Worked on classification and regression problems with recurrent neural networks on time series
data of ICU patients and visualizing the patterns in data with sophisticated techniques.
Worked on a natural language processing model to classify patient physician communication and to
improve message triage.
Implemented a novel forecasting framework which utilizes a CNN to extract features from a patient's
brain MRIs which we then fused with patient data and use RNN to track progression.
Showed that the inclusion of these customised/patient-specific features increases the F1-score of
0.4644, with recall at 0.4974 and precision of 0.4355 of forecasting the disease stages.
Developed a Neural Network classifier to identify if a chest tube is present in an X-Ray image and
achieved
an accuracy of 95% trained on 6000+ images, to help radiologists make better decisions which is
waiting on FDA
approval.
Named Entity Recognition (NER) is a process of extracting, disambiguation, and linking an entity
from raw text to insightful and structured knowledge bases. The paper explores the evolution of NER
from a rule based approach, supervised to unsupervised learning approaches.
Medical report generation could be error-prone, long and tedious task for most physicians. To
address this issue we have worked on automatic image captioning for medical images, exclusively
Chest X-Ray images as they have large repository of publicly available dataset with captions and
also datasets which have comprehensive reports.
Attempted to predict the onset of Sepsis in patients using historical patient data. Precisely
we used ICU and timestamp data of over 40,000 patients and used baseline logistic regression and
random forests to predict the onset of Sepsis as a baseline.
Developed a Long Short - Term Memory
(LSTM) model using PyTorch which led to F1 score of 0.82.
Our model could predict the onset of Sepsis 6 hours before its clinical occurrence to assist
patients with high risks of sepsis for early intervension.
SOF
is an Educational Organization popularizing academic competition and assisting development of competitive
spirit among school children.
I was awarded the School Topper in Grade 10.
Software Developer - 2016 to 2017 ; Team Technocrats is
VIT Chennai’s official robotics team.
Technocrats incorporates students from different
branches who share a common passion for robotics and each one showcases their talent and skill in their
particular field.
Participated in the MIT Hacking Medicine - May 2019 and
co-developed OraNet
which is a mobile application which would assist a clinician to screen patients for oral cancer in a quick
and cost
effective way.Participated in the HackHer413 and co-developed an
algorithm to
detect Wild animals in the images and gives information about the animals.
Represented Spain in the United Nations Development
Program and was an honorable mention in a 4 day Model United Nations in Hyderabad in August 2013.
Represented Spain in the United Nations Environmental
Program and was an honorable mention in a 3 day Model United Nations in BITS Goa in February 2015.
Part of an event at DESIRE Society, Hyderabad - Serving children affected with HIV/AIDS.
Volunteer of an event at Sivananda Rehabilation Home, Hyderabad - To serve the needs of people affected by
leprosy.
Volunteered a fund raising event, organized a 5K run in Hyderabad.