The application of smart watch monitoring to construct an overwork prediction and alarm model for emergency healthcare professionals
Summary By collecting the changes of kinds of vital signs(including heart beat rate, blood pressure, steps, and etc.), we combined the results with the testers’ fatigue scores which belonged to the matched fatigue types and their basic profiles. We extracted 780 types of structural features sets by cleaning raw data, then separating into training set, validation set and testing set. With machine learning algorithm to auto-optimize regression analysis, finally we come out the correlative features between vital signs and fatigue, and the overwork model as our deliverables.

In terms of scientific breakthroughs, previous studies have not done systematic large-scale and real-time physiological data monitoring on the topic of health care overwork. The results showed that blood pressure (especially the ratio of systolic to diastolic pressure) and heart rate variability (HRV) during the work hours were significantly correlated with the changes of the fatigue scale at the time of start and finish work. A machine learning regression model of all collected physiological data was used to analyze the changes of the fatigue scale at the time of start and finish work, and a correlation coefficient of about 0.7 could be achieved in the test set. This is much higher than the model using only the basic profiles (age, gender, shift schedule, work hours, etc.) of health care workers (correlation coefficient close to 0). Lastly, we try to define overwork as the fatigue value that exceeds two standard deviations of the distribution of all fatigue differences in the time of start and finish work, and classification results using machine learning models can yield an AUC of 0.8 or higher in the test set. If restricted to the group of nurses younger than 35 years old (64% of the total test sample), the classification AUC for overwork over two standard deviations was raised to 0.94 and obtained sensitivity (TPR) of 0.79.

This result shows that the physiological data measured by the transmission device can provide immediate feedback on the physiological status of the health care worker, which can effectively identify the overworked health care workers and facilitate the implementation of preventive protection measures to achieve the effect of preventing the occurrence of danger beforehand. Health care workers can also keep track of their physiological status by viewing real-time recorded physiological data. In addition, the treatment and follow-up measures of overworked health care workers are often done at the time of the tragedy. However, the immediate deployment of medical manpower and the investment of resources in rescuing health care workers is often more costly than early prevention, and the post-event response is doubly costly. By tracking and modeling health care overwork in advance and applying big data analysis and artificial intelligence technology, we can automatically optimize and adapt a common overwork model to provide management units with sufficient information to properly allocate and adapt medical resources, and more effectively prevent and manage health care overwork in advance.

Our purpose is to adopt AI technology to develop medical personnels’ overwork model, which could auto-optimize itself for general application.
The overwork detection model could be adopted not only in hospitals in Taiwan but also in relevant institutes abroad. In different industries, it could be adjusted based on each specific scenario. By adjusting mandatory types of raw data and features, the overwork prediction model could be tuned to fit for specific scenarios. The Ministry of labor mentioned that the average of annual working hours of employees in Taiwan equals 2,028 hours, which is top 4 in global ranking. If the service could be designed as a personalization service, the value of the prediction model could be maximized.
Technical Film
Keyword Medical engineering and medical equipment Biomedical materials and tissue engineering medicine Nano Biomedical Technology cosmetic products
Provide the latest information of AI research centers and applied industries