Koustav Biswas
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Koustav Biswas has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
Aditya Paul
2 shared publicationsShalini S
Dayananda Sagar Academy of Technology and Management (DSATM)
2 shared publications- A. Mamatha
Dayananda Sagar Academy of Technology and Management
1 shared publication - B. A. Mala
Dayananda Sagar University
1 shared publication Bhagyashri R Hanji
1 shared publicationC Nandini
1 shared publication
Latest Publications
- Open Access
- 1 September 2026
- Published by Elsevier in SSRN Electronic Journal
Medimate: An Intelligent Framework for Peer-to-Peer Exchange of Compositionally Equivalent Medicines for Affordable Healthcare Access
The rising cost of branded pharmaceuticals continues to be a significant barrier to fair healthcare access, especially for economically disadvantaged groups. Even with the availability of generic and equivalent medications, many people either do not know about these options or cannot afford them. To tackle this pressing issue, we present MediMate, a web application designed to connect users through a peer-to-peer marketplace for buying and selling medicines with similar chemical compositions at much lower prices. The platform uses a carefully selected pharmaceutical database that allows for accurate drug matching based on factors like active ingredients, dosage forms, and therapeutic categories. This ensures that users receive safe and effective alternatives to expensive branded medications. One of key innovations is its JavaScript Chatbot Assistant, which is crucial for user engagement and accessibility. The chatbot offers smart suggestions for equivalent drugs, guides users through the platform's features, answers health-related questions, and provides step-by-step help for uploading prescriptions and managing transactions. This interactive assistant also helps users make informed choices by breaking down medical jargon and giving relevant responses based on their needs. By integrating Ajax support with an easy-to-use marketplace, MediMate not only helps lower the financial burden of medications but also tackles bigger issues like pharmaceutical waste, access to drugs, and digital literacy in healthcare. This research examines the system's design, main features, and MediMate's potential to change how underserved communities access and manage essential medications in a digital world.
- 1 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessMedimate: An Intelligent Framework for Peer-to-Peer Exchange of Compositionally Equivalent Medicines for Affordable Healthcare Access
The rising cost of branded pharmaceuticals continues to be a significant barrier to fair healthcare access, especially for economically disadvantaged groups. Even with the availability of generic and equivalent medications, many people either do not know about these options or cannot afford them. To tackle this pressing issue, we present MediMate, a web application designed to connect users through a peer-to-peer marketplace for buying and selling medicines with similar chemical compositions at much lower prices. The platform uses a carefully selected pharmaceutical database that allows for accurate drug matching based on factors like active ingredients, dosage forms, and therapeutic categories. This ensures that users receive safe and effective alternatives to expensive branded medications. One of key innovations is its JavaScript Chatbot Assistant, which is crucial for user engagement and accessibility. The chatbot offers smart suggestions for equivalent drugs, guides users through the platform's features, answers health-related questions, and provides step-by-step help for uploading prescriptions and managing transactions. This interactive assistant also helps users make informed choices by breaking down medical jargon and giving relevant responses based on their needs. By integrating Ajax support with an easy-to-use marketplace, MediMate not only helps lower the financial burden of medications but also tackles bigger issues like pharmaceutical waste, access to drugs, and digital literacy in healthcare. This research examines the system's design, main features, and MediMate's potential to change how underserved communities access and manage essential medications in a digital world.
- Open Access
- 15 August 2026
- Published by Elsevier in SSRN Electronic Journal
A Scalable Food Delivery Management System with Secure Payment, Real-Time Tracking and Synthetic Data Testing
The increasing usage of digital services has led to an increase in the need for efficient and reliable food delivery system on the internet. In this paper, a webbased scalable food delivery management system has been developed using Django framework, along with eSewa payment gateway integration. The system offers user authentication, restaurant registration, menu searching, order management, secure payment online, and real-time delivery status update. The library Faker has been used for creating dummy data for testing. Experimental results have shown that the system is available 98.5% of the time, can support more than 1,000 users simultaneously, orders are processed 40% faster than with conventional methods, and 95% user satisfaction.
- 15 August 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA Scalable Food Delivery Management System with Secure Payment, Real-Time Tracking and Synthetic Data Testing
The increasing usage of digital services has led to an increase in the need for efficient and reliable food delivery system on the internet. In this paper, a webbased scalable food delivery management system has been developed using Django framework, along with eSewa payment gateway integration. The system offers user authentication, restaurant registration, menu searching, order management, secure payment online, and real-time delivery status update. The library Faker has been used for creating dummy data for testing. Experimental results have shown that the system is available 98.5% of the time, can support more than 1,000 users simultaneously, orders are processed 40% faster than with conventional methods, and 95% user satisfaction.
- Open Access
- 16 July 2026
- Published by Elsevier in SSRN Electronic Journal
Hybrid Chess Engine Using Reinforcement Learning and Minimax Algorithm with Alpha-Beta Pruning
The Artificial Intelligence field has recently become a major area of research in playing strategy board games like chess. Intelligent systems have to make complicated tactical and strategic decisions when playing this type of game. Traditional Chess programs typically use Minimax algorithm and Alpha-Beta pruning technique to explore game tree and make move evaluation. Although those techniques prove to be very effective in terms of tactics, they do not have any kind of adaptation mechanisms. On the other hand, Reinforcement Learning allows the system to acquire the right strategy and make predictions by playing the game and being rewarded for success. In this study, a combination of Reinforcement Learning and Minimax algorithm with Alpha-Beta pruning is considered. The Reinforcement Learning technique would be used for making strategic decisions and move prediction, while Minimax search would handle tactical decisions and moves. It is anticipated that this hybrid Chess work shows better performance in terms of gameplay accuracy, speed and effectiveness in comparison with separate methods. The system is evaluated using such measures as win rate, move accuracy, search time, and adaptability. The results would prove that the hybrid technique would show better performance in playing Chess.
- 16 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessHybrid Chess Engine Using Reinforcement Learning and Minimax Algorithm with Alpha-Beta Pruning
The Artificial Intelligence field has recently become a major area of research in playing strategy board games like chess. Intelligent systems have to make complicated tactical and strategic decisions when playing this type of game. Traditional Chess programs typically use Minimax algorithm and Alpha-Beta pruning technique to explore game tree and make move evaluation. Although those techniques prove to be very effective in terms of tactics, they do not have any kind of adaptation mechanisms. On the other hand, Reinforcement Learning allows the system to acquire the right strategy and make predictions by playing the game and being rewarded for success. In this study, a combination of Reinforcement Learning and Minimax algorithm with Alpha-Beta pruning is considered. The Reinforcement Learning technique would be used for making strategic decisions and move prediction, while Minimax search would handle tactical decisions and moves. It is anticipated that this hybrid Chess work shows better performance in terms of gameplay accuracy, speed and effectiveness in comparison with separate methods. The system is evaluated using such measures as win rate, move accuracy, search time, and adaptability. The results would prove that the hybrid technique would show better performance in playing Chess.
- 2 January 2026
- Published by Springer Nature
- p. 231-239
The Need for Sustainable AI-Based Green Technology for Soil Management in Agriculture
- 2 January 2026
- Published by Springer Nature
- p. 231-239
The Need for Sustainable AI-Based Green Technology for Soil Management in Agriculture
- Open Access
- 1 January 2025
- Published by Elsevier in SSRN Electronic Journal
Water Contamination and Groundwater Level Detection using Machine Learning and Internet of Things
Water scarcity and pollution compound world problems that require smart, anticipatory solutions to sustainable water resource management. Conventional water quality monitoring techniques are intrinsically bound by episodic sampling, time-lagged laboratory analysis, and limited coverage, rendering them incapable of reacting to real-time contamination spills. Concurrently, groundwater levels continue declining precipitously under overutilization and climate change, with few forecasting aids to inform effective management. This research offers an end-to-end, two-layered solution that integrates Internet of Things (IoT) technology for real- time water quality monitoring and machine learning (ML) models for forecasting groundwater availability. The system incorporates four sensor types—pH, turbidity, Total Dissolved Solids (TDS), and temperature/humidity (DHT11)—networked through ESP32 microcontrollers, which send continuous data to a centralized cloud interface using the Blynk IoT platform. This real-time data is plotted on an interactive dashboard where users can define thresholds, get notifications, and respond immediately in the event of anomalies. At the same time, historical groundwater data from Kaggle and Data.gov.in are analyzed with ML algorithms, i.e., Linear Regression, Random Forest, and Gradient Boosting. These models are trained and tested to detect temporal patterns and forecast future groundwater availability with high accuracy.
- 1 January 2025
- Published by Elsevier in SSRN Electronic Journal
Open AccessWater Contamination and Groundwater Level Detection using Machine Learning and Internet of Things
Water scarcity and pollution compound world problems that require smart, anticipatory solutions to sustainable water resource management. Conventional water quality monitoring techniques are intrinsically bound by episodic sampling, time-lagged laboratory analysis, and limited coverage, rendering them incapable of reacting to real-time contamination spills. Concurrently, groundwater levels continue declining precipitously under overutilization and climate change, with few forecasting aids to inform effective management. This research offers an end-to-end, two-layered solution that integrates Internet of Things (IoT) technology for real- time water quality monitoring and machine learning (ML) models for forecasting groundwater availability. The system incorporates four sensor types—pH, turbidity, Total Dissolved Solids (TDS), and temperature/humidity (DHT11)—networked through ESP32 microcontrollers, which send continuous data to a centralized cloud interface using the Blynk IoT platform. This real-time data is plotted on an interactive dashboard where users can define thresholds, get notifications, and respond immediately in the event of anomalies. At the same time, historical groundwater data from Kaggle and Data.gov.in are analyzed with ML algorithms, i.e., Linear Regression, Random Forest, and Gradient Boosting. These models are trained and tested to detect temporal patterns and forecast future groundwater availability with high accuracy.
- Open Access
- 1 January 2025
- Published by Elsevier in SSRN Electronic Journal
AI-Enhanced Comprehensive Autism Detection and Support Platform for all ages
With AI support, this research personalizes programs to meet each primary area of development listed for individuals on the Autism Spectrum, such as communication skills, skills for daily living, and adaptive learning through interactive modules. The built-in chatbot allows for conversational support, interaction, and real-time guidance, assisting users and caregivers in navigating the platform and gaining access to resources. In addition, the AI system helps in the early identification of Autism Spectrum Disorder (ASD) by examining patterns in language, interaction methods, and behavioral indicators for possible autism markers. To further improve user assistance, the platform features emotion recognition functions, allowing it to interpret emotional signals and facial expressions to better understand and react to the emotional condition of the user. Sign language interpretation modules are also included, providing an accessible communication tool for non-verbal users or those with speech limitations. The system connects users with local NGOs and support organizations, building a bridge to the local communities and services for therapy, training, and providing autism-friendly, enjoyable games that support cognitive development, socialization, and routine learning in a stress-free, fun environment. Underpinned by data privacy and ethical use of AI, the platform provides an integrated, holistic ecosystem for those with Autism across all ages, supporting skill development, emotional stability, and enhanced independence.
- 1 January 2025
- Published by Elsevier in SSRN Electronic Journal
Open AccessAI-Enhanced Comprehensive Autism Detection and Support Platform for all ages
With AI support, this research personalizes programs to meet each primary area of development listed for individuals on the Autism Spectrum, such as communication skills, skills for daily living, and adaptive learning through interactive modules. The built-in chatbot allows for conversational support, interaction, and real-time guidance, assisting users and caregivers in navigating the platform and gaining access to resources. In addition, the AI system helps in the early identification of Autism Spectrum Disorder (ASD) by examining patterns in language, interaction methods, and behavioral indicators for possible autism markers. To further improve user assistance, the platform features emotion recognition functions, allowing it to interpret emotional signals and facial expressions to better understand and react to the emotional condition of the user. Sign language interpretation modules are also included, providing an accessible communication tool for non-verbal users or those with speech limitations. The system connects users with local NGOs and support organizations, building a bridge to the local communities and services for therapy, training, and providing autism-friendly, enjoyable games that support cognitive development, socialization, and routine learning in a stress-free, fun environment. Underpinned by data privacy and ethical use of AI, the platform provides an integrated, holistic ecosystem for those with Autism across all ages, supporting skill development, emotional stability, and enhanced independence.