Integrating artificial intelligence-driven wound detection into simulation-based forensic medicine training

Abstract

Forensic wound detection and classification is a fundamental skill in forensic medicine education, enabling medical students to interpret injury mechanisms and provide medico-legal opinions. Traditional teaching is often limited by insufficient exposure to real forensic cases and lack of interactive, feedback-driven learning opportunities. This study aimed to develop, validate, and implement a real-time, Artificial Intelligence (AI)-assisted wound detection and classification system using You Only Look Once version 8 nano (YOLOv8n) and deploy it as a web-based application to enhance simulation-based forensic teaching for medical students. A custom dataset comprising 595 images of eight wound types (gsw_entrance, gsw_exit, wound_burn, wound_hanging, wound_hesitation, wound_laceration, wound_open_fracture, wound_strangulation) was collected from physical wound simulations and augmented using Roboflow. Images were annotated in YOLOv8 format and split into training (70%), validation (20%), and test (10%) sets. The YOLOv8n model was trained for 100 epochs on Google Colab with an NVIDIA T4 Graphics Processing Unit (GPU). Model performance was evaluated using mean average precision at IoU 0.50 (mAP@50), mAP@50–95, precision, recall, F1-score, and confusion matrix analysis. The best-performing model (best.pt, 6 MB) was deployed via a Streamlit-based web application, enabling image upload and real-time camera detection. The YOLOv8n model achieved high detection performance, with overall mAP@50 = 0.99, mAP@50–95 = 0.64, precision = 0.98, and recall = 0.99 on the test set (n = 24 images, 27 instances). Per-class F1-scores exceeded 0.99 for most wound types, with highest performance observed for gunshot exit and hanging (1.00). Real-time application testing demonstrated smooth inference on CPU-only devices, enabling immediate visual feedback during simulation sessions. This study demonstrates the feasibility and educational value of integrating Artificial Intelligence-based wound detection and classification into simulation-based teaching. The deployed web application supports cross-platform access without installation requirements, making it a practical and scalable tool for medical education. Future work will focus on expanding the dataset, improving generalizability across wound severities, and conducting formal evaluations of student learning outcomes to quantify the pedagogical impact of this approach.

References

V.A. Cărcăle, “The future of artificial intelligence applications in forensics,” rais.education, 2025. [Online]. Available: https://rais.education/wp-content/ uploads/2025/05/0523.pdf

Prithwish and R. Pinki, “A lightweight end-to-end system for wound tissue analysis,” Sci. Direct, 2025. [Online]. Available: https://www.sciencedirect.com/ science/article/abs/pii/S1746809425002459

Hajare and M. Thalor, “AI-based crime scene simulation through 3D image processing and semantic segmentation,” IJSRET, vol. 11, no. 2, pp. 1576–1580, Mar.–Apr. 2025. [Online]. Available: https://ijsret.com/ wp-content/uploads/2025/03/IJSRET_V11_issue2_552.pdf

Jocher et al., “Explore Ultralytics YOLOv8,” 2025. [Online]. Available: https://docs.ultralytics.com/ models/yolov8/

Vazquez, J. A. Nuñez, X. Fu, P. Gu, and B. Fu, “Exploring transfer learning for deep learning polyp detection in colonoscopy images using YOLOv8,” arXiv preprint arXiv:2502.00133, Jan. 2023. [Online]. Available: https://arxiv.org/html/2502.00133v1

He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2016, pp. 770–778. [Online]. Available: https://arxiv. org/abs/1512.03385

Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, and A. Rabinovich, “Going deeper with convolutions,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2015, pp. 1–9. [Online]. Available: https:// arxiv.org/abs/1409.4842

Ma, X. Zhang, H.-T. Zheng, and J. Sun, “ShuffleNet V2: Practical guidelines for efficient CNN architecture design,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2018, pp. 116–131. [Online]. Available: https://arxiv.org/abs/1807.11164New

Albuquerque, D. M. de Oliveira, A. Oliveira, and G. Rodrigues, “Deep learning-based object detection algorithms in medical imaging: A review,” Heliyon, vol. 10, no. 9, e27379, 2024. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S240584402417168X

Smith, “Simulation-based education in the artificial intelligence era,” Cureus, vol. 15, 2023. [Online]. Available: https://www.cureus.com/articles/161951 -simulation-based-education-in-the-artificial-intelli gence-era

Topol, “Applications of artificial intelligence in medical education: a systematic review,” Cureus, 2025. [Online]. Available: https://www.cureus.com/ articles/334959-applications-of-artificial-intelli gence-in-medical-education-a-systematic-review

Hayden, B. Travis, and S. Benjamin K., “Ethics of artificial intelligence-assisted image interpretation in dermatology and radiology,” NCBI, 2025. [Online]. Available: https://pmc.ncbi.nlm.nih.gov/articles/ PMC11889981/

Vidit, M. Engilberge, and M. Salzmann, “A single domain generalization approach for object detection,” in Proc. CVPR, 2023. [Online]. Available: https://openaccess.thecvf.com/content/CVPR2023/papers/Vidit_CLIP_the_Gap_A_Single_Domain_Generali zation_Approach_for_Object_CVPR_2023_paper.pdf

D. A. Budiman et al., “Effectiveness of AI-driven assessments in enhancing learning outcomes,” Int. J. Inform. Educ. Technol., vol. 15, no. 7, 2025. [Online]. Available: https://www.ijiet.org/vol15/ IJIET-V15N7-2342.pdf

Lattas, C. Davis, C. Creamer, and G. Jefferies, “The use of simulation in forensic social work education: a scoping review,” J. Soc. Work Educ., 2024. [Online]. Available: https://journals.sagepub. com/doi/abs/10.1177/14999013241301097

Hu, Z. Li, J. Yu, X. Wan, H. Tan, and Z. Lin, “Efficient lightweight YOLO: improving small object detection for robustness under real-world conditions,” Cureus, 2023. [Online]. Available: https://pmc.ncbi.nlm.nih.gov/articles/PMC10385816/