What question did this study set out to answer?

To overcome extreme class imbalance in continuous video monitoring and reduce false-positive alarms in automated fall detection systems.

August 16, 2026Scientific ReportsOpen Access

Fall detection in extreme class imbalance: a cascade architecture for real-world deployment

View Full Paper

Ask AI

Bookmark

Authors

EHEthan Henley NSNicole E.-P. StarkVirginia Tech BRBrianna ReillyVirginia Tech

Discussion

Member takes

Implication

Validation study demonstrates a two-stage cascade architecture reduces false-positive fall alarms by 40.5% in continuous elderly care footage, highlighting real-world clinical viability.

Key Points

To overcome extreme class imbalance in continuous video monitoring and reduce false-positive alarms in automated fall detection systems.
Designed a two-stage cascade model comprising a lightweight 3D CNN trained via curriculum learning with adaptive focal loss (Stage 1) and a feature-based arbiter to filter false positives (Stage 2).
Evaluated performance using 190 simulated fall videos from the LE2I dataset and 300 spontaneous falls from elderly residents in long-term care facilities from the Robinovitch dataset.
The cascade architecture achieved 82.4% recall and 32.2% precision, representing a 38% relative precision improvement compared to single-stage CNN baselines (88.5% recall, 23.3% precision).
Achieved a 40.5% reduction in false-positive detections while maintaining overall sensitivity above 80%.

Cite This Study

Henley et al. (2026) studied this question.

synapsesocial.com/papers/6a817a60f2fb91fc834ae2e9 https://doi.org/10.1038/s41598-026-62060-x

View Full Paper

Ask AI

Bookmark

Implication

Validation study demonstrates a two-stage cascade architecture reduces false-positive fall alarms by 40.5% in continuous elderly care footage, highlighting real-world clinical viability.

Fall detection in extreme class imbalance: a cascade architecture for real-world deployment

Discussion

Member takes

Implication

Key Points

Cite This Study

Fall detection in extreme class imbalance: a cascade architecture for real-world deployment

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider