Data Structures and Algorithms

2409 Submissions

[3] viXra:2409.0090 [pdf] submitted on 2024-09-16 00:21:44

A Framework for Automated Low Latency Adverse Drug Reaction Reporting Using Crowdsourced Reporting and Graph Theory Analysis

Authors: Jyotirmay Kirtania
Comments: 19 Pages.

Adverse Drug Reactions (ADRs) are a leading cause of hospital admissions and healthcare costs. Traditional methods of ADR reporting often rely on post-marketing surveillance, and manual reporting of ADRs to the local or national pharmacovigilance agencies for causality assessment and final reporting to the WHO. High-income countries have their own national (i.e., USFDA) and regional (i.e., European Medicines Agency / EMA) pharmacovigilance agencies. However, this process is slow and inefficient. This article proposes a novel framework for integrating ADR detection into clinical workflows using Electronic Medical Record (EMR) systems, crowdsourced reporting from patients and healthcare professionals, and graph theory for generating automated ADR signals and reports to the local or national pharmacovigilance agencies. The system leverages automated data collection from EMRs (drug prescriptions, clinical notes) by EMR data scraping, integrating ADR dictionaries and drug databases to automate the generation of ranked ADR signals. By applying graph theory, the system filters and upranks connections between drugs and ADRs, considering the temporal relationship between drug administration and ADR occurrence. This automated approach offers a significant improvement in ADR reporting, enabling faster detection and more accurate predictions. Methodologies, framework visualizations and python code snippets are included to aid implementation.
Category: Data Structures and Algorithms

[2] viXra:2409.0079 [pdf] submitted on 2024-09-15 23:04:23

Optimizing Procurement Efficiency: Neural Networks and Synthetic Data for Predicting Delivery Times

Authors: Oluwashola Aremu, Dan Taiye
Comments: 14 Pages.

This study explores the application of neural networks to predict product delivery times in procurement processes, utilizing a large synthetic dataset. As timely delivery is crucial for supply chain efficiency, accurate prediction of procurement timelines can significantly enhance operational planning and resource allocation. Our research employs a multi-layer neural network model trained on a synthetically generated dataset of 1 million entries. The dataset incorporates key procurement attributes including purchase value, complexity, procurement method, product type, number of potential suppliers, urgency, organizational size, team experience, budget availability, geographical location, season, and industry sector. By using synthetic data, we overcome common limitations in procurement research such as data scarcity and confidentiality issues, while still capturing the complex interrelationships between variables. The neural network model demonstrates promising results in predicting delivery times, outperforming traditional linear regression models. Our findings suggest that certain attributes, such as complexity, procurement method, geographical location and budget availability have a more significant impact on delivery time predictions. The study also highlights the potential of machine learning techniques in procurement analytics and decision support. While based on synthetic data, this research provides a foundation for future studies using real-world procurement data. It also offers insights into the key factors influencing procurement timelines and demonstrates the potential of neural networks in enhancing procurement efficiency.
Category: Data Structures and Algorithms

[1] viXra:2409.0053 [pdf] submitted on 2024-09-10 11:00:32

Theoretical Structures as Networks from the Perspective of Automated Creation and Proving of Theorems

Authors: Taha Sochi
Comments: 8 Pages.

We envisage theoretical structures (especially in pure mathematics and theoretical physics) as networks made of elementary propositions (representing nodes) interconnected through deductive relationships (representing throats). This vision can be exploited as a basis for employing traditional network modeling techniques in the automated search for new theorems as well as for automated proving of proposed theorems and conjectures. This deductive, deterministic and intuitive approach can replace some of the conventional approaches (which are generally more sophisticated and elaborate and hence they are more expensive) in certain areas of automated and assisted theorem proving in addition to its benefit in the automated search for novel theorems. However, we admit that it has a number of limitations and shortcomings although this similarly applies to other methods in this field; moreover some of these limitations and shortcomings can be overcome by the reformulation of certain theoretical structures where we rely for the viability of this reformulation on our perception of theoretical structures as elaborate high-level linguistic systems.
Category: Data Structures and Algorithms