Abstract
Children's unique physiological and psychological characteristics lead to distinct thermal comfort needs in pediatric outpatient departments, highlighting the necessity for developing personal thermal comfort prediction models. However, most studies developed personal thermal comfort models in controlled experimental chambers, with few focusing on pediatric outpatient departments. To fill this gap, this study collected 2908 paired data from a pediatric outpatient department in Nanjing and developed non-intrusive personal thermal comfort prediction models. Firstly, YOLOv8 and face mesh were used to categorize the occupants and identify regions of interest (ROIs) of the body surface. Next, the landmarks of these ROIs were transferred from the visible to infrared images, then the temperature features of the ROIs were computed. Finally, machine learning algorithms were developed to predict the subjects’ subjective thermal responses by temperature features of the ROIs and surrounding operative temperatures (Top). In addition, the importance of input parameters and the application mode of the prediction models were discussed. The results demonstrated that: (i) the developed models outperformed baseline models based solely on the environmental parameters by 4.5 %–17.6 %; (ii) Top was the most important input variable, and outdoor temperature improved model performance by 3.5 %; and (iii) the temperature features of the eyes and the median temperature of ROIs were significant input variables, and incorporating subjects’ body mass index enhanced model performance by 5.6 %. This study provides a basis for developing occupant-centric thermal environment regulation strategies that balance energy efficiency with occupant comfort in pediatric outpatient departments.
| Original language | English |
|---|---|
| Article number | 113452 |
| Journal | Building and Environment |
| Volume | 284 |
| DOIs | |
| Publication status | Published - 1 Oct 2025 |
| Externally published | Yes |
UN SDGs
This output contributes to the following UN Sustainable Development Goals (SDGs)
-
SDG 7 Affordable and Clean Energy
Keywords
- Computer vision
- Machine learning
- Non-intrusive
- Occupant thermal comfort
- Pediatric outpatient department
Fingerprint
Dive into the research topics of 'Non-intrusive personal thermal comfort modeling for occupants in the pediatric outpatient department of a hospital in Nanjing'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver