Conference Agenda

Autism Spectrum Disorder (ASD) presents neurodevelopmental challenges in social interaction, communication, and restricted or repetitive patterns of behaviour. Common conditions of ASD may include difficulties in understanding social cues and engaging in conversations. With the spread of virtual and augmented reality technologies, there has been growing interest to develop game-based learning approaches for helping patients and reducing the impact of ASD. Thus, a virtual environment has been developed to be used in a system dedicated for the emotional training of patients with ASD to improve social interaction and develop appropriate emotional responses to real-life scenarios.

Assessment of driver stress, crucial for road safety, can greatly benefit from the analysis of multimodal physiological signals. However, fusing such heterogeneous data poses significant challenges, particularly in intermediate fusion where noise can also be fused. In this study, we address this challenge by exploring a 1D convolutional neural network (CNN) with self-attention mechanisms on multimodal data. Electrocardiogram (ECG) signals (256 Hz) and respiration (RESP) signals (128 Hz) were obtained from ten subjects using textile electrodes while driving in different scenarios, namely normal driving and phone usage (calling). The obtained multimodal data is preprocessed and then applied to a self-attention mechanism (SAM) CNN (SAMcNN) to identify driver stress. Experiments are validated using Leave-one-out-subject cross validation. The proposed approach is capable of classifying driver stress. It is observed that shorter segments yield an accuracy of 64.16$\%$ compared to longer segment lengths. Thus, exploring self-attention mechanisms for multimodal signals using wearable shirts facilitates non-intrusive monitoring in real-world driving scenarios

Introduction

Point-of-care-ultrasound (POCUS) has been increasingly used in emergency departments and has shown to improve pa-tient care. Its use in prehospital emergency medicine is still uncommon and has only been used in isolated cases. However early detection of right heart strain can provide important information on the patient's condition.

Methods

A randomized, controlled, single-blind simulation study was conducted. It was approved by the Ethics-Committee of the Baden-Württemberg-Medical-Association (reference: F-2023-111). The primary endpoint was quality of the ultrasound examination and assessability of images. Subjects were 22 paramedics, randomized into two groups. The intervention group received basic training (4 hours) and an additional course by a DEGUM instructor (2 hours). The control group only got basic training. Both groups performed an eFAST-examination with telemedical support. The maximum time was limited to 10 minutes. Data were documented using a questionnaire. Theoretical knowledge was assessed on basis of a pre-/post-tests.

Results

The ultrasounds of the intervention group were rated as good (79.31%), the control group as satisfactory (71.03%). The average time to complete the test was 6:35 minutes in the intervention group and 7:30 minutes in the control group. Post-tests from both groups were also rated as good, with the intervention group answering 82.67% and the control group 76.33% of the questions correctly. Both groups were able to improve their knowledge (intervention group + 24.33%, control group + 17.0%).

Conclusion

Both groups delivered sufficient results to argue the use of mobile ultrasound devices in practice. The differences in quality between the two groups were small and do not justify the additional expense of a further training. However, neither group was able to provide a reasonable time span to perform the ultrasound examination. More studies are needed to find ways to reduce the length of the examination and thus make it safe to use on real patients.

Digital biomarkers derived from mobile sensors could enable more frequent and objective monitoring of motor disorder symptoms. In this study, we examined the utility of five tablet-based tests in predicting changes in the MDS-UPDRS-III scores following controlled Levodopa administration, utilizing regression models trained on data from 28 patients. Our experiments revealed that the best-performing single-test models exhibited an inverse prediction in five patients. Implementing a combined model incorporating two tests reduced misclassification to three patients. Further stratification of patients into symptom-specific groups, based on the presence of medication-induced tremor changes, led to only one subject being misclassified. These results underscore the potential of mobile device biomarkers for more accurate prediction of symptom changes in Parkinson's Disease patients.