Related Experiment Video
Updated: Oct 14, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
7.0K
Predicting Space Radiation Single Ion Exposure in Rodents: A Machine Learning Approach.
Matthew T Prelich1, Mona Matar1, Suleyman A Gokoglu1
1NASA Glenn Research Center, Cleveland, OH, United States.
Frontiers in Systems Neuroscience
|November 1, 2021
Summary
Machine learning models accurately predict individual space radiation exposure in rats using cognitive tests. This research shows even low doses of Galactic Cosmic Radiation (GCR) can impact rodent cognitive performance.
Area of Science:
- Astrobiology and Space Medicine
- Neuroscience and Cognitive Science
- Computational Biology and Machine Learning
Background:
- Space exploration missions, like those to Mars, expose astronauts to significant levels of Galactic Cosmic Radiation (GCR).
- Understanding the individual impact of GCR exposure on cognitive function is crucial for astronaut health and mission success.
- Existing methods for assessing radiation exposure effects may not capture individual variability.
Purpose of the Study:
- To develop and evaluate machine learning (ML) models for predicting individual GCR ion exposure based on cognitive performance data.
- To explore the correlation between specific GCR ion doses (He, O, Si, Ti, Fe) and cognitive outcomes in rodents.
- To assess the efficacy of ML classifiers in discriminating between different exposure levels.
Main Methods:
- Utilized Attentional Set-shifting (ATSET) experimental tests, a cognitive performance assay, on male Wistar rats.
- Employed machine learning techniques, including Support Vector Machine, Gaussian Naive Bayes, and Random Forest classifiers.
- Trained models using raw cognitive performance data as features and GCR ion doses as targets, exploring various normalization approaches.
Main Results:
- ML models demonstrated the capability to predict individual ion exposure using ATSET scores, with performance exceeding random chance (MCC and F1 scores).
- The study identified a decremental effect on cognitive performance in rodents exposed to ≤150 mGy of single ion GCR.
- Models could effectively discriminate between zero exposure and any GCR exposure level within the cognitive performance feature space.
Conclusions:
- Machine learning provides a viable data-driven approach for predicting individual GCR exposure based on cognitive performance.
- Even low doses of specific GCR ions can measurably impact cognitive function in rodent models.
- The findings highlight the importance of considering individual responses to radiation and suggest potential biomarkers for exposure assessment.
