A5044442880- Topic Modeling
- Natural Language Processing Techniques
- Multimodal Machine Learning Applications
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Biomedical Text Mining and Ontologies
- Machine Learning in Healthcare
- Authorship Attribution and Profiling
- Generative Adversarial Networks and Image Synthesis
- Speech and dialogue systems
- Image and Object Detection Techniques
- Glaucoma and retinal disorders
- Artificial Intelligence in Healthcare
- Ophthalmology and Eye Disorders
- Cerebral Venous Sinus Thrombosis
- Artificial Intelligence in Healthcare and Education
- Ethics in Clinical Research
- Computational and Text Analysis Methods
- Imbalanced Data Classification Techniques
- Digital Media Forensic Detection
- Semantic Web and Ontologies
- Data Mining Algorithms and Applications
- Handwritten Text Recognition Techniques
Institute of Medical Sciences
2022
Adobe Systems (United States)
2020-2021
Georgia Institute of Technology
2021
Indian Institute of Technology Roorkee
2019
Abstract Objective The objective of this study is to systematically examine the efficacy both proprietary (GPT-3.5, GPT-4) and open-source large language models (LLMs) (LLAMA 7B, 13B, 70B) in context matching patients clinical trials healthcare. Materials methods employs a multifaceted evaluation framework, incorporating extensive automated human-centric assessments along with detailed error analysis for each model, assesses LLMs’ capabilities analyzing patient eligibility against trial’s...
Clinical trial matching is the task of identifying trials for which patients may be eligible. Typically, this labor-intensive and requires detailed verification patient electronic health records (EHRs) against stringent inclusion exclusion criteria clinical trials. This process also results in many missing out on potential therapeutic options. Recent advancements Large Language Models (LLMs) have made automating patient-trial possible, as shown multiple concurrent research studies. However,...
Chart Question Answering (CQA) is the task of answering natural language questions about visualisations in chart image. Recent solutions, inspired by VQA approaches, rely on image-based attention for question/answering while ignoring inherent structure. We propose STL-CQA which improves through sequential elements localization, question encoding and then, a structural transformer-based learning approach. conduct extensive experiments proposing pre-training tasks, methodology also an improved...
Hrituraj Singh, Anshul Nasery, Denil Mehta, Aishwarya Agarwal, Jatin Lamba, Balaji Vasan Srinivasan. Proceedings of the 2021 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies. 2021.
Clinical trial matching is the task of identifying trials for which patients may be potentially eligible. Typically, this labor-intensive and requires detailed verification patient electronic health records (EHRs) against stringent inclusion exclusion criteria clinical trials. This process manual, time-intensive, challenging to scale up, resulting in many missing out on potential therapeutic options. Recent advancements Large Language Models (LLMs) have made automating patient-trial...
We propose a semantically-aware novel paradigm to perform image extrapolation that enables the addition of new object instances. All previous methods are limited in their capability merely extending already existing objects image. However, our proposed approach focuses not only on (i) present but also (ii) adding extended region based context. To this end, for given image, we first obtain an segmentation map using state-of-the-art semantic method. The, thus, obtained is fed into network...
This paper presents a Multisurface proximal SVM (MPSVM) based decision tree system, combined with ensemble methods of Random Forest and Gradient Boosting, for heart disease classification. Decision trees are very popular classification tasks normally use set axis-parallel boundaries to classify the data. The MPSVM trees, used here, can learn any orientation, are, hence, more flexible in learning dataset this study is Cleveland dataset, which 5-class problem, has data 303 patients, each 13...
Semantic parsing is the task of obtaining machine-interpretable representations from natural language text. We consider one such formal representation - First-Order Logic (FOL) and explore capability neural models in English sentences to FOL. model FOL as a sequence mapping where given sentence, it encoded into an intermediate using LSTM followed by decoder which sequentially generates predicates corresponding formula. improve standard encoder-decoder introducing variable alignment mechanism...
While recent advances in language modeling has resulted powerful generation models, their style remains implicitly dependent on the training data and can not emulate a specific target style. Leveraging generative capabilities of transformer-based we present an approach to induce certain target-author attributes by incorporating continuous multi-dimensional lexical preferences author into models. We introduce rewarding strategies reinforcement learning framework that encourages use words...
Retrieving information from EHR systems is essential for answering specific questions about patient journeys and improving the delivery of clinical care. Despite this fact, most still rely on keyword-based searches. With advent generative large language models (LLMs), retrieving can lead to better search summarization capabilities. Such retrievers also feed Retrieval-augmented generation (RAG) pipelines answer any query. However, task real-world data contained within in order solve several...
The recent success of large language models (LLMs) has paved the way for their adoption in high-stakes domain healthcare. Specifically, application LLMs patient-trial matching, which involves assessing patient eligibility against clinical trial's nuanced inclusion and exclusion criteria, shown promise. Recent research that GPT-3.5, a widely recognized LLM developed by OpenAI, can outperform existing methods with minimal 'variable engineering' simply comparing trial information summaries....
Author stylized rewriting is the task of an input text in a particular author's style. Recent works this area have leveraged Transformer-based language models denoising autoencoder setup to generate author without relying on parallel corpus data. However, these approaches are limited by lack explicit control target attributes and being entirely data-driven. In paper, we propose Director-Generator framework rewrite content style, specifically focusing certain attributes. We show that our...
Recently, research efforts have gained pace to cater varied user preferences while generating text summaries. While there been attempts incorporate a few handpicked characteristics such as length or entities, holistic view around these is missing and crucial insights on why certain should be incorporated in specific manner are absent. With this objective, we provide categorization relevant the task of summarization: one, focusing what content needs generated second, stylistic aspects output...
While recent advances in language modeling have resulted powerful generation models, their style remains implicitly dependent on the training data and can not emulate a specific target style. Leveraging generative capabilities of transformer-based we present an approach to induce certain target-author attributes by incorporating continuous multi-dimensional lexical preferences author into models. We introduce rewarding strategies reinforcement learning framework that encourages use words...
Introduction: Causes of abnormal shape eye globe are axial myopia, staphyloma, coloboma, apparent enlargement eye. The presence and types staphyloma determined by the entire in CT/MRI scans. To look for posterior eyeball patients who come brain imaging is useful preventing vision. Aim & Objectives: study various shapes using cross-sectional (CT/MRI).Role (CT/MRI) prevention vision loss. Material And Method: This a observational done department radio-diagnosis, Pacic Institute Medical...
We propose a semantically-aware novel paradigm to perform image extrapolation that enables the addition of new object instances. All previous methods are limited in their capability merely extending already existing objects image. However, our proposed approach focuses not only on (i) present but also (ii) adding extended region based context. To this end, for given image, we first obtain an segmentation map using state-of-the-art semantic method. The, thus, obtained is fed into network...
Author stylized rewriting is the task of an input text in a particular author's style. Recent works this area have leveraged Transformer-based language models denoising autoencoder setup to generate author without relying on parallel corpus data. However, these approaches are limited by lack explicit control target attributes and being entirely data-driven. In paper, we propose Director-Generator framework rewrite content style, specifically focusing certain attributes. We show that our...