Refine
Document Type
- Article (5)
- Conference Proceeding (1)
- Doctoral Thesis (1)
Language
- English (5)
- German (1)
- Multiple Languages (1)
Has Fulltext
- yes (7)
Is part of the Bibliography
- no (7)
Keywords
Faculty / Organisational entity
#Sport #Gesundheit #Digital
(2021)
In ihrem 50. Jubiläumsjahr lud die Technische Universität Kaiserslautern am 26. und 27. November 2020 zu einem Höhepunkt ein: dem Kongress #Sport #Gesundheit #Digital. Für zwei Tage wurden im Rahmen eines Online-Forums gemeinsam die Themenfelder Sport, Gesundheit und Digitalisierung diskutiert. Wir freuen uns sehr, dass die Techniker Krankenkasse die TUK als Ausrichter der Veranstaltung besonders unterstützt hat. #SGD – Der Kongress setzte an der Schnittstelle von Sport, Gesundheit und Digitalisierung an und beleuchtete Chancen und Möglichkeiten, die durch das Zusammenspiel dieser Disziplinen entstehen können. Gleichzeitig wurden Risiken und Herausforderungen der digitalen Entwicklungen in Sport und Gesundheit betrachtet und perspektivisch mit Blick in die Zukunft analysiert. Hochkarätige Beiträge aus Wissenschaft und Praxis aus allen für das Themenspektrum relevanten Fachrichtungen sorgten für ein hohes Maß an Abwechslung und Transfer. Der Kongress richtete sich dabei nicht nur an Personen aus Wissenschaft und Praxis der Bereiche Gesundheitswesen und -management, Medizin und Psychologie. Ebenso angesprochen wurden Übungsleitende und Angehörige aus Hochschulsport und Sportwissenschaft, Studierende und Mitarbeitende aller bezogenen Fachrichtungen sowie alle allgemein interessierten Personen. Der vorliegende Kongressband stellt die Sammlung der Kongressinhalte dar. Neben den schriftlichen Beiträgen lassen sich hier auch Impressionen der Kongresstage und die Vorträge als interaktiv eingebundene Videos finden.
Study aim: To find out, without relying on gait-specific assumptions or prior knowledge, which parameters are most important for the description of asymmetrical gait in patients after total hip arthroplasty (THA).
Material and methods: The gait of 22 patients after THA was recorded using an optical motion capture system. The waveform data of the marker positions, velocities, and accelerations, as well as joint and segment angles, were used as initial features. The random forest (RF) and minimum-redundancy maximum-relevance (mRMR) algorithms were chosen for feature selection. The results were compared with those obtained from the use of different dimensionality reduction methods.
Results: Hip movement in the sagittal plane, knee kinematics in the frontal and sagittal planes, marker position data of the anterior and posterior superior iliac spine, and acceleration data for markers placed at the proximal end of the fibula are highly important for classification (accuracy: 91.09%). With feature selection, better results were obtained compared to dimensionality reduction.
Conclusion: The proposed approaches can be used to identify and individually address abnormal gait patterns during the rehabilitation process via waveform data. The results indicate that position and acceleration data also provide significant information for this task.
Many machine learning models show black box characteristics and, therefore, a lack of transparency, interpretability, and trustworthiness. This strongly limits their practical application in clinical contexts. For overcoming these limitations, Explainable Artificial Intelligence (XAI) has shown promising results. The current study examined the influence of different input representations on a trained model’s accuracy, interpretability, as well as clinical relevancy using XAI methods. The gait of 27 healthy subjects and 20 subjects after total hip arthroplasty (THA) was recorded with an inertial measurement unit (IMU)-based system. Three different input representations were used for classification. Local Interpretable Model-Agnostic Explanations (LIME) was used for model interpretation. The best accuracy was achieved with automatically extracted features (mean accuracy Macc = 100%), followed by features based on simple descriptive statistics (Macc = 97.38%) and waveform data (Macc = 95.88%). Globally seen, sagittal movement of the hip, knee, and pelvis as well as transversal movement of the ankle were especially important for this specific classification task. The current work shows that the type of input representation crucially determines interpretability as well as clinical relevance. A combined approach using different forms of representations seems advantageous. The results might assist physicians and therapists finding and addressing individual pathologic gait patterns
Die Synopsis setzt sich auseinander mit dem Einsatz von Künstlicher Intelligenz (Maschinelles Lernen) im Kontext biomechanischer Daten. Potentiale der Methoden werden herausgearbeitet und ausgewählte praxisrelevante Limitationen anhand von fünf Publikationen adressiert. Unter anderem können durch Verwendung von Ensemble Feature Selection, Explainable Artificial Intelligence und Metric Learning sowie die Entwicklung eines pathologieunabhängigen Klassifikators vielversprechende Perspektiven aufgezeigt werden.
Clinical classification models are mostly pathology-dependent and, thus, are only able to detect pathologies they have been trained for. Research is needed regarding pathology-independent classifiers and their interpretation. Hence, our aim is to develop a pathology-independent classifier that provides prediction probabilities and explanations of the classification decisions. Spinal posture data of healthy subjects and various pathologies (back pain, spinal fusion, osteoarthritis), as well as synthetic data, were used for modeling. A one-class support vector machine was used as a pathology-independent classifier. The outputs were transformed into a probability distribution according to Platt’s method. Interpretation was performed using the explainable artificial intelligence tool Local Interpretable Model-Agnostic Explanations. The results were compared with those obtained by commonly used binary classification approaches. The best classification results were obtained for subjects with a spinal fusion. Subjects with back pain were especially challenging to distinguish from the healthy reference group. The proposed method proved useful for the interpretation of the predictions. No clear inferiority of the proposed approach compared to commonly used binary classifiers was demonstrated. The application of dynamic spinal data seems important for future works. The proposed approach could be useful to provide an objective orientation and to individually adapt and monitor therapy measures pre- and post-operatively.
The present study aimed to assess the effects of asymmetric muscle fatigue on the skin surface temperature of abdominal and back muscles. The study was based on a pre-post/follow-up design with one group and included a total of 41 subjects (22 male, 19 female; age, 22.63 ± 3.91; weight, 71.89 ± 12.97 kg; height, 173.36 ± 9.95). All the participants were asked to perform side bends in sets of 20 repetitions on a Roman chair until complete exhaustion. The pre-, post- and follow-up test (24 h after) skin surface temperatures were recorded with infrared thermography. Subjective muscle soreness and muscle fatigue were analyzed using two questionnaires. The results of the post hoc tests showed that skin temperature was statistically significantly lower in the post-tests than in the pre- and follow-up tests, but no meaningful differences existed between the pre- and follow-up tests. Asymmetric side differences were found in the post-test for the upper and lower areas of the back. Differences were also noted for the front in both the upper and lower areas. No thermographic side asymmetries were found at the pre- or follow-up measurement for either the back or the front. Our results support the potential of using thermographic skin surface temperature to monitor exercise and recovery in athletes, as well as its use in rehabilitational exercise selection.
Clinical classification models are mostly pathology-dependent and, thus, are only able to
detect pathologies they have been trained for. Research is needed regarding pathology-independent
classifiers and their interpretation. Hence, our aim is to develop a pathology-independent classifier
that provides prediction probabilities and explanations of the classification decisions. Spinal posture
data of healthy subjects and various pathologies (back pain, spinal fusion, osteoarthritis), as
well as synthetic data, were used for modeling. A one-class support vector machine was used as a
pathology-independent classifier. The outputs were transformed into a probability distribution according
to Platt’s method. Interpretation was performed using the explainable artificial intelligence
tool Local Interpretable Model-Agnostic Explanations. The results were compared with those obtained
by commonly used binary classification approaches. The best classification results were obtained
for subjects with a spinal fusion. Subjects with back pain were especially challenging to distinguish
from the healthy reference group. The proposed method proved useful for the interpretation
of the predictions. No clear inferiority of the proposed approach compared to commonly used
binary classifiers was demonstrated. The application of dynamic spinal data seems important for
future works. The proposed approach could be useful to provide an objective orientation and to
individually adapt and monitor therapy measures pre- and post-operatively.