Results of the summarizing process for the arXiv paper: 2110.02842v1

The recent ongoing coronavirus pandemic has brought attention to the importance of hand hygiene practices in our daily lives. Governments and worldwide health authorities have been promoting good hand hygiene practices as a crucial measure to prevent the spread of infections. In Europe alone, there are over one million cases of hospital-acquired infections annually, highlighting the need for effective hand hygiene compliance. To address this issue, the World Health Organization (WHO) conducted a study on hand hygiene gestures. They recorded and analyzed these gestures using an aluminum frame setup placed at a laboratory sink. The setup included enough space to accommodate utilities such as a soap dispenser and hand towel. It was designed to be lightweight yet sturdy so that it could be easily relocated if necessary. During the study, thirty participants were trained on proper hand hygiene gestures demonstration before their gestures were recorded. The video recordings were then converted into image files and organized into six different hand hygiene classes. These classes included gestures such as Fingers Interlaced, P2PFingers Interlaced, Rotational Rub, Rub hands palm to palm, Fingers Interlocked, and Thumb Rub. To classify these multiclass hand hygiene stages accurately, the Resnet50 framework was selected for training the model. The first set of classes (Fingers Interlaced, P2PFingers Interlaced, and Rotational Rub) were trained for 25 epochs with an accuracy of 44 percent and a loss score greater than 1.5 in the validation set. For the second set of classes (Rub hands palm to palm, Fingers Interlocked, Thumb Rub), training steps were extended to 50 epochs resulting in an accuracy of 72 percent with a loss score less than 0.8 for the validation set. Additionally, green sheets were used to hide background information while a WHO poster displaying hand hygiene gestures helped participants perform them accurately without needing to memorize them. This work represents a preliminary analysis of a robust hand hygiene dataset using transfer learning techniques with an ultimate goal of deploying a real-time prediction system for healthcare workers based on its findings. By promoting good hand hygiene compliance through such systems we can reduce both risk of infections and healthcare expenditures significantly .

• - Coronavirus pandemic highlights importance of hand hygiene practices
• - Governments and health authorities promote good hand hygiene as crucial for preventing infections
• - Over one million cases of hospital-acquired infections in Europe annually
• - World Health Organization (WHO) conducts study on hand hygiene gestures
• - Study records and analyzes gestures using aluminum frame setup at laboratory sink
• - Setup designed to be lightweight yet sturdy for easy relocation if necessary
• - Thirty participants trained on proper hand hygiene gestures before recordings were made
• - Video recordings converted into image files and organized into six different hand hygiene classes
• - Resnet50 framework selected for training the model to accurately classify the classes
• - First set of classes trained for 25 epochs with 44% accuracy and loss score >1.5 in validation set
• - Second set of classes trained for 50 epochs with 72% accuracy and loss score <0.8 in validation set
• - Green sheets used to hide background information during training
• - WHO poster displaying hand hygiene gestures helps participants perform them accurately
• - Goal is to deploy a real-time prediction system for healthcare workers based on findings
• - Promoting good hand hygiene compliance reduces risk of infections and healthcare expenditures significantly

The coronavirus pandemic has shown how important it is to keep our hands clean. Governments and health authorities say that washing our hands well can help stop infections. In Europe, there are more than one million cases of infections that happen in hospitals every year. The World Health Organization (WHO) did a study on how we wash our hands. They used a special setup with metal frames to record and analyze hand gestures at a sink in a lab. They trained 30 people on the right way to wash their hands before recording them. The recordings were turned into pictures and put into different groups based on how we wash our hands. They used a computer program called Resnet50 to learn from these pictures and classify them correctly. After training the program two times, they got better at recognizing the different ways we wash our hands. During training, they used green sheets to hide other things in the background so that the program only focused on the hand gestures. The WHO wants to use what they learned to make a system that can predict in real-time if healthcare workers are washing their hands correctly or not. By promoting good hand hygiene, we can reduce the risk of getting sick and save money on healthcare." Definitions- Coronavirus pandemic: A big sickness that is spreading around the world. - Hand hygiene practices: Washing our hands properly. - Infections: When germs get inside our bodies and make us sick. - Hospital-acquired infections: Infections that people get while staying in a hospital. - World Health Organization

The Importance of Hand Hygiene Practices in the Age of Coronavirus

The recent coronavirus pandemic has brought attention to the importance of hand hygiene practices, with governments and worldwide health authorities promoting good hand hygiene as a crucial measure to prevent the spread of infections. In Europe alone, there are over one million cases of hospital-acquired infections annually, highlighting the need for effective compliance with hand hygiene protocols. To address this issue, the World Health Organization (WHO) conducted a study on hand hygiene gestures.

Study Setup

To record and analyze these gestures, an aluminum frame setup was placed at a laboratory sink. The setup included enough space to accommodate utilities such as a soap dispenser and hand towel, and was designed to be lightweight yet sturdy so that it could be easily relocated if necessary. During the study, thirty participants were trained on proper hand hygiene gestures demonstration before their gestures were recorded via video recordings which were then converted into image files and organized into six different classes: Fingers Interlaced; P2PFingers Interlaced; Rotational Rub; Rub hands palm to palm; Fingers Interlocked; Thumb Rub.

Classification Model

To classify these multiclass hand hygiene stages accurately, the Resnet50 framework was selected for training the model. The first set of classes (Fingers Interlaced, P2PFingers Interlaced, and Rotational Rub) were trained for 25 epochs with an accuracy of 44 percent and a loss score greater than 1.5 in the validation set. For the second set of classes (Rub hands palm to palm, Fingers Interlocked, Thumb Rub), training steps were extended to 50 epochs resulting in an accuracy of 72 percent with a loss score less than 0.8 for the validation set. Additionally green sheets were used to hide background information while a WHO poster displaying hand hygiene gestures helped participants perform them accurately without needing to memorize them .

Conclusion

This work represents a preliminary analysis of a robust dataset using transfer learning techniques with an ultimate goal of deploying real-time prediction systems for healthcare workers based on its findings – thus helping promote good hand hygiene compliance through such systems which can reduce both risk of infections and healthcare expenditures significantly .