Using mmWave Radar and Deep Learning to Classify Caregiver Activities for Infection Prevention
Bibliographic record
Abstract
Healthcare-associated infections (HAIs) pose significant risks in clinical environments, and the type of caregivers' activities in patient room play a crucial role in infection transmission. To address this, we present an activity recognition system that uses 3D point cloud data from mmWave radar to classify caregivers' actions and assess their associated infection risk levels while preserving their privacy. Our dataset, collected in a simulated hospital environment, includes 30 distinct caregiver activities categorized into four risk levels based on the type of contact: walking, environmental contact, low-risk patient contact, and high-risk patient contact. We evaluated three deep learning models-PointNet++, Pointformer, and DGCNN-using 5-fold cross-validation to determine the most effective approach for real-world deployment. PointNet++ achieved the best overall performance, with a classification accuracy of 75.16% ± 3.51% and an F1-score of 67.08% ± 2.15% when distinguishing all 30 activities. After grouping activities based on risk levels, performance improved significantly, with 90.84% ± 3.49% accuracy and 88.68% ± 1.85% F1-score. Our results demonstrate the feasibility of non-intrusive, privacy-preserving activity recognition using mmWave radar, enabling automated monitoring systems to enhance infection prevention strategies in clinical settings. Future work will focus on experiments with different validation methods and improving the deep learning models' architectures.Clinical Relevance- This research helps the healthcare facilities such as Infection Prevention and Control (IPAC) experts to understand the events inside the patient rooms while not violating the privacy of patients and staff and predict and prevent possible infection transmission cases.
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) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| 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".