-generated reports) and a similar rate of acceptable sentences (97.8% for original vs. 97.4% for AI-generated reports), marking a significant improvement over prior user studies with larger gaps and higher error rates. Our results suggest that MAIRA-X can effectively assist radiologists, particularly in high-volume clinical settings.

Rishabh Jangir, Nicklas Hansen, Sambaran Ghosal et al.

Learning to solve precision-based manipulation tasks from visual feedback using Reinforcement Learning (RL) could drastically reduce the engineering efforts required by traditional robot systems. However, performing fine-grained motor control from visual inputs alone is challenging, especially with a static third-person camera as often used in previous work. We propose a setting for robotic manipulation in which the agent receives visual feedback from both a third-person camera and an egocentric camera mounted on the robot's wrist. While the third-person camera is static, the egocentric camera enables the robot to actively control its vision to aid in precise manipulation. To fuse visual information from both cameras effectively, we additionally propose to use Transformers with a cross-view attention mechanism that models spatial attention from one view to another (and vice-versa), and use the learned features as input to an RL policy. Our method improves learning over strong single-view and multi-view baselines, and successfully transfers to a set of challenging manipulation tasks on a real robot with uncalibrated cameras, no access to state information, and a high degree of task variability. In a hammer manipulation task, our method succeeds in 75% of trials versus 38% and 13% for multi-view and single-view baselines, respectively.

Siddhartha Gairola, Murtuza Bohra, Nadeem Shaheer et al.

Keratoconus is a severe eye disease affecting the cornea (the clear, dome-shaped outer surface of the eye), causing it to become thin and develop a conical bulge. The diagnosis of keratoconus requires sophisticated ophthalmic devices which are non-portable and very expensive. This makes early detection of keratoconus inaccessible to large populations in low- and middle-income countries, making it a leading cause for partial/complete blindness among such populations. We propose SmartKC, a low-cost, smartphone-based keratoconus diagnosis system comprising of a 3D-printed placido's disc attachment, an LED light strip, and an intelligent smartphone app to capture the reflection of the placido rings on the cornea. An image processing pipeline analyzes the corneal image and uses the smartphone's camera parameters, the placido rings' 3D location, the pixel location of the reflected placido rings and the setup's working distance to construct the corneal surface, via the Arc-Step method and Zernike polynomials based surface fitting. In a clinical study with 101 distinct eyes, we found that SmartKC achieves a sensitivity of 94.1% and a specificity of 100.0%. Moreover, the quantitative curvature estimates (sim-K) strongly correlate with a gold-standard medical device (Pearson correlation coefficient =0.78). Our results indicate that SmartKC has the potential to be used as a keratoconus screening tool under real-world medical settings.

Anusua Trivedi, Mohit Jain, Nikhil Kumar Gupta et al.

Malnutrition is a global health crisis and is the leading cause of death among children under five. Detecting malnutrition requires anthropometric measurements of weight, height, and middle-upper arm circumference. However, measuring them accurately is a challenge, especially in the global south, due to limited resources. In this work, we propose a CNN-based approach to estimate the height of standing children under five years from depth images collected using a smart-phone. According to the SMART Methodology Manual [5], the acceptable accuracy for height is less than 1.4 cm. On training our deep learning model on 87131 depth images, our model achieved an average mean absolute error of 1.64% on 57064 test images. For 70.3% test images, we estimated height accurately within the acceptable 1.4 cm range. Thus, our proposed solution can accurately detect stunting (low height-for-age) in standing children below five years of age.

Minesh Mathew, Mohit Jain, CV Jawahar

Inspired by the success of Deep Learning based approaches to English scene text recognition, we pose and benchmark scene text recognition for three Indic scripts - Devanagari, Telugu and Malayalam. Synthetic word images rendered from Unicode fonts are used for training the recognition system. And the performance is bench-marked on a new IIIT-ILST dataset comprising of hundreds of real scene images containing text in the above mentioned scripts. We use a segmentation free, hybrid but end-to-end trainable CNN-RNN deep neural network for transcribing the word images to the corresponding texts. The cropped word images need not be segmented into the sub-word units and the error is calculated and backpropagated for the the given word image at once. The network is trained using CTC loss, which is proven quite effective for sequence-to-sequence transcription tasks. The CNN layers in the network learn to extract robust feature representations from word images. The sequence of features learnt by the convolutional block is transcribed to a sequence of labels by the RNN+CTC block. The transcription is not bound by word length or a lexicon and is ideal for Indian languages which are highly inflectional. IIIT-ILST dataset, synthetic word images dataset and the script used to render synthetic images are available at http://cvit.iiit.ac.in/research/projects/cvit-projects/iiit-ilst

Siddhartha Gairola, Francis Tom, Nipun Kwatra et al.

Auscultation of respiratory sounds is the primary tool for screening and diagnosing lung diseases. Automated analysis, coupled with digital stethoscopes, can play a crucial role in enabling tele-screening of fatal lung diseases. Deep neural networks (DNNs) have shown a lot of promise for such problems, and are an obvious choice. However, DNNs are extremely data hungry, and the largest respiratory dataset ICBHI has only 6898 breathing cycles, which is still small for training a satisfactory DNN model. In this work, RespireNet, we propose a simple CNN-based model, along with a suite of novel techniques -- device specific fine-tuning, concatenation-based augmentation, blank region clipping, and smart padding -- enabling us to efficiently use the small-sized dataset. We perform extensive evaluation on the ICBHI dataset, and improve upon the state-of-the-art results for 4-class classification by 2.2%

Karan Ahuja, Abhishek Bose, Mohit Jain et al.

Autism often remains undiagnosed in adolescents and adults. Prior research has indicated that an autistic individual often shows atypical fixation and gaze patterns. In this short paper, we demonstrate that by monitoring a user's gaze as they watch commonplace (i.e., not specialized, structured or coded) video, we can identify individuals with autism spectrum disorder. We recruited 35 autistic and 25 non-autistic individuals, and captured their gaze using an off-the-shelf eye tracker connected to a laptop. Within 15 seconds, our approach was 92.5% accurate at identifying individuals with an autism diagnosis. We envision such automatic detection being applied during e.g., the consumption of web media, which could allow for passive screening and adaptation of user interfaces.

Shanu Kumar, Shubham Atreja, Anjali Singh et al.

Citizen engagement and technology usage are two emerging trends driven by smart city initiatives. Governments around the world are adopting technology for faster resolution of civic issues. Typically, citizens report issues, such as broken roads, garbage dumps, etc. through web portals and mobile apps, in order for the government authorities to take appropriate actions. Several mediums -- text, image, audio, video -- are used to report these issues. Through a user study with 13 citizens and 3 authorities, we found that image is the most preferred medium to report civic issues. However, analyzing civic issue related images is challenging for the authorities as it requires manual effort. Moreover, previous works have been limited to identifying a specific set of issues from images. In this work, given an image, we propose to generate a Civic Issue Graph consisting of a set of objects and the semantic relations between them, which are representative of the underlying civic issue. We also release two multi-modal (text and images) datasets, that can help in further analysis of civic issues from images. We present a novel approach for adversarial training of existing scene graph models that enables the use of scene graphs for new applications in the absence of any labelled training data. We conduct several experiments to analyze the efficacy of our approach, and using human evaluation, we establish the appropriateness of our model at representing different civic issues.

Mohit Jain, Minesh Mathew, C. V. Jawahar

Building robust recognizers for Arabic has always been challenging. We demonstrate the effectiveness of an end-to-end trainable CNN-RNN hybrid architecture in recognizing Arabic text in videos and natural scenes. We outperform previous state-of-the-art on two publicly available video text datasets - ALIF and ACTIV. For the scene text recognition task, we introduce a new Arabic scene text dataset and establish baseline results. For scripts like Arabic, a major challenge in developing robust recognizers is the lack of large quantity of annotated data. We overcome this by synthesising millions of Arabic text images from a large vocabulary of Arabic words and phrases. Our implementation is built on top of the model introduced here [37] which is proven quite effective for English scene text recognition. The model follows a segmentation-free, sequence to sequence transcription approach. The network transcribes a sequence of convolutional features from the input image to a sequence of target labels. This does away with the need for segmenting input image into constituent characters/glyphs, which is often difficult for Arabic script. Further, the ability of RNNs to model contextual dependencies yields superior recognition results.