Heatstroke Prevention Challenge

Heatstroke prevention is crucial because heatstroke is a life-threatening condition that can result in severe health consequences and even death. Heatstroke occurs when the body's internal temperature rises to a dangerous level, usually as a result of prolonged exposure to high temperatures or physical exertion in hot weather. Symptoms of heatstroke include confusion, rapid heartbeat, rapid breathing, seizures, and loss of consciousness. Furthermore, the mortality rate due to overheating is estimated to increase by 260% by the 2050s. So, it is important to understand heatstroke symptoms in advance.

Thus, forecasting the danger situation for heatstroke using physiological data and machine learning will be helpful because it can help identify individuals who are at risk for heatstroke before they develop symptoms. By analyzing physiological data such as heart rate, body temperature, and blood pressure, machine learning algorithms can detect patterns and trends that may indicate an increased risk of heatstroke.

This proactive approach to heatstroke prevention can enable timely interventions and treatments to be implemented, potentially preventing more severe health complications. For instance, if an individual's physiological data indicates that they are at risk for heatstroke, they can be advised to seek cooler environments, rest, or hydrate to mitigate the risk of developing heatstroke.

Challenge Goal

The goal of the Heatstroke Prevention Challenge is to forecast the personal thermal comfort sensations on the different days given past data. The training dataset comprises timestamped observations for 27 individuals over a span of 6 days. Each observation consists of a specific feature value at a particular time. Your goal is to create a machine-learning model that can forecast the thermal sensations for the different 2 days of unknown 9 individuals' thermal sensations based on this historical data. Participants should suggest ways for selecting the best features from these data, or they can suggest how different activities/ gender/age-based models can help to improve forecasting labels (optional). You can use 6 days of train data to build your model, and test data will consist of 2 days of data.

• Download the dataset

• Registration is now open

Published papers about the dataset

• Tahera Hossain, Yusuke Kawasaki, Kazuki Honda, Kizito Nkurikiyeyezu, Guillaume Lopez, ‘Toward Human Thermal Comfort Sensing: New Dataset and Analysis of Heart Rate Variability (HRV) Under Different Activities’, The 4th International Conference on Activity and Behavior Computing (ABC 2022), pp. 27 pages, Taylor & Francis, October, 2022. Click HERE for copy.

• Tahera Hossain, Yusuke Kawasaki, Kizito Nkurikiyeyezu, Guillaume Lopez, ‘Toward the Prediction of Environmental Thermal Comfort Sensation using Wearables’, 18th International Conference on Intelligent Environments (IE2022) (Workshop), pp. 10 pages, June, 2022.

• Tahera Hossain, Yusuke Kawasaki, Kazuki Honda, Anna Yokokubo, Guillaume Lopez, ‘Heat Comfort Prediction Using Pulse Variation’, DICOMO2022 Symposium, July, 2022.