By collecting the changes of kinds of vital signs(including heart beat / blood pressure / steps), we combined the results with the testers’ fatigue scores which belonged to the matched fatigue typestheir basic profiles. We extracted 780 types of structural features sets by cleaning raw data, then separating into training set, validation settesting set. With machine learning algorithm to auto-optimize regression analysis, finally we come out the correlative features between vital signsfatigue ,the overwork model as our deliverables.

"Using a machine learning regression model with physiological time series data,  we are able to fit the changes of fatigue score on duty time of test set with R^2 ~ 0.5. Besides, if we set the alarm limit of overwork according to the fatigue score difference between onoff duty, we are able to classify the overwork cases, changes of fatigue score greater than 2 standard deviation, with AUC ~ 0.9 from our best machine learning classification model.
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 datafeatures, 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.

天使投資人、策略合作夥伴