Methods and tools for efficient physiotherapy
Many people suffer from dysfunctions or abnormalities in their gait, due for example, to functional deficits. Large therapeutic facilities and rehabilitation centres usually have a large number of patients to analyze and treat cost-effectively in the shortest possible time. In the project "IntelliGait", St. Pölten University of Applied Sciences developed tools for automated pattern recognition in gait-analysis-data in order to facilitate the work of physiotherapists and support therapy.
Facilities typically use a combination of different analytical methods. For the detection of gait disorders, often ground reaction forces while walking are analyzed using force plates. Therapists and medical personnel usually inspect the variety of resulting biomechanical parameters manually, and from these derive clinical diagnoses upon which they base medical decisions.
New procedures in the clinical field
"In recent years, new approaches for the automatic analysis and classification of gait-analysis- data have been repeatedly presented. They are however, usually based only on artificially created data sets or include only a small number of possible functional deficits. This strongly limits the reliability of such methods", said sports scientist Brian Horsak, UAS lecturer in the Department of Health Sciences at St. Pölten UAS and head of the IntelliGait project.
The IntelliGait project can use the database of the Lower Austrian Rehabilitation Centre, Weißer Hof AUVA (Allgemeine Unfallversicherungsanstalt / General Accident Insurance Institution), which includes gait-analysis data and associated diagnoses of patients from 20 years of clinical practice. With this collection of data, a general model of normal walking patterns is generated, which takes into account different parameters such as walking speed, age and gender. Based on this, an automatic classification of different functional deficits as well as methods are being developed, which by deviation from the norm gait model, match data from the entire database in order to find similar cases and their associated diagnoses. This should assist therapists in making more efficient diagnoses and decisions in daily clinical practice.
The project is carried out in close cooperation with AUVA (Allgemeine Unfallversicherungsanstalt / General Accident Insurance Institution). The University of Vienna is a scientific partner in the area of biomechanics. The study programme, Physiotherapy at St. Pölten UAS brings clinical expertise to the project.
Pattern recognition by machine learning
For data analysis, methods of pattern recognition and machine learning should be developed. "The amount of data that we have available in the IntelliGait project makes it possible to use self-learning techniques such as neural networks: for one thing, to be able to automatically classify gait data, as well as to learn from the data typical movement patterns for different pathologies. In this way our methods can help the experts gain new insights into the data", said Matthias Zeppelzauer from the Institute for Creative\Media/Technologies (IC\M/T) at St. Pölten UAS, who heads project research in the field of pattern recognition.
In addition to the recognition of patterns and abnormalities, visualization of this data is of great importance. "With large volumes of data, analysis and interpretation of the data is often difficult. But with the right approach, information that is hidden in it can be discovered. The decisive factor is the interaction between automatic data analysis by computer and the interpretation by experts by means of interactive visualization", said Wolfgang Aigner, head of the Institute for Creative\ Media/Technologies.
Aigner together with co-workers in the project "KAVA-Time: Knowledge-Assisted Visual Analytics Methods for Time-Oriented Data", developed methods for a better analysis and visual processing of data which are applied in different subject areas. Thereby, good cooperation between human and machine is important. "With visual analytics, one lets computers do what they do best - for example, finding clusters in large amounts of data. A human however, is better at recognizing visual patterns and dealing with uncertainties and contradictions", said Aigner.
When the data from computers is properly processed, information in the confusing collection of data, which is difficult to detect and could be overlooked, can be extracted from the delineated optical patterns of information.
Research field on media technology in healthcare
In the last three years, St. Pölten UAS has set up a centre for applied research for media-supported health care with a focus on the areas of movement and activity: CARMA (Centre for Applied Research in Media Assisted Healthcare for Motion and Activity).
With the CARMA project, interdisciplinary research groups with know-how in the fields of health and technology were established. In the course of the undertaking, alongside the project IntelliGait, further innovative solutions in the health sector emerged with a focus on physiotherapeutic gait analysis.