Indication of Electromagnetic Field Exposure via RBF-SVM Using Time-Series Features of Zebrafish Locomotion

{
  "publication_year": 2020,
  "study_type": "animal",
  "exposure": {
    "band": "RF",
    "source": "laboratory magnetic exposure",
    "frequency_mhz": 6.78,
    "sar_wkg": null,
    "duration": null
  },
  "population": "Adult zebrafish (Danio rerio)",
  "sample_size": 28,
  "outcomes": [
    "Zebrafish locomotion time-series features (x,y trajectories)",
    "Classification accuracy/precision for indicating EMF exposure condition",
    "Assessment of ambient environmental parameters (AEPs) influence on locomotion"
  ],
  "main_findings": "In 14 adult zebrafish randomized to sham vs real magnetic exposure (6.78 MHz, ~1 A/m), time-series features of locomotion were used to train an RBF-SVM model that reportedly classified exposure conditions with up to 100% accuracy, with several classifier configurations achieving >95% precision. The authors report that ambient environmental parameters had little effect on locomotion based on regression verified using another group of 14 zebrafish.",
  "effect_direction": "mixed",
  "limitations": [
    "Small sample size per exposure condition (7 fish per group)",
    "Primary outcome is model classification performance rather than direct biological/health endpoints",
    "Exposure duration and other dosimetry details (e.g., SAR) not stated in the abstract",
    "Authors describe a speculative mechanism (neural effects) without direct measurement"
  ],
  "evidence_strength": "low",
  "confidence": 0.74,
  "peer_reviewed_likely": "yes",
  "stance": "neutral",
  "stance_confidence": 0.62,
  "summary": "This animal study developed an RBF-SVM classifier using highly comparative time-series analysis features from zebrafish locomotion to indicate whether fish were under sham or real magnetic exposure at 6.78 MHz (~1 A/m). Using 14 zebrafish (7 per group), the model reportedly achieved very high classification performance (up to 100% accuracy; some configurations >95% precision). A separate group of 14 zebrafish was used to quantitatively verify that ambient environmental parameters had little effect on locomotion.",
  "key_points": [
    "The study uses adult zebrafish locomotion trajectories as time-series inputs to classify EMF exposure conditions.",
    "Fourteen zebrafish were randomized into sham versus real exposure groups (7 per group) for a novel tank test.",
    "Exposure was described as a magnetic exposure at 6.78 MHz with field strength about 1 A/m.",
    "Time-series features were computed via HCTSA and selected using mRMR before RBF-SVM training.",
    "Reported classification accuracies reached 100%, and some classifier setups exceeded 95% precision.",
    "Ambient environmental parameters were assessed and reported to have little effect on locomotion using regression in another 14 zebrafish.",
    "The abstract proposes a speculative neural mechanism but does not report direct neural measurements."
  ],
  "categories": [
    "Animal Studies",
    "RF Exposure",
    "Bioeffects",
    "Methods & Measurement",
    "Machine Learning"
  ],
  "tags": [
    "Zebrafish",
    "Locomotion",
    "Time-Series Features",
    "Support Vector Machine",
    "RBF Kernel",
    "HCTSA",
    "mRMR Feature Selection",
    "Novel Tank Test",
    "Magnetic Field Exposure",
    "6.78 MHz",
    "Classification Accuracy",
    "Behavioral Assay"
  ],
  "keywords": [
    "electromagnetic field",
    "EMF exposure",
    "zebrafish",
    "Danio rerio",
    "locomotion",
    "time-series",
    "HCTSA",
    "mRMR",
    "RBF-SVM",
    "6.78 MHz",
    "magnetic exposure"
  ],
  "suggested_hubs": [],
  "social": {
    "tweet": "Zebrafish locomotion time-series features were used to train an RBF-SVM model to indicate sham vs real magnetic exposure (6.78 MHz, ~1 A/m), with reported classification accuracy up to 100% in this small animal study.",
    "facebook": "A Sensors (Basel) study trained an RBF-SVM classifier on zebrafish locomotion time-series features to distinguish sham from real magnetic exposure at 6.78 MHz (~1 A/m), reporting very high classification performance and minimal influence from ambient environmental parameters.",
    "linkedin": "This animal study (Sensors, 2020) applied HCTSA feature extraction and mRMR selection to zebrafish locomotion trajectories and trained an RBF-SVM to indicate EMF exposure condition (sham vs 6.78 MHz magnetic exposure), reporting up to 100% classification accuracy in a small sample."
  }
}