Data Description

The accelerometer data has been collected using one smartphone carried by subjects, which are caregivers and nurses when they were conducting daily works at a healthcare facility. The smartphone was carried in an arbitrary position such as a pocket. There are a total of 28 activities divided into 4 groups. All the activities are listed in the below table.

Data structure

The data were collected on May and June 2018. This time for the challenge we are providing data of 5 users ( 8, 13, 14, 15, 25). Training and testing care record data separation dates are given in the table below.

Inside the training data folder, we have provided both care records data and accelerometer data.

In the accelerometer data folder, the data is not split. So if you want to utilize it you can match it with your train and test care record file to split them as shown in the tutorial.

If you open the files you can see that the data contains: subject_id(nurse/caregiver), datetime(timestamp), and triaxial sensor data (x,y, and z).

In the care record data folder, we provided the separated care record data as shown in the table.

In the test data folder, we have provided each day’s data for each user. If you open the files you can see that the data contains: id (label id), user_id(nurse/caregiver), activity_type_id(unique id for each activity type), activity_type (activity name), target_id (patients), activity2user_id, start and finish timestamp of the activity and year-month-date-hour timestamp.

If you want you can match it with the accelerometer data that was provided. Participants should note that the start and finish time at the care record file may differ from the datetime at the accelerometer file due to the different time zone settings.

Participants are required to propose their pipelines, predict and submit the activity label for the testing dataset as shown in the tutorial. The test submission folder contains test time stamps that you need to predict for each user. The files contain time stamps, and activity_type_ids and the columns are filled up with zero as shown in Table 3 below.

The participants are required to put 1 in place of 0 when the activity is supposed to happen and regenerate the files. Please make sure to maintain the shape same as the provided files(any changes that occurred in the submission are not the liability of the organizers, please make sure that there is no difference in shape and activity_type_id with the given file)

Data use

All participants may use the data free of charge. Download the dataset

Result Submission

Please submit your results via the submission form. Access the Submission Form here