Saurabh Singh Thakur

Bio: Saurabh Singh Thakur is an academic researcher from Indian Institute of Technology Kharagpur. The author has contributed to research in topics: Instrumentation (computer programming) & Collaborative filtering. The author has an hindex of 5, co-authored 17 publications receiving 58 citations. Previous affiliations of Saurabh Singh Thakur include National Institute of Technology, Jamshedpur & Maharaja Surajmal Institute of Technology.

Artificial-Intelligence-Based Prediction of Clinical Events among Hemodialysis Patients Using Non-Contact Sensor Data.

Abstract: Non-contact sensors are gaining popularity in clinical settings to monitor the vital parameters of patients. In this study, we used a non-contact sensor device to monitor vital parameters like the heart rate, respiration rate, and heart rate variability of hemodialysis (HD) patients for a period of 23 weeks during their HD sessions. During these 23 weeks, a total number of 3237 HD sessions were observed. Out of 109 patients enrolled in the study, 78 patients reported clinical events such as muscle spasms, inpatient stays, emergency visits or even death during the study period. We analyzed the sensor data of these two groups of patients, namely an event and no-event group. We found a statistically significant difference in the heart rates, respiration rates, and some heart rate variability parameters among the two groups of patients when their means were compared using an independent sample t-test. We further developed a supervised machine-learning-based prediction model to predict event or no-event based on the sensor data and demographic information. A mean area under curve (ROC AUC) of 90.16% with 96.21% mean precision, and 88.47% mean recall was achieved. Our findings point towards the novel use of non-contact sensors in clinical settings to monitor the vital parameters of patients and the further development of early warning solutions using artificial intelligence (AI) for the prediction of clinical events. These models could assist healthcare professionals in taking decisions and designing better care plans for patients by early detecting changes to vital parameters.

Chest X-Ray Images Based Automated Detection of Pneumonia Using Transfer Learning and CNN

Abstract: Pneumonia is also known as Bronchopneumonia is a condition of the lungs which primarily affects the air sacks called alveoli. It results in dry and productive cough, fever, weakness, chest pain, and difficulties in breathing. Pneumonia is caused by infections brought upon by microorganisms such as bacteria and viruses. The diagnosis of pneumonia is compassed by scrutinizing X-ray images of the chest. This diagnosis is subjective to factors such as the unclear appearance of the disease in the X-ray image. Hence, artificial intelligence-based systems are required to beacon the clinics. In this study, we have implemented a very popular convolutional neural network model known as VGG16 for the diagnosis of pneumonia. In the training stage, we have used transfer learning and fine-tuning. We were able to achieve an accuracy of 90.54% with a 98.71% recall and 87.69% precision using the aforementioned convolutional neural network model.

An Information Retrieval and Recommendation System for Astronomical Observatories

Abstract: We present a machine-learning-based information retrieval system for astronomical observatories that tries to address user-defined queries related to an instrument. In the modern instrumentation scenario where heterogeneous systems and talents are simultaneously at work, the ability to supply people with the right information helps speed up the tasks for detector operation, maintenance, and upgradation. The proposed method analyzes existing documented efforts at the site to intelligently group related information to a query and to present it online to the user. The user in response can probe the suggested content and explore previously developed solutions or probable ways to address the present situation optimally. We demonstrate natural language-processing-backed knowledge rediscovery by making use of the open source logbook data from the Laser Interferometric Gravitational Observatory (LIGO). We implement and test a web application that incorporates the above idea for LIGO Livingston, LIGO Hanford, and Virgo observatories.

Predicting mental health using smart-phone usage and sensor data

Abstract: The prevalence of mental health problems is rising in the college-going population. To predict the mental health of students using smartphone usage and sensor data is an intriguing research problem. In this study, we aim to engineer feature variables related to daily-living behavior using smartphone usage and sensor data. Further, to develop models using these feature variables to predict if anybody is having a mental health issue or not. Independent-samples t-test has been used to compare the variation in means between the healthy group and group with mental illness. Correlation analysis is used to see the strength of the relationship between the independent and dependent variables. The classification model has been developed to predict mental health, (baseline: n = 45). The difference in means of various feature variables among the two groups is statistically significant (p ≤ 0.05). Many variables are strongly correlated with various mental health predictors. The area under curve of the prediction model for predicting stress is 82.6% and that for the depression is 74%. Our results are quite encouraging and point towards the novel application of smartphone-based data sensing in tracking or predicting mental health issues. The study has some implications for practice such as developing a smartphone-based automated system for predicting mental health that could be a useful tool for professionals in predicting mental health, especially in academic institutions.

Information Retrieval and Recommendation System for Astronomical Observatories

Abstract: We present a machine learning based information retrieval system for astronomical observatories that tries to address user defined queries related to an instrument. In the modern instrumentation scenario where heterogeneous systems and talents are simultaneously at work, the ability to supply with the right information helps speeding up the detector maintenance operations. Enhancing the detector uptime leads to increased coincidence observation and improves the likelihood for the detection of astrophysical signals. Besides, such efforts will efficiently disseminate technical knowledge to a wider audience and will help the ongoing efforts to build upcoming detectors like the LIGO-India etc even at the design phase to foresee possible challenges. The proposed method analyses existing documented efforts at the site to intelligently group together related information to a query and to present it on-line to the user. The user in response can further go into interesting links and find already developed solutions or probable ways to address the present situation optimally. A web application that incorporates the above idea has been implemented and tested for LIGO Livingston, LIGO Hanford and Virgo observatories.

The Observation of Gravitational Waves from a Binary Black Hole Merger

Abstract: On September 14, 2015 at 09:50:45 UTC the two detectors of the Laser Interferometer Gravitational-Wave Observatory simultaneously observed a transient gravitational-wave signal. The signal sweeps upwards in frequency from 35 to 250 Hz with a peak gravitational-wave strain of 1.0×10(-21). It matches the waveform predicted by general relativity for the inspiral and merger of a pair of black holes and the ringdown of the resulting single black hole. The signal was observed with a matched-filter signal-to-noise ratio of 24 and a false alarm rate estimated to be less than 1 event per 203,000 years, equivalent to a significance greater than 5.1σ. The source lies at a luminosity distance of 410(-180)(+160) Mpc corresponding to a redshift z=0.09(-0.04)(+0.03). In the source frame, the initial black hole masses are 36(-4)(+5)M⊙ and 29(-4)(+4)M⊙, and the final black hole mass is 62(-4)(+4)M⊙, with 3.0(-0.5)(+0.5)M⊙c(2) radiated in gravitational waves. All uncertainties define 90% credible intervals. These observations demonstrate the existence of binary stellar-mass black hole systems. This is the first direct detection of gravitational waves and the first observation of a binary black hole merger.

Named Entity Recognition and Normalization Applied to Large-Scale Information Extraction from the Materials Science Literature.

Abstract: The number of published materials science articles has increased manyfold over the past few decades. Now, a major bottleneck in the materials discovery pipeline arises in connecting new results with the previously established literature. A potential solution to this problem is to map the unstructured raw text of published articles onto structured database entries that allow for programmatic querying. To this end, we apply text mining with named entity recognition (NER) for large-scale information extraction from the published materials science literature. The NER model is trained to extract summary-level information from materials science documents, including inorganic material mentions, sample descriptors, phase labels, material properties and applications, as well as any synthesis and characterization methods used. Our classifier achieves an accuracy (f1) of 87%, and is applied to information extraction from 3.27 million materials science abstracts. We extract more than 80 million materials-science-related named entities, and the content of each abstract is represented as a database entry in a structured format. We demonstrate that simple database queries can be used to answer complex "meta-questions" of the published literature that would have previously required laborious, manual literature searches to answer. All of our data and functionality has been made freely available on our Github ( https://github.com/materialsintelligence/matscholar ) and website ( http://matscholar.com ), and we expect these results to accelerate the pace of future materials science discovery.

Enhancing gravitational-wave science with machine learning

Abstract: Machine learning has emerged as a popular and powerful approach for solving problems in astrophysics. We review applications of machine learning techniques for the analysis of ground-based gravitational-wave detector data. Examples include techniques for improving the sensitivity of Advanced LIGO and Advanced Virgo gravitational-wave searches, methods for fast measurements of the astrophysical parameters of gravitational-wave sources, and algorithms for reduction and characterization of non-astrophysical detector noise. These applications demonstrate how machine learning techniques may be harnessed to enhance the science that is possible with current and future gravitational-wave detectors.

Multi-hazard disaster studies: monitoring, detection, recovery, and management, based on emerging technologies and optimal techniques

Abstract: Every year man-made and natural disasters impact the lives of millions of people. The frequency of occurrence of such disasters is steadily increasing since the last 50 years, and this has resulted in considerable loss of life, destruction of infrastructure, and social and economic disruption. A focussed and comprehensive solution is needed encompassing all aspects, including early detection of disaster scenarios, prevention, recovery, and management to minimize the losses. This survey paper presents a critical analysis of the existing methods and technologies that are relevant to a disaster scenario, such as WSN, remote sensing technique, artificial intelligence, IoT, UAV, and satellite imagery, to encounter the issues associated with disaster monitoring, detection, and management. In case of emergency conditions arising out of a typical disaster scenario, there is a strong likelihood that the communication networks will be partially disrupted; thus the alternate networks can play a vital role in disaster detection and management. It focuses on the role of the alternate networks and the associated technologies in maintaining connectivity in various disaster scenarios. It presents a comprehensive study on multiple disasters such as landslide, forest fire, and an earthquake based on the latest technologies to monitor, detect, and manage the various disasters. It focuses on several parameters that are necessary for disaster detection and monitoring and offers appropriate solutions. It also touches upon big data analytics for disaster management. Several techniques are explored, along with their merits and demerits. Open challenges are highlighted, and possible future directions are given.