Center for Machine Learning and Intelligent Systems
About  Citation Policy  Donate a Data Set  Contact

Repository Web            Google
View ALL Data Sets

× Check out the beta version of the new UCI Machine Learning Repository we are currently testing! Contact us if you have any issues, questions, or concerns. Click here to try out the new site.

Daphnet Freezing of Gait Data Set
Download: Data Folder, Data Set Description

Abstract: This dataset contains the annotated readings of 3 acceleration sensors at the hip and leg of Parkinson's disease patients that experience freezing of gait (FoG) during walking tasks.

Data Set Characteristics:  

Multivariate, Time-Series

Number of Instances:




Attribute Characteristics:


Number of Attributes:


Date Donated


Associated Tasks:


Missing Values?


Number of Web Hits:



Daniel Roggen, University of Newcastle Upon Tyne, UK, daniel.roggen '@'
Meir Plotnik, Sheba Medical Center, IL, meir.plotnikPeleg '@'
Jeff Hausdorff, Tel Aviv Sourasky Medical Center, jhausdor '@'

This dataset was collected as part of the EU FP6 project Daphnet, grant number 018474-2.
Additional effort to publish this dataset was supported in part by the EU FP7 project CuPiD, grant number 288516.

Data Set Information:

The Daphnet Freezing of Gait Dataset is a dataset devised to benchmark automatic methods to recognize gait freeze from wearable acceleration sensors placed on legs and hip.

The dataset was recorded in the lab with emphasis on generating many freeze events. Users performed there kinds of tasks: straight line walking, walking with numerous turns, and finally a more realistic activity of daily living (ADL) task, where users went into different rooms while fetching coffee, opening doors, etc.

This dataset is the result of a collaboration between the Laboratory for Gait and Neurodynamics, Tel Aviv Sourasky Medical Center, Israel and the Wearable Computing Laboratory, ETH Zurich, Switzerland. Recordings were run at the Tel Aviv Sourasky Medical Center in 2008. The study was approved by the local Human Subjects Review Committee, and was performed in accordance with the ethical standards of the Declaration of Helsinki.

Attribute Information:

Each file comprises the data in a matrix format, with one line per sample, and one column per channel. The channels are as follows:
Time of sample in millisecond
Ankle (shank) acceleration - horizontal forward acceleration [mg]
Ankle (shank) acceleration - vertical [mg]
Ankle (shank) acceleration - horizontal lateral [mg]
Upper leg (thigh) acceleration - horizontal forward acceleration [mg]
Upper leg (thigh) acceleration - vertical [mg]
Upper leg (thigh) acceleration - horizontal lateral [mg]
Trunk acceleration - horizontal forward acceleration [mg]
Trunk acceleration - vertical [mg]
Trunk acceleration - horizontal lateral [mg]
Annotation [0, 1, or 2]

The meaning of the annotations are as follows:
0: not part of the experiment. For instance the sensors are installed on the user or the user is performing activities unrelated to the experimental protocol, such as debriefing
1: experiment, no freeze (can be any of stand, walk, turn)
2: freeze

Relevant Papers:

[1] Marc Bächlin, Meir Plotnik, Daniel Roggen, Nir Giladi, Jeffrey M Hausdorff and Gerhard Tröster, A Wearable System to Assist Walking of Parkinson's Disease Patients.Methods of Information in Medicine, 49:1(88-95), 2010
[2] Meir Plotnik, Marc Bächlin, Inbal Maidan, Daniel Roggen, Gerhard Tröster, Nir Giladi and Jeffrey M Hausdorff, Automated biofeedback assistance for freezing of gait in patients with Parkinson's disease. Proceedings of the International Society for Posture and Gait Research (ISPGR), Bologna, Italy, 2009
[3] Meir Plotnik, Marc Bächlin, Daniel Roggen, Noit Inbar, Inbal Maidan, Talia Herman, Marina Brozgol, Eliya Shaviv, Gerhard Tröster and Jeffrey M Hausdorff, Automated treatment of freezing of gait in Parkinson's disease using a wearable device that automatically detects freezing. Annual meeting of the Israeli Neurological Society, Israel, pages 63, 2009
[4] Marc Bächlin, Daniel Roggen, Meir Plotnik, Jeffrey M Hausdorff, Nir Giladi and Gerhard Tröster, Online Detection of Freezing of Gait in Parkinson's Disease Patients: A Performance Characterization. Proceedings of the 4th International Conference on Body Area Networks, 2009
[5] Marc Bächlin, Meir Plotnik, Daniel Roggen, Noit Inbar, Nir Giladi, Jeffrey M Hausdorff and Gerhard Tröster. Parkinson patients' perspective on context aware wearable technology for auditive assistance. Proceedings of the 3rd International Conference on Pervasive Computing Technologies for Healthcare, 2009
[6] Marc Bächlin, Daniel Roggen, Meir Plotnik, Noit Inbar, Inbal Maidan, Talia Herman, Marina Brozgol, Eliya Shaviv, Nir Giladi, Jeffrey M Hausdorff and Gerhard Tröster,
Potentials of enhanced context awareness in wearable assistants for Parkinson’s disease patients with freezing of gait syndrome. Proceedings of the 13th International Symposium on Wearable Computers (ISWC), pages 123-130, 2009
[7] Sinziana Mazilu, Michael Hardegger, Zack Zhu, Daniel Roggen, Gerhard Tröster, Meir Plotnik, Jeff Hausdorff. Online Detection of Freezing of Gait with Smartphones and Machine Learning Techniques. Proc 6th Int Conf on Pervasive Computing Technologies for Healthcare, 2012

Citation Request:

Use of this dataset in publications must be acknowledged by referencing the following publication:

Marc Bächlin, Meir Plotnik, Daniel Roggen, Inbal Maidan, Jeffrey M. Hausdorff, Nir Giladi, and Gerhard Tröster, Wearable Assistant for Parkinson's Disease Patients With the Freezing of Gait Symptom. IEEE Transactions on Information Technology in Biomedicine, 14(2), March 2010, pages 436-446

This paper describes the dataset in details. It explain the data acquisition protocol, the kind of sensor used and their placement, and the nature of the data acquired. It also provides baseline results for the automated detection of freezing of gait, against which newer methods can be benchmarked. In particular it describes detection sensitivity/specificity for 3 sensor placements and 4 kinds of derived sensor signals, it analyzes detection latency, and provides first insight into user specific v.s. user independent performance.

We also appreciate if you inform us (daniel.roggen '@' of any publication using this dataset for cross-referencing purposes.

Supported By:

 In Collaboration With:

About  ||  Citation Policy  ||  Donation Policy  ||  Contact  ||  CML