Bimodal data analysis for early detection of lameness in dairy cows using artificial intelligence
Bibliographic record
Abstract
Lameness remains a leading cause of economic loss in Canadian dairy herds while also compromising animal welfare. To address the urgent need for early detection, we introduce a novel bimodal artificial intelligence (AI) framework that leverages both facial biometric data and accelerometer-based movement metrics. Over a 21-day period, six Holstein cows were monitored to capture variations in facial expressions and locomotion, and a multimodal model was built by combining DenseNet-121 for image analysis with Long Short-Term Memory (LSTM) networks for time-series data. Crucially, our model employs a multi-head attention mechanism to fuse visual and movement features, enabling it to overcome confounding factors such as lighting conditions, barn environments, and individual behavioral differences. This approach achieved a 99.55 % accuracy—substantially exceeding single-modality baselines—and Grad-CAM interpretations revealed key facial cues (orbital tightening, ear posture, muzzle tension) linked to lameness. Lame cows also exhibited prolonged resting times, especially during peak activity hours, underscoring their discomfort. These findings illustrate how integrating facial and accelerometer data can promote timely interventions, significantly enhancing cow welfare and reducing medical expenditures and productivity losses. Moreover, our results highlight how tie-stall barn systems can exacerbate lameness by restricting natural movement, further supporting recommendations to transition toward more open, movement-friendly housing. In doing so, producers not only protect cow well-being but also safeguard vital economic returns. • AI integrates visual and accelerometer data for early lameness detection in dairy cows. • Facial biometrics reveal novel features, enhancing accurate lameness identification. • Multimodal AI model improves detection accuracy to 99.55 % for lameness in dairy cows. • Combines DenseNet and LSTM models for comprehensive analysis of cow behavior data. • Offers a non-invasive, real-time solution to enhance cow welfare and farm productivity.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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".