From Theory to Observational Results: Baryon Acoustic Oscillations Detected at High Redshift Ranges

Overview

The current mission from several space science institutions occurring on a global scale is the search for baryon acoustic oscillation (BAO) peaks. In our work, we were able to successfully discover a BAO peak detected by the Dark Energy Spectroscopic Instrument (DESI) Early Data Release (EDR) at a higher redshift range than the current published work. In our study, we chose a selection of 60,431 quasars between redshifts ranges from 2 < z < 3. We followed the methods used by current scientists which was to compute the correlation function, revealing a peak at roughly 110 h-1 Mpc at a significance value of 4 ⋅ 10-73 compared to a correlation function associated with a pure CDM model. The detection of the BAO peak in the DESI EDR is significant as it shows the potential of DESI as a powerful tool to determine the BAO standard ruler across a wide range of redshifts, with the 2025 official DESI data release providing even more data to conduct in -depth studies on BAOs . Based on this result, future analyses can help refine the position of the BAO peak at high redshifts and uncover new values for cosmological parameters within the early universe. Virtual OsteoNexus: Attention-Driven Neural Networks for Osteoporosis Detection Saanvi Chakraborty Mason Classical Academy, Naples, FL Often doctors need multiple medical tests to confirm if a patient has osteoporosis or not and this takes a lot of time leading to the disease -causing severe damage. This research focuses on developing a deep learning model for detecting osteoporosis in kne e X -rays by combining attention mechanisms, LSTM networks, and autoencoders. The goal is to improve the accuracy and efficiency of osteoporosis detection, utilizing both spatial and temporal dependencies within the image data. The procedure involves prepro cessing knee X-ray images and training a neural network that incorporates an attention mechanism to highlight key features, an LSTM layer to capture temporal patterns, and an autoencoder for unsupervised feature extraction. The model is trained using labeled knee X-ray images, and its performance is evaluated on a separate test set. Results show the model's ability to classify knee X -rays into two categories: healthy and osteoporosis affected. The model achieved a test accuracy of 90%, and the confusion mat rix analysis confirmed its classification effectiveness. These findings suggest that this model could be applied in clinical settings for automated osteoporosis detection, offering potential benefits for early diagnosis and reducing the impact of human err or on bone damages. By automating the detection process, it reduced the time and effort required for manual analysis, enabling faster diagnosis and better patient outcomes. The system could be integrated into existing healthcare infrastructure, supporting radiologists to improve patient care.

Competition history

• JSHS 2025 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.