An Approach to Avoid Hypoglycemia: A Model for Mealtime Insulin Dose Calculation for Diabetic People Deepali Javale, Sharmishta Desai U Porto Journal of Engineering, 2025 Background: It has been observed that the number of diabetes patients has steadily increased over the past decades. Particularly, Type 1 diabetics are required to use insulin therapy to maintain their blood glucose levels. Technological advancements in the production of insulin pumps and monitoring devices have aided diabetics in maintaining a healthy lifestyle. A patient with type 1 Diabetes Mellitus (DM) requires insulin therapy with profound roots. Most diabetics require a minimum of two daily insulin injections, with dosage adjustments dependent on self-monitoring of blood glucose levels. Aim: However, it is frequently observed that determining insulin dose is a somewhat perplexing and not always appropriate task. When the incorrect insulin dose is administered, hypoglycemia, which is a blood glucose level below 60 mg/dL, frequently occurs. This condition places the patient at significant risk and should therefore not be ignored. Method: This paper proposes an FCCMID (Food Carbohydrate Count Mealtime Insulin Dose) model for calculating the Mealtime Insulin Dose (MID) based on insulin sensitivity, insulin carbohydrate ratio, and automatic calculation of carbohydrate count based on food intake using the Indian Food Carbohydrate Lookup Table. Conclusion: Target Blood Glucose Monitoring value attained by injecting MID (FCCMID) value of insulin is very close to the desired target Blood Glucose Monitoring Value (assumed to be 100 mg/l), thereby preventing Hypoglycemia. The observed average target blood glucose level is 110 mg/l, which is very near to the optimal 100 mg/l value.
FER-BHARAT: a lightweight deep learning network for efficient unimodal facial emotion recognition in Indian context Ruhina Karani, Jay Jani, Sharmishta Desai Discover Artificial Intelligence, 2024 Humans' ability to manage their emotions has a big impact on their ability to plan and make decisions. In order to better understand people and improve human–machine interaction, researchers in affective computing and artificial intelligence are investigating the detection and recognition of emotions. However, different cultures have distinct ways of expressing emotions, and the existing emotion recognition datasets and models may not effectively capture the nuances of the Indian population. To address this gap, this study proposes custom-built lightweight Convolutional Neural Network (CNN) models that are optimized for accuracy and computational efficiency. These models are trained and evaluated on two Indian emotion datasets: The Indian Spontaneous Expression Dataset (ISED) and the Indian Semi Acted Facial Expression Database (iSAFE). The proposed CNN model with manual feature extraction provides remarkable accuracy improvement of 11.14% for ISED and 4.72% for iSAFE datasets as compared to baseline, while reducing the training time. The proposed model also surpasses the accuracy produced by pre-trained ResNet-50 model by 0.27% ISED and by 0.24% for the iSAFE dataset with significant improvement in training time of approximately 320 s for ISED and 60 s for iSAFE dataset. The suggested lightweight CNN model with manual feature extraction offers the advantage of being computationally efficient and more accurate compared to pre-trained model making it a more practical and efficient solution for emotion recognition among Indians.
Analysis of machine learning models for traffic accidents severity classification Akshat Dawange, Avaneesh Bhoite, Sharmishta Desai International Journal of Modeling Simulation and Scientific Computing, 2024 In the modern world, traffic accidents frequently result in fatalities and serious injuries. The ability of machine learning to foretell the severity of road traffic accidents has shown great promise. The classification of traffic accidents has shown to be a good application for algorithms like random forest. In this paper, performance on a specific dataset has been evaluated using random forest and other models. The dataset used for the analysis came from a publicly accessible source and contained information on several variables like the type of road, the time of day, and the weather. In order to analyze the severity of accidents, a number of algorithms were applied to the dataset, including decision tree, random forest classifier, and logistic regression algorithm. Each model was evaluated on parameters such as model accuracy, precision and recall of the model, and F1 score. The random forest classifier outperformed the other models, achieving an accuracy of 98.48%. The study concludes that machine learning algorithms can accurately predict the severity of road traffic accidents, which could help to reduce the number of accidents and fatalities on the road.
IndEmoVis: A Multimodal Repository for In-Depth Emotion Analysis in Conversational Contexts Ruhina Karani, Dr. Sharmishta Desai Procedia Computer Science, 2024 Emotion recognition holds significant importance in human communication, benefiting various domains like human-computer interaction, affective computing, and social robotics. Recent interest lies in exploiting multimodal data, encompassing audio, visual, and other cues, to enhance emotion recognition systems. However, most available datasets predominantly focus on Western cultures, overlooking the diverse emotional expressions in regions like India. Moreover, existing datasets often neglect complex emotions like sympathy and awe. To address these limitations, we introduce "IndEMoVis," a novel multimodal dataset of Indian emotions. It comprises 122 recorded audio visual responses during conversations between pairs of individuals. The dataset includes 61 participants, consisting of 25 females and 36 males aged 18 to 21, primarily from Maharashtra and Gujarat states in India. It encompasses nine emotions: Neutral, Happiness, Sadness, Surprise, Disgust, Anger, Fear, Awe, and Sympathy. The annotation process involves a three-step procedure, ensuring accurate emotion labeling. Additionally, annotations are provided for intensity and confidence levels. IndEMoVis dataset aims to support the research community in affective computing by improving conversation abilities, analyzing emotional intelligence, and evaluating responses in debates. Its cultural relevance and inclusion of complex emotions offer valuable insights into emotion recognition for diverse contexts.
Artificial Intelligence/Machine Learning driven decision making in business analytics for Financial Sector using Ensemble Machine Learning Techniques Ranjana Agrawal, Sharmishta Desai, Divyam Dholwani, Nikita Kedari, Anurag Banerjee 2024 IEEE 3rd World Conference on Applied Intelligence and Computing Aic 2024, 2024 In today's rapidly evolving financial landscape, banks stand in critical conditions to capitalize on advanced technologies to enhance their competitiveness. By harnessing AI/ML models, finance sectors like banks can optimize customer interactions, streamline operations, and drive down expenses while delivering unparalleled user experiences. These powerful tools enable banks to excel in three critical domains: real-time customer engagement, automated workflows, and data-driven decision-making. Deploying AI/ML solutions at scale offers a distinct advantage over competitors, fostering substantial gains for clients, stakeholders, and the institution alike. Credit card risk assessment, insurance claim prediction, and targeted marketing initiatives are just some examples where AI can deliver exceptional results. Algorithms like K-Nearest Neighbors, Random Forest, Support Vector Machines, and Logistic Regression, in this paper prove effective in classifying decision-making problems within the finance sector with great recall, precision and F1 Score above 0.8 in majority cases. We have well utilized stacking classifier ensemble technique for better results.. A comprehensive approach to AI integration encompasses all aspects of the business, including product development, risk management, compliance, and customer service. By embracing this transformational technology, banks can achieve sustainable growth, foster innovation, and maintain a strong foothold in the ever-changing world of finance. As AI continues to revolutionize the banking industry, forward-thinking organizations will reap the rewards of early adoption, setting new standards for excellence and efficiency.
Artificial Intelligence/Machine Learning Driven Decision making in Business Analytics for Financial Sector using Ensemble Machine Learning Techniques Ranjana Agrawal, Sharmishta Desai, Divyam Dholwani, Nikita Kedari, Anurag Banerjee 2024 IEEE 3rd World Conference on Applied Intelligence and Computing Aic 2024, 2024 In today's rapidly evolving financial landscape, banks stand in critical conditions to capitalize on advanced technologies to enhance their competitiveness. By harnessing AI/ML models, finance sectors like banks can optimize customer interactions, streamline operations, and drive down expenses while delivering unparalleled user experiences. These powerful tools enable banks to excel in three critical domains: real-time customer engagement, automated workflows, and data-driven decision-making. Deploying AI/ML solutions at scale offers a distinct advantage over competitors, fostering substantial gains for clients, stakeholders, and the institution alike. Credit card risk assessment, insurance claim prediction, and targeted marketing initiatives are just some examples where AI can deliver exceptional results. Algorithms like K-Nearest Neighbors, Random Forest, Support Vector Machines, and Logistic Regression, in this paper prove effective in classifying decision-making problems within the finance sector with great recall, precision and F1 Score above 0.8 in majority cases. We have well utilized stacking classifier ensemble technique for better results.. A comprehensive approach to AI integration encompasses all aspects of the business, including product development, risk management, compliance, and customer service. By embracing this transformational technology, banks can achieve sustainable growth, foster innovation, and maintain a strong foothold in the ever-changing world of finance. As AI continues to revolutionize the banking industry, forward-thinking organizations will reap the rewards of early adoption, setting new standards for excellence and efficiency. (Abstract)
DETECTION OF EPILEPTIC SEIZURES USING EEG SIGNALS Journal of Theoretical and Applied Information Technology, 2023
Design and Implementation of Multipurpose Chatbot N Pavitha, Priyanka Bhatele, Sharmishta Desai, Himangi Pande Proceedings 4th International Conference on Smart Systems and Inventive Technology Icssit 2022, 2022
Multimodal Techniques for Emotion Recognition Devangi Agarwal, Sharmishta Desai 2021 International Conference on Computational Intelligence and Computing Applications Iccica 2021, 2021
Models for Hand Gesture Recognition using Deep Learning Manasi Agrawal, Rutuja Ainapure, Shrushti Agrawal, Simran Bhosale, Sharmishta Desai 2020 IEEE 5th International Conference on Computing Communication and Automation Iccca 2020, 2020
Very fast decision tree (VFDT) algorithm on hadoop Sharmishta Desai, Sourav Roy, Brina Patel, Samruddhi Purandare, Minal Kucheria Proceedings 2nd International Conference on Computing Communication Control and Automation Iccubea 2016, 2017
