Enhancing Automated Patient Detection and Tracking in Epilepsy Monitoring Unit Videos
Bibliographic record
Abstract
Continuous patient monitoring using automated algorithms has the potential to reduce risks associated with seizures by accelerating interventions. Video monitoring offers a practical alternative; however, seizure detection algorithms could learn to associate incidental events with seizure activity, introducing biases during both the training and the testing phases. By incorporating patient tracking into seizure detection algorithms, patients can be isolated within video frames, reducing biases and improving specificity. As no existing object detection models are pre-trained to identify patients as unique class of individuals, a YOLOv8 model was fine-tuned using 1,590 manually labeled frames from 131 patients admitted to our epilepsy monitoring unit, capturing daily activities, seizures, and body occlusions in both daytime and nighttime. Hyperparameter optimization was performed using a grid-search with a five-fold cross-validation on the training and validation set (85% of the data), while 15% was reserved as a held-out test set. Data were split on a patient-wise basis to prevent data leakage across patients. Additionally, a novel post-processing pipeline for videos was developed to further improve performances. The fine-tuned model achieved 97.6% recall and 96.8% precision, detecting patients under diverse conditions. The post-processing pipeline was evaluated on four new seizure videos, correcting 577 out of 580 misclassified or missed detections in these videos. Combined with the novel post-processing step, the proposed pipeline offers a promising tool for enhancing video-based seizure detection.Clinical relevance- This approach reduces misattributions in video-based seizure detection, leading to more reliable, biasfree monitoring. Thus, it enables faster and more accurate interventions, improving patient safety.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".